Final Report
2008 Texas Adult Probation Turnover Intention Study:
Line Community Supervision Officers and Direct-Care Staff

Submitted by

Dr. Won-Jae Lee
Assistant Professor of Criminal Justice
Angelo State University, San Angelo, TX
Member, Texas Tech University System
July 4, 2008

1
TABLE OF CONTENTS
Acknowledgments………………………………………………………………………………………2
Executive Summary…………………………………………………………………………………… 3
Introduction…………………………………………………………………………………………… 5
Methodology……………………………………………………………………………………….

8

Measurement of Variables & Descriptive Analyses……………………………………………………13
Bivariate & Multivariate Regression Analyses for Line Community Supervision Officers……………42
Bivariate & Multivariate Regression Analyses for Direct-Care Staff………………………………… 60
Structural Equation Modeling for Line Community Supervision Officers and Direct-Care Staff…… 76
Conclusion & General Policy Implications…………………………………………………………… 96
References………………………………………………………………………………………………100
Appendix - Comparison between Total and Usable Responses……………………………………… 110
Supplemental Statistical Information - Salary and Tenure…………………………………………… 112

2
ACKNOWLEDGMENTS
I would like to extend my gratitude to a number of people who assisted in completing this
report.
First, I am grateful to the four probation directors and their employees who willingly
assisted in the conduct of the initial survey which served as an impetus for this 2008
Texas Adult Probation Turnover Intention Study. Were it not for the support of Arlene
Parchman (Brazos CSCD), Leighton Iles (Fort Bend CSCD), Tom Plumlee (Tarrant
CSCD), and John Wilmoth (Concho Valley CSCD), this more comprehensive study
would not have seen the light of day.
Christie Davidson, Executive Director of the National Association of Probation
Executives, and Dan Richard Beto, Editor of Executive Exchange, deserve recognition,
because they provided invaluable assistance in developing the first survey and publishing
its results. Too, they have expressed a willingness to publish the results of this more
comprehensive study.
I’d like to express my special gratitude to Steve Enders, Chair of the Probation Advisory
Committee (PAC), and the other PAC members. Chairman Enders was very supportive of
this initiative, as were the other PAC members, all of whom provided helpful comments
in the development of this state-wide study and exerted their strong leadership in
enabling the Web-based survey for this report to be successfully completed. I would be
remiss if I did not acknowledge Paul Becker (Harris CSCD), Michael Noyes (Dallas
CSCD), William Fitzgerald (Bexar CSCD), Tom Plumlee (Tarrant CSCD), Geraldine
Nagy (Travis CSCD), Joe Lopez (Hidalgo CSCD), and Stephen Enders (El Paso CSCD)
for their financial support.
Special thanks also go to Carl T. Martin, Technical Services Analyst at Angelo State
University, and my editor Michèle Sharon. Mr. Martin and his IT staff helped me create
the Web-based survey, and were ever-vigilant in protecting the confidentially of the
survey data. I am also grateful for the assistance of Ms. Sharon, my dedicated editor, for
her work in helping this report reach a statewide audience.
And finally, my heartfelt thanks are extended to the 3,241 survey respondents, whose
involvement in this process may well assist in advancing the probation profession in
Texas.

3
EXECUTIVE SUMMARY
Purpose: Heretofore there has been no readily available, cost-effective mechanism has
been implemented in Texas probation, to fully and empirically analyze actual, voluntary
turnover. Voluntary turnover (excluding retirement and termination) can be preventable
by identifying its underlying reasons, and addressing identified causes. In response, the
purpose of this report is comprehensively to investigate: (1) any determinant factors that
shape turnover intention; and (2) pay satisfaction’s influence on organizational outcomes,
such as overall job satisfaction, organizational commitment and turnover intention.
Methodology: This State-wide report, supported by the Texas Probation Advisory
Committee, has been commissioned to conduct a Web-based survey targeting all line
probation officers and all direct-care staff. Of the usable sample of 3,234 responses,
2,653 were obtained from line officers and 581 from direct-care staff.
Findings: Results from the descriptive analyses indicate that large portions of the line
probation officers and direct-care staff have their high levels of inclinations to leave. For
example, 41.3 percent reported their turnover intention: 30.3 percent were having serious
thought about leaving in the near future, and another 11 percent were actively looking to
leave. Among all organizational factors used, pay and promotion are the most negatively
perceived work-related areas in Texas probation. As evidence of low levels of pay
satisfaction, only 10.3 percent reported their pay level was good, only 13.5 percent
indicated their pay level was either adequate or more than adequate given the cost of
living in their area and only 15.4 percent reported that their pay level had a favorable
influence on their overall attitude toward their job. Moreover, the average mean of
organizational commitment was lower than that of overall job satisfaction, suggesting
employees in Texas probation have a stronger psychological/ emotional attachment to
their job and job experience than their department.
Results from the bivariate and multivariate regression statistical analyses indicate
organizational factors, rather than individual status factors, have a substantially greater
contribution to associating with and predicting the employees’ turnover intention,
suggesting that, rather than an employee, his or her organization has underlying causes
for his or her turnover intention. For both line probation officers and direct-care staff,
affective commitment, high sacrifice commitment, overall job satisfaction and pay are the
main predictors of their turnover intention. Among the four main predictors, affective
commitment has the strongest direct effect on turnover intention.
Among all individual factors, young age group and tenure group are more likely to feel
inclined to leave their job. Specifically, age, especially the 20-34 age group (42.8% of the
line probation officers sampled), is the strongest predictor of their turnover intention,
whereas tenure, particularly the 0-3 years of tenure group (45.6% of the direct-care staff),
is the strongest predictor of their inclinations to quit.
Lastly, structural equation modeling analysis compared total effects (direct and indirect)
of compensation satisfaction (pay and fringe benefits), overall job satisfaction, lack of

4
alternatives, high sacrifice, and affective commitment on turnover intention. Results from
the structural equation modeling indicate that the total effect (indirect and direct) of
compensation satisfaction on turnover intention is much greater than the total effect of
affective commitment.
Overall, these findings suggest that while affective commitment has the strongest direct
effect on turnover intention, the total influence of compensation satisfaction, especially
pay satisfaction, is much more important than that of affective commitment in reducing
high levels of turnover intention and subsequent voluntary turnover in Texas probation.
Therefore, it can be concluded that pay satisfaction is the strongest underlying cause of
high turnover intention in Texas probation.
General Policy Implications: Most importantly, Texas probation administrators should
be acutely aware of the transition from individual to organization factors, especially the
significance of pay satisfaction and affective commitment, as possible underlying causes
leading to a high voluntary turnover rate. Accordingly, a concerted effort to convince the
Texas Legislature to significantly increase basic probation funding, and designing
strategies that enhance affective commitment are strongly recommended. In addition,
Texas probation administrators should recognize the unique characteristics of the new
generation of employee, and devote considerable attention and resources to them by
developing mentoring systems, embracing shift in supervisory and managerial roles and
styles, and developing improved selection and better training of managers. Lastly, it is
believed that a strong willingness and commitment on the part of Texas probation
administrators to address these identified causes of high turnover intention will restore
the effectiveness and efficiency of Texas probation departments in the promotion of
public safety.

5

Section 1.
Introduction

6
Other than retirement, American correctional agencies have been concerned with,
and paid significant attention to voluntary turnover. High levels of employee absenteeism,
stress, poor health, low morale, and high turnover rates continue to confront executives,
both in institutional and community corrections agencies. Each of these factors
contributes to poor job-related productivity (Finn, 1999; Mitchell, Mackenzie, Styve, &
Gover, 2000; Slate & Vogel, 1997; Slate, Vogel, & Johnson, 2001, Whitehead, 1987).
Specifically, voluntary turnover in a probation setting may result in increased caseloads
for the remaining staff. This may lead to a deterioration in supervision, low morale,
increases in unnoticed violations, absconders, recidivism, and increased expenditures
related to the recruitment and training of replacements (Simmons, Cochran, & Blount,
1997). These negative consequences could diminish the promotion of public safety,
which is the definitive mission of the Texas probation system.
Texas State Auditor’s Office (2007) reported 10.8 percent statewide voluntary
turnover rate (excluding involuntary separations and retirements) among all state
agencies but institutions of higher education during the fiscal year 2007. However, the
report does not provide any information about a voluntary turnover rate of Texas adult
probation. Despite no available statewide turnover rate for Texas probation, there is much
substantial evidence that high levels of employee turnover, and its attendant causes are
critical issues faced by probation executives. Florida probation agencies, for example,
reported a turnover rate of approximately 30 percent in 1995 (Simmons et al., 1997). In a
2000 report, the Texas Juvenile Probation Commission reported a 19.7 percent turnover
rate among the State’s juvenile probation officers in 1999. The Commission also reported
a 31.4 percent turnover rate for juvenile detention and corrections officers (Texas
Juvenile Probation Commission, 2000). In addition, despite the absence of extensive
national reports addressing community correctional officer turnover, members of the
National Institute of Corrections agreed that the loss of qualified officers was a major
concern (National Institute of Corrections, 1994).
Lee and Beto (2008) explored voluntary turnover rates among Texas line
probation officers from 2004 to 2006. They sampled four adult Community Supervision
and Corrections departments based upon department size, populations served, receptivity
to meaningful research, and recognized leadership. The four County departments
surveyed included Tarrant, Tom Green, Fort Bend, and Brazos. In this survey, the
voluntary turnover rate in each department was calculated by dividing the total number of
line officers who voluntarily resigned (excluding termination and retirement) by the total
number of line officers employed during each fiscal year (FY). Based on responses from
the four sampled probation departments, line officers’ average turnover rate in each fiscal
year was estimated to be 17-24%. Also, one department experienced an unusually high
voluntary turnover rate (nearly 40% in FY 2006). Interestingly, voluntary turnover rates
increased steadily during the study period: 17% for FY 2004, 20% for FY 2005, and 24%
for FY 2006 (Lee & Beto, 2008). Their findings supported previous studies, which
suggest that probation agencies have not only experienced high turnover, but have failed
to resolve the problem.

7
Remediating extensive staff turnover ought to be a top priority for Texas
probation administrators, especially in an era of tightening administration budgets, and
expanding public expectations. Unfortunately, no readily available, cost-effective
mechanism has been implemented in Texas, to fully and empirically analyze actual,
voluntary turnover rates State-wide. Such analysis should reveal underlying reasons
among probation officers for voluntary turnover. In response, this State-wide report,
funded by the Texas Probation Advisory Committee, has been commissioned to conduct
a Web-based survey targeting all non-managerial and non-supervisory probation officers,
and all direct-care staff. The report will examine: (1) any determinant factors (both
personal and organizational) that shape turnover intention; and, (2) pay satisfaction’s
influence on organizational outcomes, such as overall job satisfaction, organizational
commitment, and turnover intention.
The purpose of this report is to assist the Texas Probation Advisory Committee
(PAC), and Community Justice Assistance Division (CJAD) to pay attention to the
employee-turnover problem. This aim is achieved by identifying underlying determinants
of turnover, and by designing strategies that both enhance the working environment, and
diminish actual, voluntary turnover rates among line officers and direct-care staff. Thus,
administration may avoid the increasing, negative consequences of voluntary turnover–
such as unnoticed violations, absconders, recidivism, diminished morale among
remaining staff, and expenditures for recruitment and training new employees. Efficiently
addressing identified causes of turnover should be beneficial to staff in Texas probation,
thereby helping to restore the effectiveness and efficiency of Texas probation
departments in the promotion of public safety.
Structure of the Report
This report consists of seven sections including this Introduction as Section 1.
Section 2 describes the research design, methodology, and data collection. The data
analyses and results are reported in the next four sections (3, 4, 5, and 6). Section 3
provides a brief measurement description and univariate statistics (variable frequencies,
means and standard deviations) for each variable included in this report. Section 4
contains both bivariate and multivariate regression analyses for line community
supervision officers. These analyses are designed to provide information on the strength
and direction of the association between each predicting variable and turnover intention,
and also identify which predicting variable(s) are found to be significant determinants of
turnover intention. Replications are made for direct-care staff in the following Section 5.
Issuing from the data analyses and results, Section 6 provides evidence of the
causal relationship between pay satisfaction and significant attitudinal and behavioral
consequences–overall job satisfaction, organizational commitment, and turnover
intention–in the Texas probation system. Also, this section provides a comparison of the
total influence of pay satisfaction on turnover intention, with that of overall job
satisfaction and organization commitment, respectively. Based on results from these
analyses, general policy implications are discussed in Section 7. At the end of the report,
supplemental statistical information on salary and tenure is provided.

8

Section 2.
Methodology

9
Sample
The Turnover Intention Study (“the Study”) was administered to all nonmanagerial, and non-supervisory line probation officers, as well as to all direct-care staff
in the122 probation departments across Texas. The sampling scope was limited to these
groups, since existing literature (e.g., Slate & Vogel, 1997; Slate, Vogel, & Johnson,
2001) suggests that these groups have less opportunity to participate in decision-making
than supervisors and managers. Not surprisingly, these groups also indicated increased
stress, and lower levels of job satisfaction, and organizational commitment; thereby
leading to high turnover intention, or actual turnover itself. In this report, a line
community supervision officer is defined as supervising at least one direct case. The line
community supervision officer group includes both full-time (working at least 40 hours
per week), and part-time (working less than 40 hours per week) community supervision
officers.
The second group targeted for the Study was direct-care staff across the State of
Texas. According to Stephen L. Enders, PAC Chair, direct-care staff are defined as,
“…all CSCD employees who have direct contact with probationers or other clients as an
assigned job duty. This would include case workers, counselors, counselor interns,
residential monitors, caseload technicians, and technicians assigned to the inter/intrastate
caseloads. It would not include secretaries, general clerks, computer technicians, fiscal
clerks, couriers, transportation specialists, and other staff not assigned to a caseload or to
have contact with clientele as part of their regular duties.” Similar to the line community
supervision officers, the direct-care staff group includes both full-time and part-time
workers.
Data Collection, Recruitment Procedures and Data Confidentiality
With minimal cost, and efficient time planning, the PAC reached a consensus that
using a Web-based, open, anonymous survey would be the most satisfactory survey
method for the 122 probation departments in Texas, and the most practical. Accordingly,
the Web-based survey was conducted via Angelo State University’s Survey System. The
survey period began on March 31 and ended on April 18, 2008. The survey used 137
questions for line community supervision officers, and 135 questions for direct-care staff
respectively. The particular survey items for the report will be explained in Section 3.
Since it was deemed that a completely open survey would not likely secure a high
response rate, question #1 on the on-line survey provided a drop-box menu listing all the
Texas probation departments. Respondents were then required to select their location
from the list, in order for the researcher to elicit a response rate for each department.
Substantial efforts were made by the PAC and department directors to secure a high
response rate, for validity and reliable analysis and reporting. Recognizing the
significance of the Study, they elicited substantial cooperation, and voluntary
participation from all line probation officers and direct care staff members, to ensure the
Web-based survey’s success.

10
Before and during the survey period, the PAC Chair and individual department
directors announced the on-line survey in a manner believed to secure the highest
possible participation rate; encouraging individual line probation officers, and direct-care
staff to access the survey Web site, and to complete it. They also informed respondents
that they could only take the survey once. Each Web-based survey included an
encouraging cover letter from PAC Chair Stephen Enders, on PAC stationary. This was
placed before a consent form, and the survey questionnaire. In his cover letter, Mr.
Enders outlined why the survey was being conducted, why it was important for the
respondent to participate, and why it was important that they complete each question.
After reading the letter of encouragement, those who chose to participate in the
survey were forwarded an electronic consent form. Each respondent was then informed
of the survey’s purpose, and the following procedures: (1) data obtained in the survey
would remain confidential and not be linked with any individual identifiers; (2) only
aggregate data would be used in any subsequent report(s) or presentation(s) describing
the results of this survey; (3) The name of the respondent’s department would be stripped
from the data and destroyed once the data was digitally stored; and, (4) the data would be
maintained under lock and key by the researcher, and would not be shared with anyone
including the CJAD, PAC, or any other probation departments in Texas, except in
summary form. At the bottom of the electronic consent form, a ‘Take the Survey’ button
was available for respondents to ‘click,’ allowing them access to the actual survey, and
establishing their willingness to participate.
To address a potential breach of data security, as with any server connected to the
Internet, the survey’s data was kept strictly confidential, and protected in the following
three ways. First, the response-data files were stored in a location that is not directly
accessible via the World Wide Web. Second, the survey data could only be accessed by
the researcher through a password-protected page. Finally, Angelo State University-IT
staff were instructed to be vigilant in protecting the survey data, and to conduct all data
handling with due diligence, in order to prevent data compromise; this meant constantly
keeping all data and response information anonymous, as well as confidential.
During the three-week Web-based survey period, a total of 108 departments
participated. Employees from the remaining 14 departments did not respond to the survey.
In this case, individual departmental directors were contacted, and asked why they did
not participate. Interestingly, the researcher found that the primary reason cited by the 12
directors for not responding was a lack of Internet capacity, to access the survey Web site.
The remaining two departments were found to have only one employee, and they were,
therefore, responsible for both line-officer and director duties. These two departments
were subsequently removed from the total 122 departments being targeted, and the total
was reduced to 120.
For the 12 departments without Internet access, the researcher obtained prior
permission from each administrator to dispatch hard copies of the survey by mail. The
hard-copy survey period lasted between April 18 and May 8, 2008, and data collection
was conducted separately at each of the 12 departments. The same questionnaire used for

11
the Web-based survey was distributed to each department. In each case of hard-copy data
collection, a consent form, and a letter emphasizing that survey participation was
voluntary, and that responses were collected anonymously, was provided. Each
respondent was given a pre-addressed, stamped envelope in order to return the survey
directly to the researcher.
Response Rate and Descriptive Statistics of the Respondents
As detailed in the Appendix, survey responses were obtained from a total of 3,241
line probation officers and direct-care staff in Texas. After examining the data on an
item-by-item, and case-by-case basis however, it appeared that of the 3,241 responses, 7
cases required deletion due to missing data: 6 for line probation-officer cases, and 1
direct-care staff case. This reduced the usable data sample to 3,234. The 3,234 responses
were obtained from 120 adult community supervision and corrections departments, with a
100 percent departmental response rate. It should be noted that the Crane and Winkler
departments were excluded from the survey, since both employ only one person, with
both line-officer and managerial duties.
Of the remaining usable sample of 3,234 responses, 2,653 were obtained from
line officers, and 581 from direct-care staff. However, there is no available official
information on the baseline population of both groups to calculate each group’s response
rate. Instead, only the total number of the community supervision officers including
supervisors and managers (N = 3,520) was available. By using this figure as the baseline
population, the response rate for the 2,653 line probation group was 75.4 percent. Since
the total number included both line officers and supervisory/managerial officers, the
response rate should be well over 75.4 percent. Social scientists agree that at least a 50
percent return rate is required for adequate analysis and reporting, at least a 60 percent
response rate is good, and a response rate of 70 percent or higher is very good (Maxfield
& Babbie, 2005). Therefore, a response rate greater than 75.4 percent response rate in the
sample of the line probation officers across the 120 Texas probation departments is
considered more than very good, thus indicating more than very good survey quality.
Individual status data listed in Table 1 represents respondents’ socio-demographic
and work-experience information. The selection of these individual status variables
incorporated into the survey was guided by an extensive literature review. Respondents
were employed by their department for an average of 7.31 years, ranging from a
minimum of 0.08 to a maximum of 34 years. Females accounted for 60.9% of the survey
population, and in terms of ethnic group, 47.3% were Caucasian, compared to Hispanic
(31.2%), African-American (18.8%), and Others (2.6%). The average age of the
respondents was 40.27 years (the minimum was 20 years, and the maximum 75 years),
with 58.9% reported to be married. 45.7% reported no children at home. In terms of
educational background, while 17.3% had less education than a Bachelor’s degree, the
majority (69.3%) had earned a Bachelor’s degree, and 13.4% had earned a Master’s, or
higher degree. Of the total respondents, 32.6% had prior employment in probation,
followed by 16.3% in institutional corrections, 10.8% in law enforcement, and 6.3% in
parole.

12

Table 1. Individual Status Variable Statistics
Variable
Employee classification
Community Supervision Officer
Direct-Care Staff
Gender
Male
Female

N (%)

N

3230
1264 (39.1)
1966 (60.9)
40.27 yrs
1520
1003
605
85

Martial staus
Currently married
Currently single

1892 (58.9)
1320 (41.1)

20

75

3203
3213

(47.3)
(31.2)
(18.8)
(2.6)
3212

No. of children at home

0.94

0 (none)

3*

3215
3219

402
154
2231
413
19

(12.5)
(4.8)
(69.3)
(12.8)
(0.6)

Tenure in current department

* 3 or more children at home.

Max

2653 (82)
581 (18)

Ethnicity
Caucasian
Hispanic
African American
Other

Prior employment in CJ system
Probation
Law enforcement
Corrections
Parole

Min

3234

Age

Education level
High school diploma or GED
Associate degree
Bachelor's degree
Master's degree
Doctorate degree

Mean

7.31 yrs
If "yes"
727 (32.6)
348 (10.8)
525 (16.3)
201 (6.3)

0.08

34

3196
3214

13

Section 3.
Measurement of Variables &
Descriptive Analyses

14
Means, standard deviations, and reliability for all 24 organizational variables may
be found in Table 2. All responses to all organizational variables were based on a
respondent’s experience over the past six month time frame before the beginning date of
the survey. Of all 24 organizational variables, Turnover Intention is the main dependent
variable; the remaining 23 organizational variables are independent. An extensive
literature review indicates that these independent variables have been theoretically and
empirically proven to be important correlates with turnover intention, and with actual
turnover. Thus, the organizational independent variables were selected and incorporated
into the survey. All scale items were measured using the five-point Likert scale.
Not tabulated in Table 2, is the validity of the scale items of each organizational
variable. All scales but three were found to have an acceptable validity: affective
commitment, operational procedures, and procedural justice. The validity test for each of
these scales found one item to be heterogeneous to the original scale. Then, the
heterogeneous item was discarded for a better, more accurate scale. To insure the
reliability of all scales, the Cronbach Alpha statistical reliability procedure was applied,
to test for the internal consistency of each scale. Alpha reliability coefficients for each
additive scale ranged from 0.71 to 0.94, well above the minimal level of acceptability
(α = 0.70). In other words, all 24 scales are valid and reliable.
By using two cut-off points (2.5 and 3.5, on the 5-point scale), the average mean
of each scale was broken into the following three groups: a low-average group, a neither
low- nor high-average group, and a high-average group. First, utilizing the cut-off point
of 3.5 (the midpoint between Neither disagree, nor agree and Agree), six variables were
identified as belonging to the high-average group. The six variables included overall job
satisfaction, supervision satisfaction, co-workers’ satisfaction, nature of work satisfaction,
social support, and empowerment.
On the other hand, utilizing the cut-off point of 2.5 (the midpoint between
Disagree and Neither disagree, nor agree), three variables were identified as belonging
to the low-average categorical group: pay satisfaction, promotion, and role ambiguity.
Among the three variables, role ambiguity (uncertainty about what job actions are
expected) was found to have the lowest average mean, closely followed by promotion,
and pay satisfaction. Finally, the average mean of the other variables ranged between 2.5
and 3.5; thereby identified as neither in the low- nor high-average group, and not
supporting any one particular view.

15
Table 2. Organizational Variable Descriptions, Statistics, and Reliability
No. of
Final Items

Item Mean*

SD**

Reliablity (α)

N

Turnover intention

4

2.71

0.96

0.81

3227

Organizational commitment
Affective commitment
High sacrifice
Lack of alternative

5
3
3

3.17
3.21
3.26

0.95
1.06
0.99

0.84
0.77
0.75

3212
3222
3213

Overall job satisfaction

5

3.52

0.82

0.83

3225

Pay
Promotion
Supervision
Fringe benefits
Contingent rewards
Operating procedures
Co-workers
Nature of work
Communication

5
4
4
4
4
3
4
4
4

2.44
2.33
3.86
2.87
2.53
2.53
3.62
3.80
2.91

0.77
0.89
0.97
0.87
0.95
0.94
0.78
0.80
0.95

0.87
0.80
0.86
0.77
0.84
0.74
0.75
0.79
0.81

3227
3204
3207
3198
3212
3200
3212
3216
3220

Stress
Role overload
Role conflict
Role ambiguity
Dangerousness
Job stress

5
5
4
5
5

3.09
2.77
2.17
2.88
3.12

1.00
0.89
0.74
0.84
0.97

0.91
0.82
0.71
0.80
0.90

3220
3213
3207
3207
3221

Organizational justice
Distributive justice
Procedural justice

5
6

2.55
2.86

1.00
0.84

0.94
0.81

3202
3207

Social support

4

3.55

0.85

0.83

3212

Participatory Management
Participatory climate
Empowerment

7
12

2.89
3.64

0.88
0.55

0.88
0.83

3204
3210

Variable

Satisfaction

* Responses to each item are made on a 5-point Likert scale with anchors labeled (1) strongly disagree
and (5) strongly agree.
** Standard Deviation.

16
Turnover Intention
As the main dependent variable in this report, a respondent’s intention to leave in
Table 3 was measured using the four items developed by Shore and Martin (1989). These
turnover intention items were measured on a 1-5 Likert scale (1 = strongly disagree;
5 = strongly agree), by the level of agreement (α = 0.81).
Table 3. Itemized Turnover Intention Analysis
Item

N (%)

1. Which of the following most clearly reflects your feelings about
your future with this department in the next year?
I definitely will not leave.
I probably will not leave.
I am uncertain.
I probably will leave.
I definitely will leave.

562
816
1012
573
270

Mean

SD

Total N

2.74

1.18

3233

3.03

1.17

3233

2.38

1.15

3231

2.71

1.32

3230

2.71

0.96

3227

(17.4)
(25.2)
(31.3)
(17.7)
(8.4)

2. How do you feel about leaving this department?
It is very unlikely that I would ever consider leaving this department 207 (6.4)
As far as I can see ahead, I intend to stay with this department.
1190 (36.8)
I have no feeling about one way or the other.
500 (15.5)
I am seriously considering leaving in the near future.
980 (30.3)
I am presently looking and planning to leave.
356 (11.0)
3. If you were completely free to choose, would you prefer or not
prefer to continue working with this department?
I prefer very much to continue working for this department.
I prefer to work here.
I don't care either way.
I prefer not to work here.
I prefer very much not to continue working for this department.

747
1395
363
577
149

(23.1)
(43.2)
(11.2)
(17.9)
(4.6)

4. How important is it to you personally that you spend your career in
this department rather than some other organization?
It is very important for me to spend my career in this department.
It is fairly important.
It is of some importance.
I have mixed feelings about its importance.
It is of no importance at all.

748
856
556
738
332

(23.2)
(26.5)
(17.2)
(22.8)
(10.3)

Average

17
Understandably, there might be a reasonable suspicion that even if an employee
shows an inclination to leave their employment, their intention may be influenced by the
economic climate and by circumstances in the labor market, and therefore might not
necessarily manifest in his or her actual turnover. However, Steel and Ovalle (1984), in
their meta-analysis, found that turnover intention was better than job satisfaction and
organizational commitment in predicting actual turnover, and suggested that turnover
intention does eventually lead to actual turnover. Furthermore, Hom and Griffeth (1995)
found that among 35 variables presumably related to actual voluntary turnover, turnover
intention had the strongest association with actual voluntary turnover. Recently, Griffeth
et al. (2000), in their updated meta-analysis, found that turnover intention was the best
predictor of the actual turnover process. These findings reach a similar conclusion;
turnover intention is the most immediate precursor of actual turnover.
The respondents’ inclinations to quit is mixed with an overall mean of 2.55, on a
Likert scale. However, many respondents indicated a strong inclination to leave their
department, in all questions. For example, the first item in Table 2, related to immediate
prediction of voluntary turnover, demonstrated that only 42.6 percent of the respondents
had no intention to leave their department in the next year. The remaining 57.4 percent
indicated they had either an uncertain, or certain intention to leave in the next year
(31.3% and 26.1%, respectively). In addition, 43.2 percent of the respondents indicated
that they did not plan to leave the department. However, a little over 30 percent reported
that they were having serious thoughts about leaving in the near future, and another 11
percent were actively looking to leave. These negative findings reveal substantial
evidence that identifying turnover intention should be a top priority for Texas probation
administrators, in order to reduce staff turnover in an era of tightening budgets and
expanding expectations.
Organizational Commitment
Organizational commitment has been found to be associated with both turnover
intention and actual turnover (Griffeth et al., 2000; Hom & Griffeth, 1995). Most recently,
Moynihan and Landuyt (2008), in their analysis of turnover intention among 34,668
employees of 53 different state agencies in Texas, found increased organizational
commitment reduced turnover intention. Briefly, organizational commitment is the
emotional link between an employee and his or her organization, referring to the strength
of his or her identification with, and involvement in his or her organization (Meyer &
Allen, 1997). There have been many definitions of organizational commitment.
According to Mowday, Porter and Steers (1982), organizational commitment is defined
as the extent to which an employee is involved in and identifies with his or her
organization, and it is useful in predicting performance reflective of organizational
effectiveness, and turnover. In other words, an employee who is committed to his or her
organization is more likely to both work towards the organization’s goals, and stay with
the organization. In recent years, three different dimensions of organizational
commitment were developed by Allen and Meyer (1990), and have been widely
recognized, and used in organizational research. The three dimensions which characterize

18
an employee’s commitment to the organization include affective, continuance, and
normative commitment.
According to Meyer and Allen (1997), affective commitment is defined as an
employee’s emotional attachment to, identification with, and involvement in an
organization. The employee commits to the organization because he or she wants to. In
contrast, continuance commitment is defined as the extent to which an employee
perceives high costs, such as economic costs and social costs, which would be incurred
by leaving the organization. Here, the employee remains with the organization because he
or she needs to. It should be noted that the continuance commitment construct has two
sub-dimensional constructs: high personal sacrifice and lack of alternatives (Mathieu &
Zajac, 1990; Meyer and Allen, 1984; Powell & Meyer, 2004). The high personal
sacrifice refers to the commitment related to personal accumulated investments: This
commitment develops when an employee realize that he or she would lose accumulated
investments associated with leaving the organization, and therefore the employee needs
to stay with the organization. On the other hand, the lack of alternatives denotes the
commitment related to an employee’s lack of employment alternatives which increase the
costs associated with leaving the organization, leading to the employee’s decision to stay
with the organization.
Finally, normative commitment represents an employee’s feeling obligated to
continue employment. For example, the organization may have invested resources in
training an employee who then feels a moral obligation to put forth effort on the job and
stay with the organization to repay the debt. In other words, the employee stays with the
organization because he or she ought to. Since organizational commitment generally
reduces turnover (Mowday et al., 1982), all of the three dimensions of organizational
commitment are considered to contribute to reducing turnover intention, and actual
turnover. Specifically, each is useful in predicting what may cause an employee to remain
committed to an organization and also, in predicting what will cause an employee to
leave (Allen & Meyer, 1990; Meyer & Allen, 1997). This report, however, adopted
affective and continuance commitment constructs since some empirical literature (Jaros,
Jermier, Koehler, & Sincich, 1993; Meyer & Herscovitch, 2001; Ko, Price, & Mueller,
1997; Vandenberg & Scarpello, 1990) expressed recurring criticism of the poor
discriminant validity between normative commitment and affective commitment. Mainly,
these findings indicate normative commitment is not a unique predictor of turnover
intention, and actual turnover, due to its strong association with affective commitment.
Initially, six affective commitment, three high personal sacrifice, and three lack of
alternative items were adopted from Meyer and Allen (1997). The validity and reliability
of each adopted scale was evaluated. For the validity of the affective commitment items,
a ‘principal components’ factor analysis was conducted to isolate any underlying
dimensions present in the scale construction. One item was found to be heterogeneous
and was thereby discarded, bringing the original six items to five. The five items factored
together with an appropriate eigenvalue of 2.55—greater than 1.00 through a
discontinuity test—and factor loadings all over 0.50, suggesting substantial loadings
(Comrey & Lee, 1992). The Cronbach’s Alpha reliability coefficient for the additive

19
scale produced from the five items was 0.84, indicating a slight increase from that of the
original six items (0.82). Therefore, as seen in Table 5, affective commitment was
defined and operated using the five-item scale.
To assess the accuracy of the three high personal sacrifice and three lack of
alternative items, a principal components factor analysis was conducted. As demonstrated
in Table 4, all factor loading scores exceeded the 0.50 cut-off, suggesting substantial
loadings (Comrey & Lee, 1992). Not listed in the table, the KMO measure of sampling
adequacy (0.62) indicates that the factor analysis is appropriate. Results from the factor
analysis indicate that high personal sacrifice and lack of alternatives are loaded on
different measures, supporting the validity of the two sub-dimensional constructs of
continuance commitment (Mathieu & Zajac, 1990; Meyer and Allen, 1984; Powell &
Meyer, 2004). The Alpha reliability coefficient test for each additive scale was above the
minimal level of acceptability (α = 0.77 for high personal sacrifice, and α = 0.75 for the
lack of alternative scale). Hence, high personal sacrifice was defined and operated using
its original three-item scale, and a lack of alternatives was operationalized by its original
three-item scale.
Table 4. Factor Analysis on the Dimensions of Continuance Commitment
Rotated Factor Loadings*
High
Lack of
Sacrifice
Alternatives

Item*
1. It would be very hard for me to leave my department right now, even if I
wanted to.

0.90

2. Too much of my life would be disrupted if I decided I wanted to leave
my department right now.

0.88

3. One of the major reasons I continue to work for this department is that
leaving would require considerable personal sacrifice; another agency
may not match the overall benefits I have here.

0.64

4. Right now, staying with my department is a matter of necessity as much
as desire.

0.56

5. I believe that I have too few options to consider leaving this department.

0.84

6. One of the few negative consequences of leaving this department would
be the scarcity of available alternatives.

0.89

Eigenvalue
Explanation of Variance
Cronbach α

2.151
35.85
0.77

2.112
35.20
0.75

Note : Responses to each item are made on a 5-point scale; Principal components factor analysis with a
varimax rotation.

20
As seen in Table 5, the respondents displayed an overall average of 3.17 for the
level of their affective commitment to the department which is a mixed result and
therefore does not support any one particular view. However, many respondents reported
lower levels of emotional attachment to, identification with, and involvement in their
department. For example, the respondents either disagreed or agreed whether they ‘feel
like part of the family’ at their organization (31.8% disagreed vs. 44.6% agreed), and
whether they feel emotionally attached to their organization (33.7% disagreed vs. 43.4%
agreed). There is evidence of low levels of respondents’ affective commitment: 26.6
percent of the respondents (vs. 48.4%) did not want to spend the rest of their career in
their current department, and 29.5 percent (vs. 46.4%) did not feel a strong sense of
belonging to their department.
Table 5. Itemized Organizational Commitment Analysis
Item*

Mean

SD

N

Affective Commitment

3.17

0.95

3212

1.
2.
3.
4.
5.

3.28
3.15
3.10
3.23
3.21

1.23
1.26
1.24
1.14
1.21

3230
3230
3229
3224
3223

High sacrifice

3.21

1.06

3222

1. It would be very hard for me to leave my department right now, even if I
wanted to.
2. Too much of my life would be disrupted if I decided I wanted to leave
my department right now.
3. One of the major reasons I continue to work for this department is that
leaving would require considerable personal sacrifice; another agency
may not match the overall benefits I have here.

3.32

1.29

3228

3.28

1.27

3228

3.05

1.27

3228

Lack of Alternatives

3.26

0.99

3213

1. Right now, staying with my department is a matter of necessity as much
as desire.
2. I believe that I have too few options to consider leaving this department.
3. One of the few negative consequences of leaving this department would
be the scarcity of available alternatives.

3.59

1.15

3228

3.02
3.16

1.24
1.25

3222
3223

I would be very happy to spend the rest of my career in this department.
I do not feel like "part of the family" at my department. (R)
I do not feel "emotionally attached" to this department. (R)
This department has a great deal of personal meaning for me.
I do not feel a strong sense of belonging to my department. (R)

Continuance Commitment

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree.
(R) indicates a reverse-keyed item (scoring is reversed).

21
Existing literature (i.e., Meyer & Allen, 1997; Meyer, Stanley, Herscovitch, &
Topolnytsky, 2002) has empirically supported the contention that affective commitment,
compared to normative and continuance commitments, has the strongest correlations with
turnover intention and actual turnover. In other words, employees with strong affective
commitment to the organization are more valuable employees for any organization.
However, compared to the average of high personal sacrifice (3.21) and lack of
alternative (3.26), the average of affective commitment (3.17) was found to be slightly
lower. Unfortunately, this finding appears to indicate that the main reason why
respondents in Texas probation are committed to their department is their awareness of
the costs associated with leaving—such as their personal accumulated investments and
limited employment opportunities—rather than their strong emotional attachment to,
identification with, and involvement in the department. Regarding the high level of the
respondents’ high personal sacrifice and lack of alternatives, for example, 49.7% of the
respondents (vs. 30.3%) would stay with the department because too much of their life
would be disrupted, and 46.2% (vs. 33.5%) would not leave due to the scarcity of
available alternatives.
Job Satisfaction
In a study conducted for 35 members of an adult probation department in another
state, Leonardi and Frew (1991) found a lower level of job satisfaction among adult
probation officers than the national average. Job satisfaction is generally defined as an
employee’s affective reactions to their job, based upon the level of congruence between
an employee’s job expectations and the actual situational attributes present (Cranny,
Smith, & Stone, 1992). Tett and Meyer (1993), in their meta-analysis study, tested
several variables related to turnover and found that job satisfaction is a more important
correlate with turnover intention than organizational commitment. Like organizational
commitment, job satisfaction is based on an employee’s emotional and psychological
state. However, job satisfaction is different from organizational commitment. Job
satisfaction is a linkage between an employee and his or her job, resulting from the
appraisal of his or her job, and job experiences (Locke, 1976). In contrast, organizational
commitment is a linkage between an employee and his or her organization (Meyer &
Allen, 1997).
There are two measures of job satisfaction: overall job satisfaction and
satisfaction with specific aspects of the job such as pay, supervision, promotion, coworkers, and the job itself. Overall job satisfaction was included in the report because
Griffeth et al. (2000), in their meta-analysis, suggests that overall job satisfaction is a
better indicator than job-facet satisfaction in predicting turnover, although both are
related to turnover. However, the facet approach is useful to define which parts of the job
produce satisfaction or dissatisfaction, as a useful tool to help an organization identify
areas of dissatisfaction that it can improve (Spector, 1997). In the report, job satisfaction
was assessed using the five items of overall job satisfaction-scale developed by Brayfield
and Roth (1951), and the Job Satisfaction Survey (JSS) developed by Spector (1997). The
JSS is composed of 36 items measuring nine facets of job satisfaction. Both are
standardized, and widely used survey instruments.

22
Overall Job Satisfaction
The five items in Table 6, with a five-point subscale (1 for “strongly disagree” to
5 for “strongly agree”) were designed to measure the respondents’ overall job satisfaction.
The additive scale produced by these five items had a Cronbach’s alpha reliability
coefficient of 0.83, well above the minimal level of acceptability. Overall, a moderately
high level of job satisfaction was reported, with an average mean of 3.52 (close to the
midpoint between Neither disagree nor agree and Agree. More than half agreed that: “I
am seldom bored with my job” (55.6%; Average = 3.40); “I like my job better than the
average worker does” (56.6%; Average = 3.56); “I find real enjoyment in my job”
(59.6%; Average = 3.60); “Most days I am enthusiastic about my job” (58.6%; Average
= 3.52); and, “I feel fairly well satisfied with my job” (60.1%; Average = 3.52). These
findings indicate that over half of the respondents in Texas probation are satisfied with
their job.
Table 6. Itemized Overall Job Satisfaction Analysis
Item*

Mean

SD

N

1. I am seldom bored with my job.

3.40

1.24

3229

2. I like my job better than the average worker does.

3.56

0.96

3229

3. I find real enjoyment in my job.

3.60

1.00

3231

4. Most days I am enthusiastic about my job.

3.52

1.03

3230

5. I feel fairly well satisfied with my job.

3.52

1.02

3229

3.52

0.82

3225

Average

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree.

Nine Facets of Job Satisfaction
As discussed, it is impractical to define overall job satisfaction by a discrete
definition that limits this category to specific or static facets. The Job Satisfaction Survey
(JSS) by Spector (1997) was originally developed to evaluate nine-faceted job
satisfaction in public organizations. Therefore, the JSS was also included in this report to
specify which areas of dissatisfaction may affect turnover intention in Texas probation
specifically. The nine facets include: Pay (satisfaction with pay and pay raises);
Promotion (satisfaction with promotion opportunities); Supervision (satisfaction with the
person’s immediate supervisor); Fringe benefits (satisfaction with fringe benefits);
Contingent rewards (satisfaction with rewards, not necessarily monetary, given for good
performance); Operating procedures (satisfaction with rules and procedures); Coworkers (satisfaction with co-workers); Nature of work (satisfaction with the type of

23
work done); and Communication (satisfaction with communication within the
organization).
Pay Satisfaction
This report employed both the four items of pay satisfaction in the JSS, and the
five items of satisfaction with financial rewards from the Index of Organizational
Reactions (IOR) developed by Dunham and Smith (1979). Williams, McDaniel and
Nguyen (2006), and Williams, Malos and Palmer (2002), in their meta-analysis of the
antecedents and consequences of pay-level satisfaction, recognized the five items of pay
satisfaction developed by Dunham and Smith (1979) as an accurate tool for measuring
multi-dimensional pay satisfaction, (not uni-dimensional pay level satisfaction),
reflecting a better understanding of the nature and domain of pay satisfaction. The
validity and reliability of these two pay satisfaction scales are compared in Table 7.
Table 7. Comparison of Two Pay Satisfaction Scales
Two Pay Satisfaction Scales
IOR developed by
Spector (1997)

JSS developed by
Dunham and Smith (1979)

4

5

2.16

2.44

1

1

0.71

0.86

2636.18
6
0.000

7458.43
10
0.000

Eigenvalue

2.199

3.264

Explanation of Variance

54.97

65.27

0.73

0.87

Descriptive Statistics
N of Items*
Mean
Validity Test**
N of Component(s)
KMO***
Bartlett's Test of Specificity
Approximate Chi-Square
df
Significance

Reliablity Test
Cronbach α

* Responses to each item are made on a 5-point scale. Higher scores indicate favorable responses.
** Principal components factor analysis.
*** Kaiser-Mayer-Okline measure of sampling adequacy.

24
Results from the KMO measure of sampling adequacy indicate that the factor
analysis of Dunham and Smith’s pay satisfaction was more appropriate than Spector’s
pay satisfaction (KMO = 0.86 and KMO = 0.71, respectively). Likewise, more total
variance was explained by Dunham and Smith’s pay satisfaction items than Spector’s pay
satisfaction items (65.27% and 54.97%, respectively). These results suggest the better
validity of the five items of pay satisfaction scale developed by Dunham and Smith
(1979). Similarly, the reliability of Dunham and Smith’s pay satisfaction scale (α = 0.87)
was found to be higher than that of Spector’s pay satisfaction scale (α = 0.73). Therefore,
the report adopted only Dunham and Smith’s pay satisfaction scale for further statistical
analysis.
As shown in Table 8, the respondents displayed an average of 2.44 for the level of
their pay satisfaction. The average of 2.44 is below the midpoint between “disagree” and
“neither disagree or agree,” suggesting their low levels of satisfaction with pay. There is
more substantial evidence that more than a half had their low levels of pay satisfaction:
68.3 percent of the respondents did feel their pay level poor; 52.9 percent reported their
pay level inadequate given the cost of living in their area; and 52.5 percent indicated that
their pay level made an unfavorable influence on their overall attitude toward their job.
Conversely, only small percent of the respondents revealed their high levels of pay
satisfaction. For example, 10.3 percent did feel their pay level good; 13.5 percent
indicated their pay level either adequate or more than adequate given the cost of living in
their area; and 15.4 percent reported that their pay level made a favorable influence on
their overall attitude toward their job.
In addition, Spector’s pay satisfaction scale provides valuable information on the
low levels of respondents’ pay satisfaction, although it is no longer used for further
analysis. Regarding pay raises, 75 percent indicated that the pay raises they received were
too few and far between. Furthermore, 54.7 percent of the respondents reported that they
were unappreciated by the department considering the pay level they received. On the
contrary, only 12.1 percent reported satisfaction with pay raises, and only 20.7 percent
revealed that they were appreciated by the department, considering what the department
paid them. These findings clearly indicate that the respondents in Texas probation are not
satisfied with their pay, and therefore probation administrators need to pay substantial
attention to the low levels of pay satisfaction among employees.

25
Table 8. Itemized Pay Satisfaction Analysis
Item

N (%)

1. For the job I do, I feel the amount of money I make is:
very poor.
fairly poor.
neither poor nor good.
good.
extremely good.

1051
1157
692
320
12

202
879
1396
674
81

708
1001
1088
425
9

636
1165
1136
252
41

526
1171
1039
378
118

0.98

3232

2.86

0.90

3232

2.39

0.98

3231

2.35

0.92

3230

2.50

1.01

3232

2.44

0.77

3227

(19.7)
(36.1)
(35.2)
(7.8)
(1.3)

5. How does the amount of money you now make influence your
overall attitude toward your job?
It has a very unfavorable influence.
It has a slightly unfavorable influence.
It has no influence one way or the other.
It has a fairly favorable influence.
It has a very favorable influence.

2.10

(21.9)
(31.0)
(33.7)
(13.2)
(0.3)

4. Does the way pay is handled here make it worthwhile for a
person to work especially hard?
It definitely discourages hard work.
It tends to discourage hard work.
It makes little difference.
It tends to encourage hard work.
It definitely encourages hard work.

Total N

(6.3)
(27.2)
(43.2)
(20.9)
(2.5)

3. Considering what it costs to live in this area, my pay is:
very inadequate.
inadequate.
barely adequate.
adequate.
more than adequate.

SD

(32.5)
(35.8)
(21.4)
(9.9)
(0.4)

2. To what extent are your needs satisfied by the pay you receive?
almost none of my needs are satisfied.
very few of my needs are satisfied.
a few of my needs are satisfied.
many of my needs are satisfied.
almost all of my needs are satisfied.

Mean

(16.3)
(36.2)
(32.1)
(11.7)
(3.7)

Average

26
The Other Eight Faceted Job Satisfaction Scales
Table 9 summarizes the other eight-facet job satisfaction scales used, the average
mean, and the standard deviation. Not tabulated here, each of the eight scales was found
to have an acceptable validity and reliability. Note that the validity test for the operatingprocedures scale found that one item was heterogeneous, and was therefore discarded,
bringing the original four items to three, as indicated in Table 9. Responses to all items
were made on a 5-point Likert scale. By using two cut-off points (2.5 and 3.5 on the 5point scale), the average mean of each satisfaction scale was broken into the following
three groups: unsatisfied job aspect, neither satisfied nor satisfied job aspect, and satisfied
job aspect. First, utilizing the cut-off point of 3.5 (the midpoint between Neither disagree
nor agree and Agree), three satisfied job aspects were identified. Among the job aspects,
satisfaction with the respondents’ immediate supervisor (average = 3.86) was found to be
the highest, closely followed by satisfaction with the type or their work done (average =
3.80) and satisfaction with co-workers (average = 3.62).
Second, the average mean of other job aspects except for promotion varied
between 2.5 and 3.5, and therefore were identified as neither “satisfied” nor “dissatisfied”
job aspects, not supporting any one particular view. The group includes fringe benefits,
contingent rewards, operating procedures, and communication. Lastly, utilizing the cutoff point of 2.5 (midpoint between disagree and neither disagree nor agree), only
promotion (average = 2.33) in Table 9 was identified as an unsatisfied job aspect. In other
words, respondents reported low satisfaction with promotion opportunities. There is
evidence to indicate that more than half of the respondents had low levels of promotion
satisfaction: nearly 66 percent of the respondents perceived too little chance for
promotion in their department; 50 percent did not feel they were given a fair chance of
promotion for those who performed well on the job; and 54.7 percent reported high levels
of dissatisfaction with their chances for promotion. Only a small percentage of
respondents expressed high levels of promotion satisfaction. For instance, only 14.1
percent perceived much chance for promotion in their department; 25.2 percent felt they
had a fair chance of being promoted for those who performed well on the job; and 16.2
percent reported high levels of satisfaction with their chances for promotion.
Taken together, the respondents had moderately high levels of overall job
satisfaction. This finding suggests a strong linkage between an employee, and his or her
job and job experience in Texas probation. However, average means of pay satisfaction
(average = 2.44), and promotion satisfaction (average = 2.33) were found to be lower
than the midpoint between Disagree and Neither disagree nor agree. In other words, pay
and promotion are the parts of the job that substantially contribute to dissatisfaction.
Probation administrators need to pay serious attention to these low levels of pay and
promotion satisfaction among employees in Texas probation, and therefore need to
develop substantial strategies to enhance these two aspects of the job’s terms and
conditions.

27
Table 9. Itemized Job Satisfaction Facet Analysis (excluding Pay Satisfaction)
Item*

Mean

SD

N

Promotion

2.33

0.89

3204

1.
2.
3.
4.

2.15
2.58
2.19
2.40

1.14
1.20
1.02
1.12

3220
3215
3213
3208

Supervision

3.86

0.97

3207

1. My supervisor is quite competent in doing his/her job.
2. My supervisor is unfair to me. (R)
3. My supervisor shows too little interest in the feelings of subordinates.
(R)
4. I like my supervisor.

3.81
3.96
3.64

1.20
1.11
1.26

3215
3216
3216

4.02

1.03

3210

Fringe Benefits

2.87

0.87

3198

1.
2.
3.
4.

2.97
3.06
3.12
2.33

1.24
1.17
1.07
1.05

3216
3212
3212
3207

Contingent Rewards

2.53

0.95

3212

1. When I do a good job, I receive the recognition for it that I should
receive.
2. I do not feel that the work I do is appreciated. (R)
3. There are few rewards for those who work here. (R)
4. I don't feel my efforts are rewarded the way they should be. (R)

2.64

1.19

3217

2.77
2.29
2.41

1.21
1.10
1.12

3217
3215
3214

Operating Procedures

2.53

0.94

3200

1. Many of our rules and procedures make doing a good job difficult. (R)
2. I have too much to do at work. (R)
3. I have too much paperwork. (R)

2.73
2.54
2.33

1.21
1.11
1.16

3213
3212
3204

Co-workers

3.62

0.78

3212

1. I like the people I work with.
2. I find I have to work harder at my job than I should because of the
incompetence of people I work with. (R)
3. I enjoy my co-workers.
4. There is too much bickering and fighting at work. (R)

4.10
3.20

0.83
1.19

3224
3223

4.00
3.16

0.82
1.22

3222
3221

There is really too little chance for promotion on my job. (R)
Those who do well on the job stand a fair chance of being promoted.
People get ahead as fast here as they do in other places.
I am satisfied with my chances for promotion.

I am not satisfied with the benefits I receive. (R)
The benefits we receive are as good as most other organizations offer.
The benefit package we have is equitable.
There are benefits we do not have which we should have. (R)

Table continued…

28
Table 9, continued
Item*

Mean

SD

N

Nature of Work

3.80

0.80

3216

1.
2.
3.
4.

3.37
3.95
4.11
3.75

1.26
0.88
0.92
0.98

3222
3219
3222
3220

Communication

2.90

0.95

3220

1.
2.
3.
4.

2.61
3.11
2.73
3.16

1.23
1.20
1.21
1.14

3226
3226
3226
3224

I sometimes feel my job is meaningless. (R)
I like doing the things I do at work.
I feel a sense of pride in doing my job.
My job is enjoyable.

Communications seem good within this department.
The goals of this department are not clear to me. (R)
I often feel that I do not know what is going on with the department. (R)
Work assignments are often not fully explained. (R)

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree.
(R) indicates a reverse-keyed item (scoring is reversed).

Job Stress
Job stress was found positively correlated with turnover intention (Begley &
Czajka, 1993). Theoretically and empirically, job stress and stressors are concepts that
should be distinguished from each other. There have been various definitions of stress.
For example, stress is defined as the non-specific response of the body to any demand
(Selye, 1956), the person-environment fit (Whitehead, 1985; 1987), and the lack of
congruity between individuals and their physical or social environment (Chesney &
Rosenman, 1980). Also, stress may be defined as the psychological discomfort or tension
caused by exposure to stressors which place unreasonable or distinctive demands on an
individual (Cullen, Link, Wolfe, & Frank, 1985). The current definition of job stress
includes both the ‘person-environment fit,’ and its equivalent, the ‘incongruity between
an individual and one’s physical or social environment.’
In conjunction with the person-environment fit perspective of stress, stressors
have been succinctly defined as “the conditions which place excessive or unusual
demands on a person and are capable of engendering psychological discomfort,
physiological pathology, and/or social disability” (Cullen, et al., 1985, p. 507). Stressors
are operationally defined as “circumstances which place unreasonable or distinctive
demands on an individual, and are usually capable of producing emotional/psychological
discomfort” (Grossi & Berg, 1991, p. 76). Both definitions are quite similar, and reflect
that the conditions of situations or events are stressors, and consequently produce jobrelated stress.
The results of the job-burnout study are informative. Job burnout, resulting from
prolonged job stress, is a syndrome wherein individuals working in human service

29
agencies experience three distinct outcomes: emotional exhaustion (work overextension
and exhaustion); depersonalization (an impersonal and cynical approach to clients); and
lack of personal accomplishment (a negative self-evaluation) (Maslach & Jackson, 1981).
Negative outcomes of job burnout symptoms include low job satisfaction, decreased
employee productivity, increased absenteeism, and high voluntary or actual turnover
(Gerstein, Topp, & Correll, 1987; Simmons, et al., 1996). Extensive literature (e.g.,
Cherniss, 1980; Maslach, 1982; Whitehead, 1985; 1987) suggests that role structure, such
as role overload, role conflict, and role ambiguity, is important source of job stress and
burnout.
Existing correctional literature concurs that role ambiguity and role conflict play a
negative impact on job-related stress and burnout (Byrd, Cochran, Silverman, & Blount,
2000; Cullen, et al, 1985; Finn, 1999; Grossi & Berg, 1991; Lindquist & Whitehead,
1986; Thomas, 1988; Walters, 1999; Whitehead, 1983; 1985; 1987; Whitehead &
Lindquist, 1996). In the prison setting, dangerousness of the job was found to be an
additional stressor to the role structure problem (Cullen, et al, 1985). Likewise, in a study
conducted for 457 adult probation officers in Texas, Sheeley (2008) found the need for
safety to be the fourth priority need following better communication, more resources, and
greater recognition. The literature review indicates that both job stress and stressors are
important correlates with turnover intention. Therefore, both the four stressors and job
stress levels shown in Table 10 were selected and incorporated into the survey.
Job stress (α = 0.90) was assessed using the five items developed by Crank,
Regoli, Hewitt and Wolfe (1989). Among the three role characteristics used, role
overload (α = 0.91) refers to having too much to do in the amount of time or the lack of
available resources for completing workload demands, and was measured using five
items developed by Peterson and his associates (1995). The other two role characteristics
are role conflict and role ambiguity which have been widely used in job-role research.
Both are different. Role conflict refers to conflicting requests from different people,
whereas role ambiguity refers to unclear expectations in fulfilling a role. Both role
conflict (α = 0.82) and ambiguity (α = 0.71) were measured using the total nine items
adopted from Lambert, Hogan, Paoline and Clarke (2005). Lastly, dangerousness of the
job (α = 0.80) was assessed using five items adopted from Cullen, Link, Cullen and
Wolfe (1989). All scale items used were measured using five-point Likert scales.
Respondents displayed an average of 3.12 for their job stress level, which was
between 2.5 and 3.5 on a 1-5 Likert scale, and therefore not supporting any one particular
view. However, 46.8 percent of the respondents reported that they were usually under a
lot of pressure at work, whereas 29.9 percent reported that they were not under the
pressure. Among the four stressors, role overload (average = 3.09) was found to be the
strongest stressor, closely followed by dangerousness of the job (average = 2.88) and role
ambiguity (average = 2.77). At 2.17, the average level of role ambiguity is below the
midpoint of 2.5 between “disagree” and “neither disagree or agree,” and suggesting that
uncertainty about what actions are expected, is not a particularly stressful condition in
Texas probation.

30
Table 10. Itemized Stress Analysis
Item*

Mean

SD

N

Role Overload

3.09

1.00

3220

1.
2.
3.
4.
5.

3.19
3.10
2.65
3.16
3.37

1.15
1.15
1.06
1.19
1.24

3223
3223
3222
3222
3222

Role Conflict

2.77

0.89

3213

1. I regularly receive conflicting requests at work from two or more
people.
2. When a problem comes up here, people seldom agree on how it should
be handled.
3. Sometimes, I am criticized by one supervisor for doing something
ordered by another supervisior.
4. I sometimes have to bend a rule or policy to get an assignment done.
5. I often receive an assignment without adequate resources and materials
to get it done.

2.83

1.15

3219

3.00

1.13

3219

2.61

1.19

3217

2.60
2.78

1.13
1.18

3218
3217

Role Ambiguity

2.17

0.74

3207

1.
2.
3.
4.

2.00
2.44
1.96
2.25

0.94
1.07
0.96
1.06

3212
3211
3212
3208

Dangerousness

2.88

0.84

3207

1. Most of the time when I'm at work I don't feel that I have much to worry
about. (R)
2. In my job, a person stands a good chance of getting hurt.
3. I work at a dangerous job.
4. My job is a lot more dangerous than most other jobs.
5. A lot of people I work with have been physically injured on the job.

2.90

1.19

3213

3.13
3.16
3.20
2.00

1.23
1.18
1.17
0.84

3212
3213
3211
3210

Job Stress

3.12

0.97

3221

1.
2.
3.
4.
5.

3.20
3.26
3.11
2.93
3.10

1.20
1.16
1.16
1.10
1.14

3221
3222
3222
3223
3223

There is a need to reduce some parts of my role.
I feel overburdened in my role.
I have been given too much responsibility.
My workload is too heavy.
The amount of work I have to do interferes with the quality I want to
maintain.

I clearly know what my work responsibilities are. (R)
The rules that we're supposed to follow seem to be very clear. (R)
I am unclear to whom I report and/or who reports to me.
I do not always understand what is expected of me at work.

A lot of the time my job makes me very frustrated or angry.
I am usually under a lot of pressure when I am at work.
When I'm at work I often feel tense or uptight.
I am usually calm and at ease when I'm working. (R)
There are a lot of aspects of my job that make me upset.

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree; (R). indicates a reverse-keyed item (scoring is reversed).

31
Many respondents reported high levels of role overload. Also, 51.1 percent of the
respondents felt that their workload, if heavy, negatively affected the quality of their
work, whereas only 30 percent reported they felt otherwise. Interestingly, an example
related to highly perceived dangerousness of the job, 45 percent of the respondents
reported their probation job was much more dangerous than most other jobs, while only
30 percent reported no differences of dangerousness between probation job and other jobs.
Regarding their high role conflict, for example, almost equal numbers of the respondents
either agreed or disagreed with the following statement: When a problem comes up here,
people seldom agree on how it should be handled (36.8% agreed and 38.6% disagreed,
respectively). Overall, these findings suggest that role overload, such as excessive
paperwork and expectations to complete job duties in too little time, substantially
contributes to stress-induced role characteristics. In addition, like a prison setting, the
dangerousness of the work need to be recognized as a substantial stressor in adult
probation field.
Organizational Justice
Organizational justice is related to fairness perception (Campbell & Pritchard,
1976; Cropanzano & Greenberg, 1997). Basically, if organizational injustice is perceived,
one feels relative deprivation or a feeling of discontent, which in turn may lead to a range
of attitudinal and behavioral effects, such as higher stress, lower job satisfaction, lower
organizational commitment, and higher turnover intention or actual turnover (Campbell
& Pritchard, 1976; Hendrix, Robbins, Miller, & Summers, 1999; Martin 1981). Empirical
research has supported the important theoretical link between organizational justice and
its organizational outcomes. For example, turnover intention is an aspect of motivation
that was found to be influenced by an employee’s perceived organizational fairness
(Acquino, Griffeth, Allen, & Hom, 1997; Hendrix, Robbins, Miller, & Summers, 1999)
Organizational justice conceptually includes two aspects of justice: distributive
justice and procedural justice. Distributive justice is the degree of fairness in distributing
rewards (Price & Mueller, 1986), while procedural justice is the degree of fairness in the
procedures used for distribution (Folger & Greenberg, 1985). Both distributive justice
and procedural justice are based upon employee judgments regarding the fairness of
outcomes and the fairness of the procedures.
Developed by Price and Mueller (1986), the five items shown in Table 11 were
designed to measure the respondents’ perceived fairness of outcome, which is distributive
justice. The scale was well above the minimal level of acceptability, evidenced by high
Cronbach’s Alpha reliability scores (α = 0.94). Procedural justice refers to the fairness of
the procedures in distributing outcomes, and was assessed through the use of seven items
adopted from Lambert, Hogan and Griffin (2007). However, the validity test found one
item to be heterogeneous to the original scale. Then, the heterogeneous item was
discarded for a better accurate scale, brining the original seven items to the five items.
The Cronbach’s Alpha reliability coefficient for the six items was 0.81, indicating a slight
increase from that of the original seven items (α = 0.76).

32
Table 11. Itemized Organizational Justice Analysis
Item*

Mean

SD

N

Distributive Justice

2.55

1.00

3202

1. the amount of effort that you have put forth?

2.53

1.10

3211

2. the responsibilities that you have at work?

2.58

1.09

3211

3. the stresses and strains of your job?

2.48

1.05

3209

4. the amount of education and training you have?

2.63

1.19

3208

5. the work that you have done well?

2.53

1.14

3207

Procedural Justice

2.86

0.84

3207

1. Promotions here are seldom related to employee performance. (R)

2.55

1.19

3214

2. Promotions are more related to whom you know rather than the quality
of work. (R)

2.54

1.26

3210

3. There is a fair opportunity to be promoted.

2.55

1.11

3214

4. My own hard work will lead to recognition as a good performer.

2.97

1.19

3216

5. The standards used to evaluate my performance at this department have
been fair and objective.

3.06

1.09

3215

6. I have little trust in my supervisor's evaluation of my work performance.
(R)

3.49

1.18

3214

The department has been fair in rewarding you considering:

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree.
(R) indicates a reverse-keyed item (scoring is reversed).

Respondents reported an average mean of 2.55 for their perceived level of
distributive justice. The average mean closely approached the cut-off point of 2.5
(midpoint between disagree and neither disagree nor agree), suggesting negative
judgments regarding the fairness of distributing rewards, such as pay and promotion. For
all five items, more than half of the respondents reported they perceived unfairness of
distributing rewards, given: the amount of the effort (53.8%), the responsibilities (51.6%),
the stresses and strains of the job (55.5%), the amount of education and training (48.5%),
and the work done well (53.4%).
Compared to the average mean of distributive justice (2.55), procedural justice
had a relatively higher average mean of 2.86. Respondents perceived procedural fairness
in the recognition of: hard work (38.6% vs. 36.2%), fair and objective standards to
evaluate their performance (38.2% vs. 28.1%), and the supervisor’s reliable evaluation of
their work performance (56.6% vs. 20.3%). However, nearly 50 percent of the
respondents reported procedural unfairness in performance-based promotions (49.8%)
and a fair opportunity for promotions (49.8%). In addition, 49.9 percent of respondents
perceived that promotions are given based on who you know rather than what you know.

33
Overall, these findings indicate a lack of fairness of distributing rewards such as pay and
promotion, as well as a lack of fairness in promotional procedures in the Texas probation.
Social Support
Job stress was found negatively correlated with social support (Etzion, Eden, &
Lapidot, 1998). As a provision of instrumental and emotional assistance, social support
can be obtained from both his/her supervisors and fellow officers. It can function as a
successful coping factor to alleviate job stress, preventing job dissatisfaction, enhancing
high levels of organizational commitment, and reducing turnover intention. Also, feelings
of high personal empowerment (Crozier, 1964; Spreitzer, 1996) and job competence
(Gist & Mitchell, 1992) may result from good social support at work. According to
Cullen and his associates (1985), successful social support at work depends on the quality
of interpersonal support from supervisors and fellow officers. There is substantial,
empirical evidence indicating that support from supervisors is essential in allowing
correctional officers to display positive, job-related attitudinal and behavioral outcomes
(Gardner, 1981; Jurik & Halemba, 1984; Poole & Regoli, 1980; Veneziano, 1984). Also,
feelings of low personal empowerment (Crozier, 1964; Spreitzer, 1996) and job
competence (Gist & Mitchell, 1992) may result from a lack of social support at work.
Developed by Spreitzer (1996), four items with a five-point subscale (1 = strongly
disagree to 5 = strongly agree) were adopted and slightly modified to reflect a
departmental support. The four items in Table 12 were amalgamated together to form
social support (α = 0.83), and used to measure interpersonal support from the members of
a respondent’s work group, peers, immediate supervisor, and department.
Table 12. Itemized Social Support Analysis
Item*

Mean

SD

N

1. I have the support I need from my workgroup or team to do my job well.

3.62

1.00

3218

2. I have the support I need from my peers to do my job well.

3.69

0.96

3219

3. I have the support I need from my immediate supervisor to do my job well.

3.68

1.08

3218

4. I have the support I need from my department to do my job well.

3.20

1.15

3213

3.55

0.85

3212

Average

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5)
strongly agree.

Respondents reported their levels of social support as positive, with an overall
mean of 3.55, exceeding the midpoint of 3.5 between neither disagree nor agree and
agree). All social supports used but support from the respondent’s department were
found to be moderately high. The highest level of social support reported came from

34
peers (3.69), closely followed by their immediate supervisors (3.68) and their workgroup
members (3.62). However, support from the respondents’ department is mixed with an
average mean of 3.20, not exceeding the cut-off point of 3.5. Compared to the other
social support, departmental support, which helps respondents perform their jobs well,
seems weak at helping respondents perform their jobs well. .
Participatory Management
In response to organizational issues surrounding the management of government,
President Clinton created the National Performance Review (NPR) in 1993 (Vernon &
Byrd, 1996). “Reinventing Government,” borne out of the NPR, criticized malfunctions
of hierarchical, centralized bureaucracies and envisioned the new roles of government
executives, which included developing a clear vision, creating a team environment,
empowering employees, putting customers first, communicating with employees, cutting
red tape, and creating clear accountability (Gore, 1993). Basically, bureaucratization
tends to reduce workers’ control over the means of production and alienate line workers
from the decision-making process by exerting extreme limitations on individual freedoms
and democracy (Kohn, 1976; Mouzelis, 1968).
Participatory management seeks to balance the involvement of superiors and
subordinates in information-sharing, decision-making, and problem-solving related to
production and quality control (Wagner, 1994). Research in the organizational and
correctional literature that focuses on participation in decision-making by employees has
emphasized its significant relationship to turnover intention and actual turnover (Eby,
Freeman, Rush, & Lance, 1999; Jackson, 1983; Slate & Vogel, 1997; Slate et al., 2001).
Furthermore, correctional research literature not empirically focused on the impact of
participatory management, has discussed and recommended utilizing this strategy as a
critical mechanism to improve officers’ job satisfaction, and mitigate job stress and/or
burnout, which are significantly associated with inclinations to leave (i.e., Byrd et al.,
2000; Simmons et al., 1997).
Based upon the literature, it has been recognized that a participatory management
structure is more beneficial than a rigid, autocratic structure for enhancing employee job
satisfaction. It does so by increasing employees’ ability to profoundly influence and
improve their stressful work environment through their own decision-making process.
This, in turn, leads to better job productivity and less absenteeism and turnover.
Accordingly, more attention needs to be paid to the impact of participatory management
in Texas Probation, with its voluntary turnover rate. In response, the report included both
participatory climate and empowerment, which have been recognized as important
elements of participatory management.
Participatory Climate
Developed by Slate, Wells, & Johnson (2003), seven items with a five-point
subscale (1 = strongly disagree to 5 = strongly agree) shown in Table 13 were designed
to measure the respondents’ perception of atmosphere for participation in decision-

35
making in their probation department. The scale examined was well above the minimal
level of acceptability evidenced by high Cronbach’s Alpha reliability scores (0.88). The
respondents recorded an average of 2.89 for the level of atmosphere for participation in
decision-making – neither agree nor disagree – which is considered mixed and therefore
does not support any one particular view.
Table 13. Itemized Participatory Climate Analysis
Item*

Mean

SD

N

1. My superiors ask me for input on decisions that affect me at work.

3.04

1.19

3211

2. I am encouraged to offer my opinion at work.

3.15

1.19

3211

3. There is opportunity for me to have a say in the running of this
department on matters that concern me.

2.59

1.16

3210

4. Management responds in a satisfactory manner to what I have to say.

2.72

1.12

3207

5. From past experience at this department, I feel it is a waste of time. (R)

3.06

1.16

3207

6. I feel comfortable about offering my opinion to supervisors at work.

3.24

1.15

3207

7. Those who actually do the work are involved in the writing of policies
at this department.

2.40

1.07

3207

2.89

0.88

3204

Average

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree.
(R) indicates a reverse-keyed item (scoring is reversed).

Despite no indication of one particular view, item analysis demonstrates
substantial evidence that the respondents’ opinions were not sought and respected. For
example, nearly 50 percent of the respondents (vs. 25.4%) felt that hey had no
opportunity to have a say in the running of their agency on matters that concern them,
41.4 percent (vs. 26.7%) indicated unsatisfactory response or feedback to their input, and
53.2 percent (vs. 15.5%) did not feel involved in the writing of policies. This evidence
clearly indicates the low levels of atmosphere for participation in their departments.
Empowerment
Participatory climate is related to empowerment; it is a non-traditional
organizational culture with an emphasis on facilitating, coaching, and consulting
employees (participatory climate), to facilitate a sense of control and self-efficacy
(empowerment). Empowerment is succinctly defined as a “process by which individuals
and groups gain power, access to resources and control over their own lives. In doing so,
they gain the ability to achieve their highest personal and collective aspirations and
goals” (Robbins, Chatterjee, & Canda, 1998, p. 91). The concept of empowerment as
intrinsic motivation implies that if an employee loses a sense of control and the selfefficacy to enhance or enrich their job, he or she will feel powerlessness. This feeling,

36
which includes low self-esteem and a diminished sense of autonomy and responsibility,
leads not only to poor quality job performance, but also a low level of desire to remain
attached to an organization (affective commitment), increasing the inclination to leave
(Hammer, Landau, & Stern, 1981; Mowday et al., 1982).
Empirical research has demonstrated a strong relationship between empowerment
and organizational commitment, and between turnover intention and subsequent
voluntary turnover. For example, empowerment was found to enhance organizational
commitment (Wu & Short, 1996), which in turn reduces voluntary turnover (Spreitzer &
Mishra, 2002). Koberg, Boss, Senjem, & Goodman (1999) found a negative relationship
between empowerment and turnover intention. Most recently, Moynihan and Landuyt
(2008), in their analysis of turnover intention among 34,668 Texas state employees,
found that a sense of empowerment reduces turnover intention. Literature on employment
suggests that fostering empowerment among employees can be target through
organizational intervention, to increase organizational commitment and mitigate turnover
intention and voluntary turnover.
Empowerment was assessed through the use of the Index of Empowerment
developed by Spreitzer (1995), which is composed of 12 items. The Index of
Empowerment measures four dimensions of empowerment (meaning, competence, selfdetermination and impact). These four dimensions, reflecting an employee’s orientation
to his or her work role, were combined into an overall measure of empowerment. This
scale yielded adequate reliability (α = 0.83). As demonstrated in Table 14, respondents
reported an average of 3.64 for their level of empowerment. The average mean exceeded
the midpoint of 3.5 between neither disagree nor agree and agree), suggesting that
respondents believe that they have a moderately high level of empowerment in their
department.
Separate analysis of the four dimensions of empowerment can provide probation
administrators with valuable managerial information since empowerment is a concept of
intrinsic motivation resulting from the four dimensions. The average means of all subdimensional scales but impact were found to be moderately high or very high. The
respondents reported the highest level of competence (confidence in one’s effectiveness
and job-performance), followed by meaning (the fit between the value of a work role and
an employee’s own values and standards), self-determination (an employee’s autonomy
and independent decision-making in the initiation and continuation of work behavior and
processes). However, the respondents’ report of the degree of their impact on workrelated outcomes is very negative, with an average mean of 2.29 not exceeding the
midpoint of 2.5 between disagree and neither disagree nor agree. Therefore, compared
to the other dimensions of empowerment, the low degree of impact on work-related
outcomes does not appear to make employees feel empowered.

37
Table 14. Itemized Empowerment Analysis
Item*

Mean

SD

N

Meaning

4.19

0.77

3211

1. The work I do is very important to me.

4.27

0.79

3211

2. My job activities are personally meaningful to me.

4.10

0.86

3211

3. The work I do is meaningful to me.

4.19

0.83

3211

Competence

4.37

0.64

3211

1. I am confident about my ability to do my job.
2. I am self-assured about my capabilities to perform my work activities.

4.51

0.64

3211

4.47

0.66

3211

3. I have mastered the skills necessary for my job.

4.14

0.85

3211

Self-determination

3.72

0.92

3210

1. I have significant autonomy in determining how I do my job.

3.85

0.98

3211

2. I can decide on my own how to go about doing my work.
3. I have considerable opportunity for independence and freedom in how I
do my job.

3.78
3.53

1.05
1.13

3211
3210

Impact

2.29

0.98

3211

1. My impact on what happens in my department is large.

2.66

1.17

3211

2. I have a great deal of control over what happens in my department.

2.08

1.02

3211

3. I have significant influence over what happens in my department.

2.14

1.06

3211

3.64

0.55

3210

Average

* Responses to each item are made on a 5-point scale with anchors labeled (1) strongly disagree and (5) strongly
agree.
(R) indicates a reverse-keyed item (scoring is reversed).

To summarize, the descriptive analysis for turnover intention reveals substantial
evidence of why Texas probation administrators need to pay immediate and serious
attention to the high levels of voluntary turnover intention among their employees. For
example, in response to the question, “How do you feel about leaving this department?”
41.3 percent of the respondents revealed inclinations to leave. Specifically, 30.3 percent
reported that they were having serious thoughts about leaving in the near future, and
another 11 percent were actively looking to leave.
In the descriptive analysis for organizational commitment as a predictor of
turnover intention, the average mean of affective commitment was lower than that of
continuance commitment (both high personal sacrifice and lack of alternatives). This
suggests that the Texas probation respondents are committed to their department only in
so far as they are aware of the costs associated with leaving, such as their personal
accumulated investments and limited employment opportunities, rather than their strong

38
emotional attachment to, identification with, and involvement in their department. Note
that those with strong affective commitment to the organization are more valuable
employees for any organization than those with strong continuance commitment.
Furthermore, the link between the respondents and their department (organizational
commitment) was found to be weaker than the linkage between the respondents and their
job and job experience (job satisfaction).
The accumulated findings from the descriptive analyses provide useful
managerial information, which may help probation administrators identify areas that they
need to improve. Relatively, overall job satisfaction, supervision satisfaction, co-worker
satisfaction, nature of work satisfaction, social support, and empowerment are the
positively perceived working areas. However, probation administrators in Texas need to
pay serious attention to the two negatively perceived working areas: pay and promotion
satisfaction. Related to pay and promotion satisfaction, more than half of the respondents
indicated a lack of fairness in distributing rewards, such as pay and promotion, and a lack
of fairness in promotional procedures. Finally, a low level of perceived atmosphere for
participation in decision-making was found, recommending a shift in supervisory and
managerial roles and styles from directing and controlling line officers and direct-care
staff in a traditional, autocratic culture, to facilitating, coaching, and consulting with them
in participatory management.
Separate Analyses between Two Groups and Pre-Analysis Data Screening
The previous descriptive analyses were used to provide a brief description and
univariate statistics, such as variable frequencies, means, and standard deviations for each
variable. Descriptive analysis is useful to summarize each individual variable but it can
not explore any differences and relationships between the values of the dependent
variable (turnover intention) and those of the independent variable. Also, descriptive
analysis can not determine which predicting variable(s) are found to be significant
determinants of turnover intention (Hamilton, 1990). Furthermore, it can not provide the
causal relationship of pay satisfaction with its significant attitudinal and behavioral
consequences including turnover intention, nor provide comparisons of the total influence
of pay satisfaction on turnover intention with those of other significant attitudinal and
behavioral consequences, respectively. Therefore, in the following three sections,
therefore, bivriate, multivariate regression and structural equation modeling analyses will
be employed.
A series of t-test analyses along with each of all organizational variables were
conducted to examine whether there is any significant mean difference between the line
probation officers and direct care staff. As depicted in Table 15, there were statistically
significant mean differences in all organizational variables but six. Interestingly,
compared to the line probation officers, the direct care staff tended to have more negative
feelings about turnover intention and stress-related variables, and more positive feelings
or perceptions about the other variables, such as affective commitment and job
satisfaction. These findings indicate the two groups are different, possibly due to the
nature of their work, and thereby further, separate analysis may be required.

39
Table 15. Independent-Samples t-test for Comparing the Means of Line Community
Supervision Officers and Direct Care Staff.
a

b

CSO
Mean (N)

DCS
Mean (N)

Turnover intention

2.763 (2647)

2.485 (580)

0.278

6.622 ***

891

Organizational commitment
Affective commitment
High sacrifice
Lack of alternative

3.164 (2637)
3.209 (2643)
3.265 (2635)

3.347 (575)
3.239 (579)
3.223 (578)

-0.183
-0.030
0.043

-4.200 ***
-0.647
0.936

3210
883
3211

Satisfaction
Overall job satisfaction

3.492 (2646)

3.661 (579)

-0.170

-4.536 ***

3223

Pay
Promotion
Supervision
Benefits
Contingent rewards
Operating procedures
Co-workers
Nature of work
Communication

2.408
2.300
3.859
2.853
2.483
2.436
3.614
3.762
2.873

(2648)
(2631)
(2632)
(2622)
(2635)
(2624)
(2639)
(2637)
(2641)

2.585
2.479
3.853
2.965
2.745
2.971
3.621
3.946
3.047

(579)
(573)
(575)
(576)
(577)
(576)
(573)
(579)
(579)

-0.177
-0.179
0.005
-0.112
-0.263
-0.535
-0.007
-0.185
-0.173

Stress
Role overload
Role conflict
Role ambiguity
Dangerousness
Job stress

3.196
2.782
2.173
2.957
3.198

(2642)
(2639)
(2634)
(2629)
(2644)

2.618
2.688
2.130
2.525
2.759

(578)
(574)
(573)
(578)
(577)

Organizational justice
Distributive justice
Procedural justice

2.502 (2625)
2.829 (2631)

Social support
Participatory Management
Participatory climate
Empowerment

Mean
Difference

t

-5.005
-4.380
0.123
-2.803
-6.048
-13.151
-0.192
-5.061
-3.979

df

***
***

***
***

3225
3202
3205
3196
3210
885
3210
3214
3218

0.579
0.094
0.043
0.432
0.439

12.921 ***
2.311
1.248
11.393 ***
10.017 ***

3218
3211
3205
3205
3219

2.767 (577)
3.004 (576)

-0.265
-0.175

-5.807 ***
-4.628 ***

3200
859

3.532 (2635)

3.616 (577)

-0.085

-2.160 *

3210

2.841 (2628)
3.601 (2632)

3.089 (576)
3.833 (578)

-0.248
-0.232

-6.192 ***
-8.907 ***

3202
806

**
***
***

a

Community Supervision Officer; b Direct-Care Staff
* p < 0.05; ** p < 0.01; *** p < 0.001

Before conducting further analysis, pre-analysis data screening is essential to
secure the accuracy of the data and to prevent any biased result. According to Mertler and
Vannatta (2005), the pre-analysis data screening include the following five data screening
processes: missing data, normality, linearity, and homoscedasticity and outliers. First, the
minimum number of missing values was replaced with the means of the variable. For
handling normality, linearity and homoscedasticity, six different data transformations

40
were utilized to normalize all metric variable including age and tenure. Based on each of
the six data transformations, the kurtosis (peakedness of the distribution) and skewness
(asymmetry of the distribution) values were compared and then selected when the values
were most close to zero, reflective of the normal distribution with the least skewness and
without being too peaked or too flat (Tabachnick & Fidell, 1996).
Table 16 interprets which data transformation more closely approximated a
normal curve of each metric variable. Also, in a preliminary test for residuals as another
way to secure the accuracy of the data and to prevent any biased result, the assumption of
normality, linearity and homoscedasticity was met. As for outliers, another preliminary
test identified 18 extreme outliers by which any results can be very misleading. Therefore
16 cases of the usable responses from 2653 line probation officers were deleted, brining
the data-sample size to 2637. Likewise, 2 cases of the useable responses from 581 direct
care staff were deleted, brining the data-sample size to 579. Further analysis will be
conducted with the data wherein all outliers were deleted.

41
Table 16. Transformation of Metric Variables
Skewness

Kurtosis

Transformation Method

-0.12

-0.68

Square Root

0.42
0.35
0.38

-0.45
-0.75
-0.62

Square
Square
Square

0.24

-0.21

Square

Pay
Promotion
Supervision
Fringe benefits
Contingent rewards
Operating procedures
Co-workers
Nature of work
Communication

-0.12
-0.13
0.76
0.55
-0.21
-0.14
0.31
-0.02
-0.43

-0.54
-0.79
0.11
0.16
-0.62
-0.58
-0.46
-0.55
-0.31

Square Root
Square Root
Cubic
Square
Square Root
Square Root
Square
Square
Square Root

Stress
Role overload
Role conflict
Role ambiguity
Dangerousness
Job stress

-0.40
-0.16
0.01
-0.37
-0.33

-0.29
-0.24
-0.29
-0.20
-0.37

Square Root
Square Root
Square Root
Square Root
Square Root

Organizational justice
Distributive justice
Procedural justice

-0.11
-0.45

-0.70
-0.10

Square Root
Square Root

Social support

0.24

-0.30

Square Root

Participatory management
Participatory climate
Empowerment

0.50
0.42

-0.10
0.34

Square Root
Square Root

Age

-0.09

-0.96

Logarithmic

Tenure in current department

-0.58

-0.44

Logarithmic

Variable
Turnover intention
Organizational commitment
Affective commitment
High sacrifice
Lack of alternative
Satisfaction
Overall job satisfaction

42

Section 4.
Bivariate & Multivariate Regression
Analyses for Line Community
Supervision Officers

43
Bivariate Analyses
To determine the strength and direction of the association between each
predicting variable and turnover intention, three commonly used bivariate analytical
techniques were employed. Three bivariate analyses include Pearson’s zero-order
correlation, independent-samples t-test, and one-way analyses of variance (ANOVA).
Pearson’s zero-order correlation is used to assess the strength and direction of the
relationship between each independent variable and turnover intention. Additionally,
independent-samples t-test and one-way analyses of variance (ANOVA) are used to
assess group differences of turnover intention: The independent-samples t-test is used for
gender differences relating to turnover intention, and the one-way analyses of variance
(ANOVA) is used to evaluate educational-level differences relating to turnover intention.
Table 17. Zero-Order Correlations of Turnover Intention by Both Individual Status and
Organizational Variables among Texas Community Supervision Officers (N = 2,637)
Organizational Variable

Correlation
Coefficient

Individual Status Variable

Correlation
Coefficient

Affective commitment

-0.63 **

Gender

High sacrifice

-0.46 **

Age

-0.21 **

Lack of alternative

-0.14 **

Ethnicity

-0.15 **

Overall job satisfaction

-0.53 **

Martial status

-0.13 **

Pay

-0.47 **

No. of children at home

-0.07 **

Promotion
Supervision

-0.37 **
-0.20 **

0.13 **
-0.13 **

Benefits

-0.29 **

Education level
Tenure in current department
Probation

-0.03

Contingent rewards

-0.38 **

Law enforcement

-0.03

Operating procedures

-0.27 **

Corrections

-0.03

Co-workers

-0.21 **

Parole

-0.02

Nature of work

-0.46 **

Communication

-0.33 **

Role overload

0.22 **

Role conflict

0.31 **

Role ambiguity

0.25 **

Dangerousness

0.18 **

Job stress

0.37 **

Distributive justice

-0.42 **

Procedural justice

-0.35 **

Social support

-0.34 **

Participatory climate

-0.36 **

Empowerment

-0.36 **

** Correlation is significant at the 0.01 level (2-tailed).

0.02

44
Relationships between Turnover Intention and Organizational Variables
As for organizational variables, only Pearson’s zero-order correlation in Table
17 was conducted to assess the strength and direction of the relationship between
variables so that values for the dependent variable, turnover intention, can actually be
predicted based on the values for each of the organizational variables. The strength and
direction of the relationship between two variables is referred to as their correlation, and
the standardized measure is generally known as Pearson’s r or simply r; where r varies
between -1 (a perfect negative relationship) and +1 (a perfect positive relationship). A
correlation coefficient of 0 (r = 0) indicates no relation between the two variables.
According to Davis (1971), correlation coefficients between -0.09 and +0.09 represent a
negligible relationship, those between 0.10 and 0.29 (either positive or negative) indicates
a low relationship, those between 0.30 and 0.49 represent a moderate relationship and
those larger than ± 0.50 indicate a strong relationship between two variables. Note that
“some social scientists call correlations of ± 0.30 strong” (Hamilton, 1990, p. 773).
The correlation matrix in Table 17 reveals that all twenty-three organizational
variables were found to be significantly correlated with turnover intention. Consistent
with existing literature, organizational commitment, overall job satisfaction, job facet
satisfaction, organizational justice, social support and participatory management had
significantly negative effects on a line officer’s turnover intention. Also, all stress-related
variables, role overload, role conflict, role ambiguity, dangerousness of the job and job
stress level were positively correlated with turnover intention.
Interpretatively, officers who reported lower levels of affective and continuance
commitments were more likely to express higher levels of turnover intention. Related to
job satisfaction, those who reported lower levels of both overall job satisfaction and
specific satisfaction with pay, promotion, fringe benefits, contingent rewards, operating
procedures, co-workers, nature of work and communication were less inclined to quit
their employment. In terms of organizational justice, those with lower levels of perceived
distributive justice tended to perceive more unfairness of departmental decisions
concerning the distribution of rewards, such as pay and promotions, more likely leading
to higher levels of turnover intention. Also, those with lower levels of perceived
procedural justice had a tendency to perceive more unfairness of departmental procedures
in decision-making processes for the distribution of rewards, more likely exhibiting
higher inclinations to leave their employment. As officers’ perception of social support
and participatory management decreased, their turnover intention increased. Finally,
officers who found their job roles incompatible, unclear, and demanding tended to feel
more stress, more likely leading to higher levels of turnover intention.
Interestingly, all organizational variables were found to be statistically correlated
with turnover intention. As the correlation matrix illustrates, however, the correlation
coefficient for each variable demonstrates a different strength in association with the
dependent variable of turnover intention. Using the criteria Davis (1971) suggested, the
strength of the correlation coefficients for all organizational variables can be categorized
into three groups: the first group with strong relationships, the second group with

45
moderate relationships and the final group with weak relationships. The first group
includes two variables: affective commitment (r = -0.63) and overall job satisfaction
(r = -0.53), exceeding the cut-off point of ± 0.50. Affective commitment was found to
have the strongest relationship to turnover, whereas overall job satisfaction turned out to
be the second strongest variable in association with turnover intention. The second group
representing moderate relationships with turnover intention includes the following
thirteen variables: high sacrifice commitment, pay, promotion, contingent rewards, nature
of work, communication, role conflict, job stress, distributive justice, procedural justice,
social support, participatory climate and empowerment. The third and last group
representative of weak relationships with turnover intention includes the other eight
variables: lack of alternative commitment, supervision, fringe benefits, operating
procedures, co-workers, role overload, role ambiguity and dangerousness of the job.
In the second group, the correlation coefficients for high sacrifice commitment,
pay, nature of work and distributive justice were -0.46, -0.47, -0.46 and -0.42; well
exceeding -0.40 and approaching the cut-off point of ± 0.50. Considering their relatively
high strength, these four variables in addition to affective commitment and overall job
satisfaction seem to have substantial relationships with an officer’s turnover intention. In
addition, among these six variables found to have substantial association with turnover
intention, affective commitment was found to be the strongest in association with
turnover intention, followed by overall job satisfaction, pay, high sacrifice commitment
(r = -0.461), nature of work (r = -0.457) and distributive justice.

46
Relationships between Turnover Intention and Individual Status Variables
First of all, Pearson’s zero-order correlation was utilized to assess the strength
and direction of the relationship between variables so that values for the dependent
variable, turnover intention, can actually be predicted based on the values for each of the
individual status variables. For the Pearson’s zero-order correlation, the following seven
non-metric individual status variables were dummy-coded: Gender (0 = female,
1 = male); ethnicity (0 = non-Caucasian, 1 = Caucasian); marital status (0 = single,
1 = married); and prior employment with probation (0 = no, 1 = yes), law enforcement
(0 = no, 1 = yes), corrections (0 = no, 1 = yes) and parole (0 = no, 1 = yes). The
correlation matrix in Table 17 reveals that, among all eleven individual status variables,
gender and tenure in current department and prior employment in probation, law
enforcement, corrections and parole were not found to be significantly correlated with
turnover intention.
In contrast, the other six individual status variables were found to be statistically
correlated with turnover intention. These significantly correlated individual status
variables include age, ethnicity, marital status, number of children at home, educational
level and tenure in current department. While education level was found to have
statistically significant positive association with a line officer’s turnover intention, age,
ethnicity, marital status, number of children at home, and tenure in their current
department had a statistically significant negative association with turnover intention.
Among these six variables found to have statistically significant associations with
turnover intention, age (r = -0.21) was found to be the strongest in association with
turnover intention, followed by ethnicity (r = -0.15), tenure in current department
(r = -0.132), educational level (r = 0.126), marital status (r = -0.125) and number of
children at home (r = -0.07).
Younger officers with less seniority were more likely to express higher levels of
turnover intention than were older officers with more seniority. Race was significantly
associated with turnover intention, with minorities expressing greater turnover intention
than Caucasians. In addition, single officers with no children or fewer children at home
were more likely to express higher levels of turnover intention than married officers with
a greater number of children at home. Finally, education level was positively correlated
to turnover intention: As educational level increased, turnover intention increased as well.
Despite the significance of the correlation coefficients of these five individual
status variables, age, ethnicity, marital status, educational level, and tenure in current
department fell between 0.10 and 0.29 as an absolute value, and thereby demonstrated
weak relationships with turnover intention. In addition, the correlation coefficient for the
number of children at home (r = -0.07) was found to be significant but was less than
-0.10, thereby having a negligible association with turnover intention. These findings
suggest that these five variables do not seem to have substantial relationships, but rather
weak or negligible relationships with an officer’s turnover intention.

47
Even with weak or negligible relationships between these six individual status
variables and turnover intention, further investigation for a significant mean difference of
turnover intention between/among subgroups of each of the six variables is valuable.
Further investigation may provide important managerial information in preventing
turnover-related problems with certain individual status groups, restoring their
effectiveness, and the efficiency of Texas probation. Accordingly, independent-samples ttest analysis and/or one-way ANOVA tests were utilized to determine a significant mean
difference between subgroups of each variable and turnover intention (McKean & Byers,
2000). For a one-way ANOVA test, the metric variables of age and tenure in current
department were recorded into categorical variables by dividing the lowest to the highest
ranges of each variable into intervals.
The finding from the zero-order correlation test indicates that younger officers
were more likely than older officers to exhibit higher levels of turnover intention. As
Figure 1 illustrates, ranging from the lowest age of 20 years to the highest of 73 years, the
continuous metric variable of age was reduced to nine groups: the youngest group (20-24
years) to the oldest group (60 years or more). Although not listed in the figure, age group
differences were tested using one-way ANOVA. Based on p < 0.001, The ANOVA
reveal that a significant difference (F = 15,438, df =8), existed when comparing turnover
intention among the nine age groups.
However, this statistically significant F value cannot specify which groups differ
from each other significantly, and thereby the Tukey HSD Post-Hoc test was utilized to
compare each group mean to every other group mean (Hair, Black, Anderson, & Tatham,
2006). By using this post-hoc procedure, many significant mean differences were found
between groups. Briefly, three younger groups—group 1 (20-24 years old; Mean = 3.07);
group 2 (25-29 years old; Mean = 3.06); and group 3 (30-34 years old; Mean = 2.89)—
were found significantly different from the other, older groups. Supportive of the finding
from the zero-order correlation test, these findings indicate that high turnover intention
was strongly prevalent among officers whose age range were somewhere between 20 and
34 years old. This age-range group accounts for 42.8% (991 out of 2,618) of the nonmanagerial and non-supervisory line officers.

48
Figure 1. Turnover Intention by Age Group among Texas Community Supervision Officers

3.2

Turnover Intention

3.0
2.8
2.6
2.4
2.2
2.0
20 - 24

25 - 29

30 - 34

35 - 39

40 - 44

45 - 49

50 - 54

55 - 59

60+

Age Group (years)
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

49
The finding from the zero-order correlation test suggests that minority officers
were more likely than Caucasian officers to express higher levels of turnover intention.
This finding is based upon the binary variable of ethnicity. Figure 2 indicates that three
minority groups exhibit much higher levels of turnover intention than Caucasian officers
and that in particular African American officers shows the highest levels of turnover
intention. Ethnic group differences were tested using one-way ANOVA. The ANOVA
reveal that a significant difference, F (3) = 21,319, p < 0.001, existed when comparing
turnover intention among the four ethnic groups.
By using the Tukey HSD Post-Hoc test, significant mean differences (p < 0.05)
were found between the African American group (Mean = 2.97, N = 477) and the
Caucasian group (Mean = 2.61, N = 1,241), between the Other ethnic groups (Mean =
2.95, N = 75) and the Caucasian group, and between the Hispanic group (Mean = 2.85,
N = 831) and the Caucasian group. Consistent with the finding from the zero-order
correlation test, these results indicate that minority officers accounting for 52.7% (1,383
out of 2,624) of the non-managerial and non-supervisory line officers demonstrated much
higher levels of turnover intention than Caucasian officers.
Figure 2. Turnover Intention by Ethnic Group among Texas Community Supervision Officers

2.97

2.95

3.0

Turnover Intention

2.85

2.8
2.61
2.6

2.4
Caucasian

Hispanic

African American

Other

Ethnic Group
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

50
The findings from the zero-order correlation test indicates that officers with
higher levels of education were more likely to express higher levels of turnover intention
than their counterparts. As Figure 3 illustrates, the original five educational levels in the
survey were reduced to three. Although not indicated in Figure 3, educational background
group differences were tested using one-way ANOVA. The ANOVA indicates that a
significant difference, F (2) = 21,703, p < 0.001, existed when comparing turnover
intention among the three educational background groups.
Utilizing the Tukey HSD Post-Hoc procedure, significant mean differences
(p < 0.05) were found: between the Master’s degree group (Mean = 2.99, N = 369) and
the Bachelor’s degree group (Mean = 2.75, N = 2,123); between the Master’s degree
group and the Associate degree group (Mean = 2.38, N = 135); and, between the
Bachelor’s degree group and the Associate degree group. Consistent with the finding
from the zero-order correlation test, the results indicate that as education level increased,
turnover intention increased proportionately.
Figure 3. Turnover Intention by Educational Level among Texas Community Supervision
Officers

2.99
3.0

2.75

Turnover Intention

2.8
2.6

2.38

2.4
2.2
2.0
Associate Degree or less

Bachelor's Degree

Master's Degree or more

Educational Level
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

51
As Figure 4 illustrates, single officers (Mean = 2.90, N = 1,079) demonstrate
higher turnover intention than married officers (Mean = 2.66, N = 1,544). Even if not
illustrated in the figure, marital status difference was tested using an independent samples
t-test. Based upon a mean difference (0.243) of marital status groups, the two groups did
significantly differ at p < 0.001 (t = 6.435, df = 2,621). Consistent with the finding from
the zero-order correlation test, the result evidences that single officers accounting for
41.1% of the non-managerial and non-supervisory line officer population were more
likely than married officers to exhibit turnover intention.
Figure 4. Turnover Intention by Marital Status among Texas Community Supervision Officers

2.90

Trunover Intention

3.0

2.66

2.8

2.6

2.4
Single

Married

Marital Status
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

52
The finding from the zero-order correlation test indicates that officers with no
children, or fewer children at home were more likely to express higher levels of turnover
intention than were officers with more children at home. In Figure 5, turnover intention
decreases as the number of children at home increases. A significant difference,
F (3) = 5,339, p < 0.05, existed when turnover intention was compared among the four
groups. More specifically, the Tukey’s HSD Post-Hoc test on this significant ANOVA
measure was conducted to determine any significant difference among the four groups.
The Tukey’s HSD test indicates that significant differences were found between the group
without children (Mean = 2.83, N =1,199) and the two-children group (Mean = 2.67,
N = 576), and between the group without children and the three-children-or-more group
(Mean = 2.64, N = 236). It should be noted that the childless group accounts for 45.7% of
the population. Consistent with the findings from the zero-order correlation test, the
results indicate that as the number of children at home increased, turnover intention
decreased.
Figure 5. Turnover Intention by Number of Children among Texas Community Supervision
Officers

3.0

Turnover Intention

2.83
2.76
2.8

2.67

2.64

2.6

2.4
None

1

2

3 or more

Number of Children at Home
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

53
The finding from the zero-order correlation test indicates that officers with less
seniority were more likely to express higher levels of turnover intention than were those
with more seniority. As Figure 6 illustrates, the continuous metric variable of tenure—
ranging from the lowest at 0.08 years to the highest at 34 years—was divided into seven
groups, from the least senior group (0-3 years service) to the most senior (19+ years
service). There is a slight negative relationship with turnover intention, up to 16-18 years
of service. This group demonstrated the lowest turnover intention. However, the curve
rose for the next group, with19+ years of tenure. Even though there was some variation in
the curve as of the 13–15 years-tenure group, the test for linearity indicates the significant
linear relationships (p < 0.001) between seniority and turnover intention. This finding
suggests that turnover intention tends to start rising again among the most senior officers
nearing their retirement. In addition, tenure group differences were tested using one-way
ANOVA. The ANOVA reveal that a significant difference, F (6) =11,349, p < 0.001,
existed when comparing turnover intention among the seven tenure groups.
Figure 6. Turnover Intention by Tenure Group among Texas Community Supervision Officers

Turnover Intention

3.0

2.8

2.6

2.4
0-3

4-6

7-9

10 - 12

13 - 15

16 - 18

19+

Tenure Group (years of service)
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

54
The Tukey’s HSD Post-Hoc test was performed on this significant ANOVA
measures to determine any significant difference among the seven groups. The Tukey’s
HSD test indicates that significant differences were found between group 1 (0-3 years;
Mean = 2.92) and group 3 (7-9 years; Mean = 2.71) and the other higher tenure groups.
Also, only one additional significant difference was found between group 2 (4-6 years;
Mean = 2.75) and group 6 (16-18 years; Mean = 2.49). Supporting the finding from the
zero-order correlation test, this finding indicates that high turnover intention was
strongly prevalent among officers whose tenure range was somewhere between 0-6 years.
This tenure range group accounts for 55.4% (1,447 out of 2,614) of the non-managerial
and non-supervisory line officers.
Summary
From the findings from all bivariate analyses, three important results are worthy
of mention. First, of particular interest is the finding that organizational factors were
more important than individual status factors in association with turnover intention. All
organizational variables were found to be significantly associated with turnover intention,
although the strength of the relationships with turnover ranged from weak to strong. In
contrast, the individual status variables out of the eleven were found to be significantly
associated with turnover intention and their associations with turnover intention were
found to be weak or negligible. Second, among all organizational variables, affective
commitment, overall job satisfaction, pay, high sacrifice commitment, nature of work and
distributive justice appear to have substantial associations with an officer’s turnover
intention. Finally, among the six individual status variables, age was found to be the
strongest in association with turnover intention. High turnover intention was strongly
prevalent in the 20-34 years age range (equivalent to 42.8% of the Community
Supervision Officers population), and tenure group analyses indicate the similar pattern
(between 0-6 years tenure, equivalent to 55.4% of the Community Supervision Officers
population).

55
Multivariate Regression Analyses
Bivariate analyses were used to provide the direction and strength of each
individual variable in its association with turnover intention. Bivariate analysis is
valuable in summarizing the relationship between each of the independent variables and
turnover intention at once, but it cannot test any significant effect of one independent
variable on turnover intention after holding all other independent variables constant
(Hamilton, 1990). In response, three stepwise Ordinary Least Square (OLS) regression
analyses were employed to assess whether various individual and organizational variables
influence a non-managerial and non-supervisory line officer’s turnover intention (see
Table 18). In each equation of the table, standardized regression coefficient, generally
known as Beta coefficient, varies between -1 (a perfect negative prediction power) to 1 (a
perfect positive prediction power). A Beta coefficient of 0 indicates no power in
predicting turnover intention.
The three OLS analyses are necessary to seek for any evidence to support the
findings from the previous zero-order correlation, independent-Samples t-test or oneway ANOVA analyses and examine whether or not the findings are maintained after
statistically controlling for the effects of individual status variables on a line officer’s
turnover intention. In each equation, the turnover intention was the dependent variable.
Accordingly, the first equation examined the effects of the individual status variables, as
an indication of individual factors, on an officer’s turnover intention. On the other hand,
the second equation determined the effects of only organizational variables, indicative of
organizational factors, on an officer’s turnover intention. These two separate equations
were designed to compare which factors—individual or organizational—have more
influence on a line officer’s turnover intention. Finally, the third equation as the most
complete equation model, determined whether the organizational factors are still
statistically significant, after controlling for the effects of the individual factors.
Two multicollinearity diagnostics were conducted to determine any violation of
the multivariate regression assumption, that there are no highly correlated independent
variables measuring the same thing (Hair et al., 2006). The simplest way of finding
multicollinearity is to check correlation (Fox, 1981; Hy, Feig & Regoli, 1983). None of
the previous Pearson’s correlations (Table 17) among all independent variables were
higher than ± 0.7. The second method of identifying multicollinearity is to examine all
individual variation inflation factor (VIF) scores for each individual variable: a VIF of
1.0 reflects total independence, and if a VIF is higher than 1.0, more collinearity is
reflected in the variable (Hair et al., 2006). According to Stevens (1992), all independent
variables’ coefficients in a multivariate regression model are unbiased and efficient when
all VIF scores do not exceed 10. As seen in Table 18, none of the VIF exceeded 2.18.
Taken together, neither the results from zero-order correlations nor the results from the
VIF methods indicate substantial evidence that multicollinearity is an issue in this
analysis and does not substantially alter any of the findings or subsequent conclusions
drawn from the analysis.

56
Table 18. The Determinants of Turnover Intention among Texas Community Supervision
Officers (N = 2,637)
Equation 1

Included Variables

Beta

b

VIF

Equation 2
c

Beta

b

Equation 3
c

VIF

Betab

VIFc

Individual Status Variables
Gender (male = 1)

0.066 ***

1.041

Age

-0.189 ***

1.061

-0.104 ***

1.109

Ethnicity (Caucasian = 1)

-0.132 ***

1.043

-0.046 ***

1.081

Martial statusa

-0.071 **

1.213

-0.038 **

1.052

No. of children at home

-0.041 *

1.163
0.043 **

1.060

Education level

0.138 ***

1.007

Organizational Variables
Affective commitment

-0.373 ***

1.837

-0.356 ***

1.855

High sacrifice

-0.262 ***

1.134

-0.241 ***

1.167

Overall job satisfaction

-0.190 ***

2.143

-0.170 ***

2.169

Pay

-0.161 ***

1.508

-0.144 ***

1.558

Promotion

-0.048 **

1.520

-0.054 ***

1.519

Co-workers

-0.034 *

1.457

-0.041 **

1.294

Nature of work

-0.072 ***

2.150

-0.082 ***

2.176

Communication

-0.050 **

1.731

-0.045 **

1.654

0.057 ***

1.528

0.050 **

1.513

Distributive justice

-0.073 ***

1.823

-0.069 ***

1.823

Social support

-0.040 *

1.925

Job stress

a

R -square =

0.090

0.595

0.612

F =

23.809

351.198

295.353

Significance =

0.000

0.000

0.000

1 = currently married; b Standardized Coefficients; c Variance Inflation Factor

* p < 0.05; ** p < 0.01; ** p < 0.001

Equation 1 of Table 18 examines only the impact of individual status variables on
an officer’s turnover intention and shows a significant and good model fit. The chisquare test of the model indicates that Equation 1 significantly predicted a line officer’s
turnover intention (χ2 = 23.809, df = 6, p < 0.001). Six individual status variables were
found to have statistically significant effects on an officer’s turnover intention. The
remaining five individual status variables—tenure in current department and prior

57
employment in probation, law enforcement, corrections, and parole—were excluded from
the final best-fit equation since each of them didn’t have a statistically significant high
partial correlation (Hair et al., 2006).
The six statistically significant determinants of turnover intention were: gender,
age, ethnicity, marital status, number of children at home, and education level.
Specifically, males and single officers were more likely to express higher levels of
turnover intention than females and married officers. Race was significantly related to
turnover intention, with minorities expressing more turnover intention than Caucasians.
In addition, younger officers and officers either without children, or with fewer children
at home, were more likely to express higher levels of turnover intention than were older
officers and officers with more children at home. Finally, education level was positively
related to turnover intention: the higher the education level, the higher the turnover
intention. However, despite the significance of the standardized coefficients of all six
individual status variables, only 9% of the variance in the dependent variable, turnover
intention, was accounted for (R-square = 0.090).
Equation 2 examines only the impact of effects of organizational variables on an
officer’s turnover intention and shows a significant and good-model fit (χ2 = 351.198,
df = 11, p < 0.001) in predicting turnover intention. Out of twenty-three organizational
variables, eleven variables based upon each statistically significant high partial
correlation were included in Equation 2. The remaining twelve variables were excluded
from the final best-fit equation since they did not meet the entry significance. The
excluded variables include a lack of alternatives, supervision, fringe benefits, contingent
rewards, operating procedures, role overload, role conflict, role ambiguity, dangerousness
of the job, procedural justice, participatory climate, and empowerment.
The eleven statistically significant determinants of turnover intention were
affective commitment, high sacrifice commitment, overall job satisfaction, pay,
promotion, co-workers, nature of work, communication, job stress, distributive justice
and social support. Particularly, job stress was positively related to turnover intention: as
officers’ levels of job stress increased, their turnover intention also increased. On the
other hand, the other determinants were negatively related to turnover intention.
Specifically, officers who reported lower levels of affective commitment, high sacrifice
commitment, and overall job satisfaction were more likely to express higher levels of
turnover intention. In addition, those who reported lower levels of satisfaction with pay,
promotion, co-workers, nature of work, and communication were more inclined to leave.
Finally, those who reported lower levels of perceived distributive justice and social
support were more likely to have an inclination to leave.
Two additional findings relating to Equation 2 are important. First, the included
eleven significant independent variables accounted for 59.5% of the variance in the
dependent variable, turnover intention. This portion of variance explained by Equation 2
(R-square = 0.595) is almost 6.6 times higher than that explained by Equation 1 (R-square
= 0.090). This finding suggests that organizational factors have a more substantial
contribution to make in predicting an officer’s turnover intention than individual factors.

58
Second, the standardized regression coefficients for promotion, co-workers, nature of
work, communication, job stress, distributive justice, and social support were all
significant, but lower than ± 0.1. On the other hand, the standardized regression
coefficients for affective commitment, high sacrifice commitment, overall job satisfaction
and pay were -0.373, -0.262, -0.190, and -0.161, respectively. All coefficients well
exceeded ± 0.1. These four organizational variables, therefore, appear to have both
statistical and substantive significance in predicting an officer’s turnover intention. Given
the standardized regression coefficients, affective commitment had the strongest
statistically significant, negative effect on turnover intention, followed by high sacrifice
commitment, overall job satisfaction, and pay.
Equation 3 in Table 18 is the final and most complete best-fit regression model.
Here the individual status variables are treated as statistical control variables to mainly
determine whether the significant organizational variables found in Equation 2 are still
statistically significant after controlling for the effects of the individual status variables.
Equation 3 shows a significant, and a good-model fit: The chi-square test of the model
indicates that Equation 3 significantly predicted an officer’s turnover intention
(χ2 = 295.353, df = 14, p < 0.001). The proportion of variance explained by Equation 3
(R-square = 0.612) is minimally higher than that explained by Equation 2 (R-square =
0.595) and is almost 6.8 times higher than that explained by Equation 1 (R-square =
0.090). This finding indicates that the organizational variables have a much greater
contribution to make in predicting an officer’s inclination to leave even after controlling
for the effects of the individual status variables.
Fourteen variables based upon each statistically significant high partial correlation
were included in Equation 3: four individual status variables and ten organizational
variables. In comparison with Equation 1, four individual status variables, age, ethnicity,
marital status and education level were still included as being statistically significant,
whereas gender and the number of children at home were excluded from the final best-fit
equation after organizational factors were included in Equation 3. This finding indicates
that the effects of gender and the number of children at home on turnover intention are
indirect and mediated through organizational factors.
In addition, Equation 3 is statistically supportive of the direction of the four
significant individual status variables found in Equation 1: single, younger, and minority
officers with higher levels of education and no children, or fewer children at home were
more likely to express higher levels of turnover intention than their counterparts.
However, despite the significance of the standardized coefficients of all four significant
individual status variables, the strength of all standardized coefficients were largely
reduced by including organizational factors in the final model. Also, only the
standardized coefficient for age (Beta = -0.104, p < 0.001) exceeded the cut-off point of
± 0.1. This finding suggests that ethnicity, marital status and educational level contribute
significantly, but weakly, to predict an officer’s turnover intention, whereas age has its
strongest direct effect on turnover intention. That is, age has a much more substantial
contribution to make in predicting an officer’s turnover intention than other individual
factors.

59
As for organizational factors, after entering the individual factors as control
variables into the final regression equation, the effect of social support on turnover
intention was excluded from the final model. This finding indicates that social support
didn’t have significant contribution to predicting turnover intention beyond the predictive
power of the control variables. The direction and strength of each of the ten significant
organizational variables in Equation 3 are consistent with the findings in Equation 2. All
included organizational variables except for job stress were negatively related to turnover
intention. Utilizing the cut-off point of ± 0.1, the standardized regression coefficients for
promotion, co-workers, nature of work, communication, job stress, and distributive
justice were all significant but lower than ± 0.1. This finding indicates that these
organizational variables contribute significantly, but weakly, to predict an officer’s
turnover intention.
In contrast, the standardized regression coefficients for affective commitment,
high sacrifice commitment, overall job satisfaction, and pay were -0.356, -0.241, -0.170,
and -0.144 respectively; well exceeding the cut-off point. Like the findings of Equation 2,
these four organizational variables substantially contribute to predict turnover intention.
Finally, these findings show affective commitment to be the strongest predictor of turnover intention, followed by high sacrifice commitment, overall job satisfaction and pay.
Summary
Taken together and consistent with the findings from the bivariate analyses, these
regression analyses reveal that organizational factors, rather than individual status factors,
have a substantially greater contribution to make in predicting an officer’s inclinations to
leave employment. Among the organizational factors, affective commitment, high
sacrifice commitment, overall job satisfaction, and pay satisfaction each have a
significant direct effect on turnover intention after holding all other independent variables
constant. In addition age, among the four significant individual status predictors of
turnover intention, has its strongest direct effect after controlling for all other independent
variables, and has a much more substantial contribution to make in predicting an officer’s
turnover intention than other individual factors. This finding is consistent with the overall
finding from the bivariate analyses for individual status factors. Recall the finding from
the Tukey’s HSD Post-Hoc test for the nine age groups, that high turnover intention was
strongly prevalent among officers whose age range was somewhere between 20-34 years.
This age range group accounts for 42.8% (991 out of 2,618) of the non-managerial and
non-supervisory line officers.

60

Section 5.
Bivariate & Multivariate Regression
Analyses for Direct-Care Staff

61
Bivariate Analyses
The previous descriptive analyses were used to provide a brief description and
univariate statistics, such as variable frequencies, means, and standard deviations for each
variable. Although useful to summarize each individual variable, descriptive analysis
cannot explore any differences and relationships between the values of the dependent
variable (turnover intention) and those of the independent variable of the interest
(Hamilton, 1990). To determine the association between two variables, three commonly
used bivariate analytical techniques—Pearson’s zero-order correlation, independentSamples t-test and one-way analyses of variance (ANOVA)—were employed.
Table 19. Zero-Order Correlations of Turnover Intention by Both Individual Status and
Organizational Variables Among Texas Direct Care Staff (N = 579)
Organizational Variable

Correlation
Coefficient

Individual Status Variable

Correlation
Coefficient

Affective commitment

-0.55 **

Gender

High sacrifice

-0.43 **

Age

-0.11 **

Lack of alternative

-0.13 **

Ethnicity

-0.09 *

Overall job satisfaction

-0.42 **

Martial status

-0.10 *

Pay

-0.45 **

No. of children at home

-0.05

Promotion
Supervision

-0.38 **
-0.22 **

Education level

Benefits

-0.29 **

Tenure in current department
Probation

0.05
-0.15 **

Contingent rewards

-0.39 **

Law enforcement

Operating procedures

-0.30 **

Corrections

-0.01

Co-workers

-0.21 **

Parole

-0.03

Nature of work

-0.40 **

Communication

-0.37 **

Role overload

0.24 **

Role conflict

0.30 **

Role ambiguity

0.24 **

Dangerousness

0.17 **

Job stress

0.41 **

Distributive justice

-0.38 **

Procedural justice

-0.35 **

Social support

-0.35 **

Participatory climate

-0.38 **

Empowerment

-0.32 **

* Correlation is significant at the 0.05 level (2-tailed).
** Correlation is significant at the 0.01 level (2-tailed).

0.16 **

-0.02
0.06

62
Relationships between Turnover Intention and Organizational Variables
for Direct-Care Staff
As for organizational variables, only Pearson’s zero-order correlation in Table
19 was conducted to assess the strength and direction of the relationship between
variables so that values for the dependent variable—turnover intention—can actually be
predicted based on the values for each of the organizational variables. The strength and
direction of the relationship between two variables is referred to as their correlation, and
the standardized measure is generally known as Pearson’s r or simply r: r varies between
-1 (a perfect negative relationship) and +1 (a perfect positive relationship). A correlation
coefficient of 0 (r = 0) indicates no relation between the two variables. According to
Davis (1971), correlation coefficients between -0.09 and +0.09 represent a negligible
relationship; those between 0.10 and 0.29 (either positive or negative) indicates a low
relationship; those between 0.30 and 0.49 represent a moderate relationship; and, those
greater than ± 0.50 indicate a strong relationship between two variables. Note that “some
social scientists call correlations of ± 0.30 strong” (Hamilton, 1990, p. 773).
The correlation matrix in Table 19 reveals that, all twenty-three organizational
variables were found to be significantly correlated with turnover intention. Consistent
with existing literature, organizational commitment, overall job satisfaction, specific job
satisfaction, organizational justice, social support, and participatory management had
significantly negative effects on direct-care staff’s turnover intention. Also, all stressrelated variables, role overload, role conflict, role ambiguity, dangerousness of the job,
and job stress level were positively correlated with turnover intention.
Interpretatively, direct-care staff who reported lower levels of affective and
continuance commitments were more likely to express higher levels of turnover intention.
Related to job satisfaction, those who reported lower levels of both overall job
satisfaction and specific satisfaction with pay, promotion, fringe benefits, contingent
rewards, operating procedures, co-workers, nature of work, and communication were less
inclined to leave their employment. Related to organizational justice, those with lower
levels of perceived distributive justice tended to perceive greater unfairness in
departmental decisions concerning the distribution of rewards, pay and promotions, likely
leading to higher levels of turnover intention. Also, those with lower levels of perceived
procedural justice had a tendency to perceive more unfairness of departmental procedures
in the decision-making process for the distribution of rewards, more likely exhibiting
higher inclinations to leave their employment. Regarding social support and participatory
management, as direct-care staffs’ perception of social support and participatory
management decreased, their turnover intention increased. Finally, staff who found their
job roles incompatible, unclear, and demanding tended to feel more stress, likely leading
to higher levels of turnover intention.
Interestingly, all organizational variables were found to be statistically correlated
with turnover intention. As the correlation matrix illustrates, however, the correlation
coefficient for each variable indicates a different strength in association with the
dependent variable of turnover intention. Using the criteria Davis (1971) suggested, the

63
strength of the correlation coefficients for all organizational variables can be categorized
into three groups: the first group with strong relationships, the second group with
moderate relationships and the final group with weak relationships. The first group
includes only one variable: affective commitment (r = -0.55), exceeding the cut-off point
of ± 0.50. This finding suggests that affective commitment was found to have the
strongest relationship to turnover. The second group representing moderate relationships
with turnover intention includes the following fifteen variables: high sacrifice
commitment, overall job satisfaction, pay, promotion, contingent rewards, operating
procedures, nature of work, communication, role conflict, job stress, distributive justice,
procedural justice, social support, participatory climate, and empowerment. The third and
last group, representing weak relationships with turnover intention, includes the other
seven variables: lack of alternative commitment, supervision, fringe benefits, co-workers,
role overload, role ambiguity, and dangerousness of the job.
In the second group, the correlation coefficients for high sacrifice commitment,
overall job satisfaction, pay, nature of work and job stress were -0.43, -0.42, -0.45, -0.40,
and 0.41 respectively, exceeding ± 0.40 and approaching the cut-off point of ± 0.50.
Considering their relatively high strength, these five variables in addition to affective
commitment seem to have substantial relationships with direct-care staff’s turnover
intention. In addition, among these six variables found to have substantial association
with turnover intention, affective commitment was found to be the strongest in
association with turnover intention, followed by pay, high sacrifice commitment, overall
job satisfaction, job stress, and nature of work.

64
Relationships between Turnover Intention and Individual Status Variables for
Texas Direct-Care Staff
Pearson’s zero-order correlation was utilized to assess the strength and direction
of the relationship between variables so that values for the dependent variable, turnover
intention, can actually be predicted based on the values for each of the individual status
variables. For the Pearson’s zero-order correlation, the following seven non-metric
individual status variables were dummy-coded: Gender (0 = female, 1 = male); ethnicity
(0 = non-Caucasian, 1 = Caucasian); marital status (0 = single, 1 = married); and, prior
employment with probation (0 = no, 1 = yes), law enforcement (0 = no, 1 = yes),
corrections (0 = no, 1 = yes), and parole (0 = no, 1 = yes). The correlation matrix in Table
19 reveals that among all eleven individual status variables, the number of children at
home, education level, and prior employment in probation, law enforcement, corrections
and parole were not found to be significantly correlated with turnover intention.
In contrast, the other five individual status variables were found to be statistically
correlated with turnover intention. These significantly correlated individual status
variables include gender, age, ethnicity, marital status, and tenure in current department:
While gender was found to have statistically significant positive association with a staff
member’s turnover intention, age, ethnicity, marital status, and tenure in current
department had a statistically significant, negative association with turnover intention.
Among these five variables found to have statistically significant associations with
turnover intention, gender (r = 0.16) was found to be the strongest in association with
turnover intention, followed by tenure in current department (r = -0.15), age (r = -0.11),
marital status (r = -0.10), and ethnicity (r = -0.09).
Male staff were more likely than female staff to demonstrate higher turnover
intention. Younger staff members with less seniority were more likely to express higher
levels of turnover intention than were older staff members with more seniority. In
addition, single staff were more likely to express higher levels of turnover intention than
were married staff. Finally, race was significantly associated with turnover intention, with
minorities expressing more turnover intention than Caucasians.
Despite the significance of the correlation coefficients of these five individual
status variables, the correlation coefficients for gender, tenure in current department, age
and marital status were somewhere between 0.10 and 0.29 in an absolute value, being
closer to 0.10 in an absolute value and thereby showed very weak relationships with
turnover intention. Also, the correlation coefficient for ethnicity (r = -0.09) was less than
-0.10, having a negligible association with turnover intention. The findings suggest that
these five variables do not seem to have substantial relationships, but instead weak or
negligible ones with direct-care staff’s turnover intention.
Even with weak or negligible relationships between these five individual status
variables and turnover intention, further investigation for a significant mean difference of
turnover intention between and among subgroups of each of the five variables is valuable
in providing important managerial information. This information may be used to prevent

65
turnover-related problems with certain individual status groups. In restoring the
effectiveness of these status groups; thereby recovering some efficiency to Texas
probation. Accordingly, independent-Samples t-test analysis and/or one-way ANOVA
tests were utilized to determine a significant mean difference between subgroups of each
variable and turnover intention (McKean & Byers, 2000). For a one-way ANOVA test,
the metric variables of age and tenure in current department were recorded into
categorical variables by dividing the lowest to the highest ranges of each variable into
intervals.
As Figure 7 illustrates, male staff (Mean = 2.37, N = 240) demonstrate lower
turnover intention than female staff (Mean = 2.66, N = 339). Even if not illustrated in the
figure, gender difference was tested using an independent samples t-test. Based upon a
mean difference (-0.29) of marital status groups, the two groups did significantly differ at
p < 0.001 (t = -3.911, df = 577). Consistent with the finding from the zero-order
correlation test, the result evidences that male direct-care staff, 41.5% of the direct-care
staff population, were more likely than female staff to express turnover intention.
Figure 7. Turnover Intention by Gender among Texas Direct Care Staff

2.66

Turnover Intention

2.8

2.6

2.37
2.4

2.2
Female

Male

Gender
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

66
The finding from the zero-order correlation test indicates that younger staff
members were more likely than older staff members to exhibit higher levels of turnover
intention. As Figure 8 illustrates, ranging from the lowest age of 20 years to the highest
of 75, the continuous metric variable of age was reduced to nine groups: the youngest
group (20-24 years) to the oldest group (60+ years). Graphically, age group shows a
slight negative relationship with turnover intention up to the age group (40-44 years) in
which turnover intention was the lowest at which point, however, the overall curve
slightly rose from the age group (45-49 years). Even though there was some variation in
the curve as of the 45–49 age group, the test for linearity indicates the significant linear
relationships (p < 0.001) between age and turnover intention. This finding suggests that
turnover intention tends to start rising again among the older staff members for their early
retirement.
In addition, age group differences were tested using one-way ANOVA. Based on
p < 0.01, The ANOVA reveal that a significant difference (F = 2,611, df =8), existed
when comparing turnover intention among the nine age groups. To specify which groups
differ from each other significantly, the Tukey HSD Post-Hoc test was utilized to compare
each group mean to each other group mean. However, this post-hoc procedure indicates
no significant mean difference between groups. This finding suggests that younger age
groups were more likely than older age groups to exhibit higher levels of turnover
intention but, compared to the older age groups, such high turnover intention was not
statistically prevalent among the younger age groups.
Figure 8. Turnover Intention by Age Group among Texas Direct Care Staff

3.0

Turnover Intention

2.8

2.6

2.4

2.2

2.0
20 - 24

25 - 29

30 - 34

35 - 39

40 - 44

45 - 49

50 - 54

55 - 59

60+

Age Group (years)
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

67
The finding from the zero-order correlation test suggests that minority staff
members were more likely than Caucasian staff members to express higher levels of
turnover intention. This finding is based upon the binary variable of ethnicity. Figure 9
indicates that three minority groups exhibit much higher levels of turnover intention than
Caucasian staff, and African American staff in particular demonstrate the highest levels
of turnover intention. Ethnic group differences were tested using one-way ANOVA. The
ANOVA reveal that a significant difference, F (3) = 2,183, p < 0.05, existed when
comparing turnover intention among the four ethnic groups.
By using the Tukey HSD Post-Hoc test, only one significant mean difference
(p < 0.05) was found between the African American group (Mean = 2.67, N = 123), and
the Caucasian group (Mean = 2.39, N = 275). There was no significant mean difference
between the Hispanic and the Caucasian group, and between the Other ethnic groups and
the Caucasian group, respectively. The findings indicate that African American staff
(accounting for 21.5% of the staff population) were more likely to express higher levels
of turnover intention, relative to the Caucasian group.
Figure 9. Turnover Intention by Ethnic Group among Texas Direct Care Staff

Turnover Intention

2.8

2.67

2.53

2.53
2.6

2.40
2.4

2.2
Caucasian

Hispanic

African American

Other

Ethnic Group
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

68
As Figure 10 illustrates, single staff (Mean = 2.61, N = 234) demonstrate higher
turnover intention than married staff (Mean = 2.42, N = 337). Even if not illustrated in
the figure, marital status difference was tested using an independent samples t-test. Based
upon a mean difference (0.19) of marital status groups, the two groups did significantly
differ (t = 2.470, df = 569, p < 0.05). Consistent with the finding from the zero-order
correlation test, the result evidences that single staff accounting for 41% of the directcare staff population, were more likely than married staff to have higher levels of
turnover intention.
Figure 10. Turnover Intention by Marital Status among Texas Direct Care Staff

Turnover Intention

2.8

2.61

2.6

2.42

2.4

2.2
Single

Married

Marital Status
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

69
The finding from the zero-order correlation test indicates that staff with less
seniority were more likely to express higher levels of turnover intention than those with
more seniority. As Figure 11 illustrates, ranging from the lowest tenure of 0.08 years to
the highest of 32 years, the continuous metric variable of tenure was reduced to seven
groups: the least senior group (0-3 years of service) to the most senior group (19+ years
of service). Graphically, tenure group shows a slight negative relationship with turnover
intention up to the tenure group (10-12 years) in which turnover intention was the lowest
at which point, however, the curve rose in the tenure group (13-15 years) and dropped in
the tenure group (19 years or more). Even though there was some variation in the curve
of the 13-15 years tenure group, the test for linearity indicates the significant linear
relationships (p < 0.001) between seniority and turnover intention. This finding suggest
that turn over intention tends to start rising among the senior staff prior to their early
retirement.
Figure 11. Turnover Intention by Tenure Group among Texas Direct Care Staff
2.8

Turnover Intention

2.6

2.4

2.2

2.0
0-3

4-6

7-9

10 - 12

13 - 15

16 - 18

19+

Tenure Group (years of service)
Note : Responses to turnover intention are made on a 5-point scale. Higher scores indicate higher turnover intention.

70
Not listed in the figure, tenure group differences were tested using one-way
ANOVA. The ANOVA reveal that a significant difference, F (6) =3,397, p < 0.01,
existed when comparing turnover intention among the seven tenure groups. More
specifically, the Tukey’s HSD Post-Hoc test on this significant ANOVA measure was
performed to determine any significant difference among the seven groups. The Tukey’s
HSD test indicates that two significant differences was found between the 0–3 years
tenure group (Mean = 2.92) and the 7-9 years tenure group (Mean = 2.26), and between
the 0–3 years tenure group (Mean = 2.92) and the 10-12 years tenure group (Mean =
2.21). Supportive of the finding from the zero-order correlation test, this finding
indicates that high turnover intention was strongly prevalent among direct-care staff
whose tenure range was between 0 and 3 years. This tenure group accounts for 45.6%
(257 out of 564) of the direct-care staff population.
Summary
From the findings from all bivariate analyses, three important results are worth
mentioning. First, it can be surmised that organizational factors were more important than
individual status factors in association with turnover intention. All organizational
variables were found to be significantly associated with turnover intention, and the
strength of the relationships with turnover ranged from weak to strong. In contrast, only
five individual status variables out of the eleven were found to be significantly associated
with turnover intention, but their associations with turnover intention were found to be
very weak. Second, among all organizational variables, affective commitment, pay, high
sacrifice commitment, overall job satisfaction, job stress, and nature of work appear to
have substantial associations with an officer’s turnover intention.
Finally, among the five individual status variables, gender was found to have the
strongest association with turnover intention, where male direct-care staff members
(41.5% of the population) were more likely than female staff to express turnover
intention. In addition, tenure in current department was found to be the second strongest
factor in association with turnover intention, where high turnover intention was strongly
prevalent in the 0-3 years tenure range (45.6% of the population). However, the similar
pattern cannot be applied to the age group analysis since no significant mean difference
existed between age groups. For direct-care staff, it seems that turnover intention is a
matter of tenure not age.

71
Multivariate Regression Analyses
The bivariate analyses were used to provide the direction and strength of each
individual variable in its association with turnover intention. Bivariate analysis is
valuable in summarizing the relationship between each of the independent variables and
turnover intention at once, but it can not test any significant effect of one independent
variable on turnover intention, after holding all other independent variables constant. In
response, three stepwise Ordinary Least Square (OLS) regression analyses were
employed to assess whether various individual and organizational variables influence a
direct-care staff member’s turnover intention (see Table 20). In each equation of the table,
standardized regression coefficient, generally known as Beta coefficient, varies between 1 (a perfect negative prediction power) to 1 (a perfect positive prediction power). A Beta
coefficient of 0 indicates no power in predicting turnover intention.
The three OLS analyses are necessary to seek for any evidence to support the
findings from the previous zero-order correlation, independent-Samples t-test, or oneway ANOVA analyses and to examine whether or not the findings are still maintained
after statistically controlling for the effects of individual status variables on a staff
member’s turnover intention. In each equation, the turnover intention was the dependent
variable. Accordingly, the first equation examined the effects of the individual status
variables, indicative of individual factors, on a staff member’s turnover intention. On the
other hand, the second equation determined the effects of only organizational variables,
indicative of organizational factors, on a staff’s turnover intention. These two separate
equations were designed to compare which factors, individual or organizational, have
more influence on a direct-care staff member’s turnover intention. Finally, the third
equation, as the most complete equation model, determined whether the organizational
factors are still statistically significant after controlling for the effects of the individual
factors.
Two multicollinearity diagnostics were conducted to determine any violation of
the multivariate regression assumption, that there are no highly correlated independent
variables measuring the same thing (Hair et al., 2006). The simplest way of finding
multicollinearity is to check correlation (Fox, 1981; Hy, Feig & Regoli, 1983). None of
the previous Pearson’s correlations (Table 19) among all independent variables were
higher than ± 0.7. The second method of identifying multicollinearity is to examine all
individual variation inflation factor (VIF) scores for each individual variable: a VIF of
1.0 reflects total independence, and for a VIF higher than 1.0, more collinearity is
reflected in the variable (Hair et al., 2006). According to Stevens (1992), all independent
variables’ coefficients in a multivariate regression model are unbiased and efficient when
all VIF scores do not exceed 10. As seen in Table 20{renumber Table 20?}, none of the
VIF exceeded 2.11. Taken together, neither the results from zero-order correlations nor
the results from the VIF methods indicate substantial evidence that multicollinearity is
not an issue in this analysis, and does not substantially alter any of the findings or
subsequent conclusions drawn from the analysis.

72
Table 20. The Determinants of Turnover Intention among Texas Direct-Care Staff (N = 579)
Equation 1

Included Variables

c

Beta

Equation 2
d

VIF

Beta

c

VIF

Equation 3
d

Betac

d

VIF

Individual Status Variables
Gender (male = 1)

0.168 ***

1.031

0.084 **

Ethnicity (Caucasian = 1)
Martial status

a

-0.109 **

Tenure

-0.123 **

-0.064 *

1.111

0.069 *

1.138

1.053

Education level
b

1.117

1.034

-0.108 ***

1.124

Organizational Variables
Affective commitment

-0.347 ***

1.669

-0.332 ***

1.683

High sacrifice

-0.300 ***

1.144

-0.266 ***

1.191

Overall job satisfaction

-0.132 ***

1.374

-0.128 ***

1.430

Pay

-0.159 ***

1.379

-0.124 ***

1.468

Promotion

-0.097 **

1.462

-0.123 ***

1.494

Operating procedures

-0.096 **

1.464

-0.106 **

1.571

Role conflict

0.099 *

1.921

0.121 **

1.963

Job stress

0.178 ***

1.977

0.223 ***

2.076

Social support

-0.094 *

2.101

-0.086 *

R -square =

0.074

0.531

0.564

F =

4.120

27.350

20.700

Significance =

0.000

0.000

0.000

a

1 = currently married; b in current department;

c

Standardized Coefficients; d Variance Inflation Factor

2.113

* p < 0.05; ** p < 0.01; ** p < 0.001

Equation 1 of Table 20 examines only the impact of individual status variables on
a direct-care staff’s turnover intention, and demonstrates a significant, and good-model fit.
The chi-square test of the model indicates that Equation 1 significantly predicted a staff’s
turnover intention (χ2 = 4.120, df = 3, p < 0.001). Three individual status variables were
found to have statistically significant effects on direct-care staff’s turnover intention. The
remaining eight individual status variables--age, ethnicity, educational level, number of
children at home, tenure in current department, and prior employment in probation, law
enforcement, corrections, and parole—were excluded from the final best-fit equation,
since each lacked a statistically significant, high partial correlation (Hair et al., 2006).

73
The three statistically significant determinants of turnover intention were gender,
marital status, and tenure in current department. Males, and single direct-care staff were
more likely to express higher levels of turnover intention than were females and married
direct-care staff. In addition, tenure in current department was negatively related to
turnover intention: staff with less seniority were more likely to express higher levels of
turnover intention. However, despite the significance of the standardized coefficients of
all three individual status variables, only 7.4% of the variance in the dependent
variable—turnover intention—was accounted for (R-square = 0.074).
Equation 2 examines only the impact of effects of organizational variables on
direct-care staff’s turnover intention, and demonstrates a significant and good-model fit
(χ2 = 27.350, df = 9, p < 0.001) in predicting turnover intention. Out of twenty-three
organizational variables, nine variables based upon each statistically significant, high
partial correlation were included in Equation 2. The remaining fourteen variables were
excluded from the final best-fit equation since they did not meet the entry significance for
turnover intention. The excluded variables include: lack of alternatives, fringe benefits,
contingent rewards, supervision, co-workers, nature of work, communication, role
overload, role ambiguity, dangerousness of the job, distributive justice, procedural justice,
participatory climate, and empowerment.
The nine statistically significant determinants of turnover intention were affective
commitment, high sacrifice commitment, overall job satisfaction, pay, promotion,
operating procedures, role conflict, job stress, and social support. Particularly, job stress
and role conflict were positively related to turnover intention: as direct-care staffs’ levels
of job stress and perceived role conflict increased, their turnover intention increased also.
On the other hand, the other determinants were negatively related to turnover intention.
Specifically, direct-care staff who reported lower levels of affective commitment, high
sacrifice commitment, and overall job satisfaction were more likely to express higher
levels of turnover intention. In addition, those who reported lower levels of satisfaction
with pay, promotion and operating procedures were more inclined to leave. Finally, those
who reported lower levels of perceived social support were more likely to have an
inclination to leave.
Two additional findings relevant to Equation 2 are worth mentioning. First, the
nine independent variables included accounted for 53.1% of the variance in the
dependent variable, turnover intention. This portion of variance, explained by Equation 2
(R-square = 0.531) is almost 7.2 times higher than that explained by Equation 1 (R-square
= 0.074). This finding suggests that organizational factors have a much more substantial
contribution to make in predicting a staff’s turnover intention than the individual factors.
Second, the standardized regression coefficients for promotion, operating procedures,
role conflict, and social support were all significant but lower than ± 0.1. On the other
hand, the standardized regression coefficients for affective commitment, high sacrifice
commitment, overall job satisfaction, pay, and job stress were -0.347, -0.300, -0.132,
-0.159, and 0.178, respectively. All coefficients well exceeded ± 0.1. These five
organizational variables, therefore, appear to have both statistical and substantive
significance in predicting a direct-care staff’s turnover intention. Given the standardized

74
regression coefficients, affective commitment had the strongest statistically significant
effect on turnover intention, followed by high sacrifice commitment, job stress, pay, and
overall job satisfaction.
Equation 3 in Table 20 is the final and most complete best fit regression model.
Here the individual status variables are treated as statistical control variables to mainly
determine whether the significant organizational variables found in Equation 2 are still
statistically significant after controlling for the effects of the individual status variables.
Equation 3 shows a significant and good model fit: The chi-square test of the model
indicates that Equation 3 significantly predicted an officer’s turnover intention
(χ2 = 20.700, df = 13, p < 0.001). The proportion of variance explained by Equation 3
(R-square = 0.564) is higher than that explained by Equation 2 (R-square = 0.531) and is
almost 7.6 times higher than that explained by Equation 1 (R-square = 0.074). This
finding indicates that the organizational variables have a greater contribution to make in
predicting a direct-care staff member’s inclination to leave, even after controlling for the
effects of the individual status variables.
Thirteen variables based upon each statistically significant, high partial
correlation were included in Equation 3: four individual status variables and nine
organizational variables. In comparison with Equation 1, two individual status variables,
gender and tenure in current department were still included as being statistically
significant. Ethnicity and education level were excluded from Equation 1. However, after
being associated with organizational factors in Equation 3, both variables were included
as statically significant predictors of turnover intention. This finding suggests that the
effects of ethnicity and education level on turnover intention become direct through the
mediation of organizational factors. Furthermore, marital status was excluded from the
final best-fit equation after organizational factors were included in Equation 3. This
finding indicates that the effect of marital status on turnover intention is indirect and is
mediated through organizational factors as well.
In addition, Equation 3 is statistically supportive of the direction of the four
significant individual status variables found in Equation 1: male, minority direct-care
staff with less seniority and higher levels of educational background were more likely to
express higher levels of turnover intention than their counterparts. However, despite the
significance of the standardized coefficients of all four significant individual status
variables, only the standardized coefficient for tenure in current department (Beta =
-0.108, p < 0.001) exceeded the cut-off point of ± 0.1. This finding suggests that gender,
ethnicity and education level contribute significantly, but weakly, to predict direct-care
staff’s turnover intention, whereas tenure exerts the strongest direct effect on turnover
intention. Clearly, tenure has a more substantial contribution to make in predicting staff’s
turnover intention than other, individual factors.
As for organizational factors, even after entering individual factors as control
variables into the final regression equation, no organizational variable was excluded from
the final model. This finding indicates that all included variables found in Equation 2 are
still maintained due to their significant contributions to predicting turnover intention

75
beyond the predictive power of the control variables. Also, the direction and strength of
each of the nine significant organizational variables in Equation 3 are consistent with the
findings in Equation 2. All included organizational variables except for role conflict and
job stress were negatively related to turnover intention. Utilizing the cut-off point of ± 0.1,
the standardized regression coefficient for social support was all significant but lower
than ± 0.1. This finding indicates that social support contributes significantly, but weakly,
to predict a staff’s turnover intention.
In contrast, the standardized regression coefficients for affective commitment,
high sacrifice commitment, overall job satisfaction, pay, promotion, operating procedures,
role conflict, and job stress were -0.332, -0.266, -0.128, -0.124, -0.123, -0.106, 0.121, and
0.223, respectively, well exceeding the cut-off point. Like the findings of Equation 1,
affective commitment, high sacrifice commitment, overall job satisfaction, pay, and job
stress substantially contribute to predict turnover intention. Unlike the findings of
Equation 2, promotion, operating procedures and role conflict exceeded the cut-off point
of ± 0.1 and became substantial contributing predictors of turnover intention. Finally,
these findings show affective commitment to be the strongest predictor of turnover
intention, followed by high sacrifice commitment, job stress, overall job satisfaction, pay,
promotion, role conflict, and operating procedures.
Summary
Taken together, consistent with the findings from the bivariate analyses, these
regression analyses reveal that organizational factors, rather than individual status factors,
have a substantially greater contribution to make in predicting direct-care staff’s
inclinations to leave employment. Among the organizational factors, affective
commitment, high sacrifice commitment, overall job satisfaction, pay satisfaction,
promotion satisfaction, operating procedural satisfaction, role conflict, and job stress
have a significant, direct effect on turnover intention, after holding all other independent
variables constant. In addition, tenure in current department, among the four significant,
individual status predictors of turnover intention, has the strongest direct effect after
controlling for all other independent variables. Therefore, the length of tenure in the
staff’s current department makes the most substantial contribution in predicting turnover
intention, moreso than other, individual factors. This finding is consistent with the overall
finding from the bivariate analyses for individual status factors. Recall the findings from
the Tukey’s HSD Post-Hoc test for the seven tenure groups: high turnover intention was
strongly prevalent among direct-care staff whose tenure range was somewhere between
0-3 years. This tenure group accounts for 45.6% (257 out of 564) of the direct-care staff
population at the present time.

76

Section 6.
Structural Equation Modeling for
both Line Community Supervision
Officers and Direct-Care Staff

77
In the previous separate multivariate regression analyses for both community
supervision officers and direct-care staff, organizational variables were found to have a
substantially greater contribution to make in predicting turnover intention than individual
status variables. Focusing solely on the organizational variables for both groups (see
Tables 18 and 20)–pay satisfaction, overall job satisfaction, high sacrifice commitment,
and affective commitment, after controlling for the effects of individual status factors–turned out to be significant predictors of turnover intention. However, as noted by Hair et
al. (2006), the separate multivariate regression analyses used are limited in measuring
only the direct effects of these organizational variables on turnover intention. Therefore,
they cannot provide any results for indirect effect and total effect (direct and indirect), for
each of the significant four organizational predictors of turnover intention. In addition,
the previous regression analyses are limited to the assumption that one variable can be
either an independent or a dependent variable, and in that way cannot provide and test
any hypothetical causal link model between pay satisfaction, overall satisfaction, high
sacrifice commitment, affective commitment and turnover intention.
Both community supervision officers and direct-care staff need to be collapsed
into one population in this section. Of the two main purposes of this report, one exists to
probe the causal relationship of pay satisfaction with four significant attitudinal and
behavioral consequences–overall job satisfaction, high sacrifice commitment, affective
commitment, and turnover intention–in the Texas probation system. Based upon this
investigation, one important concern may be addressed: the role of pay satisfaction in
preventing high voluntary turnover. Using Amos 16.0 for structural equation modeling
techniques, a hypothetical, causal link from pay satisfaction to turnover intention,
established through solid theoretical practices, may be evaluated. Structural equation
modeling (SEM) techniques can compare the indirect, direct, and total effects of pay
satisfaction, overall satisfaction, high sacrifice commitment, and affective commitment
on turnover intention. These analyses are believed to be helpful in providing important
managerial strategies in preventing and curbing turnover-related problems in Texas
probation.
Theoretical and Empirical Ground for a Hypothetical Model
Before specifying theoretical grounds and a hypothetical causal model, it should
be noted that any individual status variables were not included in the causal model. There
are two reasons behind the exclusion. First, there have been a number of studies to
examine the individual characteristic correlates of turnover. Basically, age, gender,
education level, and tenure have been found to correlate with turnover (e.g., Cotton &
Tuttle, 1986, Griffeth et al, 2000, Huselid & Day, 1991).
However, focusing on the effects of these individual status variables, the results
from the previous multivariate regression analyses were considered inconsistent across
the two groups; while age and educational level were found to be significant correlates of
turnover intention among community supervision officers, tenure, gender, and education
level were found to be significant correlates among direct-care staff. Though educational
level and ethnicity were found to have significant effects on turnover intention for both

78
groups, their effects were almost negligible. In other words, these findings seems to be
inconsistent across the two groups and don’t support the previous empirical literature.
Another reason for the exclusion is to make the hypothetical model simple and
thereby provide the simplest of explanations in the hypothetical model of complex
turnover intention processes (Hair et al., 2006). This seems to be supported by the
previous finding that individual status variables, in comparison with organizational
variables, were found to have a substantially weak or negligible contribution in
associating and predicting turnover intention. Therefore, the individual status variables
were not included as control variables when examining the hypothetical, causal link from
pay satisfaction to turnover intention. Additionally, the lack of alternatives, one
component of organizational commitment, was excluded from the hypothetical model
since the findings from the previous analysis indicate that it is not a significant predictor
of turnover.
Due to the lack of literature on pay satisfaction and its organizational outcomes, it
is difficult to identify a causal model of voluntary turnover processes from pay
satisfaction, and to explain causal relationships between a subset of the variables.
Therefore, considerable research based upon the theoretical ground, and empirical
findings, should be required in order to identify causal relationships between pay
satisfaction, overall satisfaction, high sacrifice commitment, affective commitment, and
turnover intention.
Pay Satisfaction and Organizational Justice
To both practitioners and researchers, pay satisfaction has long been a topic of
interest. At the basis of pay satisfaction studies, there are two theoretical grounds: equity
theory (Adams, 1963) and discrepancy theory (Lawler, 1971). According to Adams
(1963), the most highly motivated employee is the one who perceives his or her output,
such as pay and benefits, equal to his or her input, such as effort. If his or her ratio of
input to output is significantly different from a referent other’s ratio, he or she tends to
feel under-rewarded, and judge that he or she is not being treated fairly. This may lead to
a range of attitudinal and behavioral effects, such as higher stress, lower job satisfaction,
lower organizational commitment, and higher turnover intention or actual turnover
(Campbell & Pritchard, 1976; Martin 1981).
When explaining the relationship between pay satisfaction and its outcomes,
Lawler’s discrepancy theory expanded Adams’ equity theory by incorporating the
concept of valence (how much one values the reward). Valence is a product of
expectancy theory (Vroom, 1964). In other words, like equity theory, pay satisfaction is a
matter of matching actual pay level with the pay level one expects he or she should receive in
comparison with those of a referent other. Unlike equity theory, however, only if one
highly values satisfaction with pay level, his or her reaction to a negative discrepancy
between actual and expected receipts would result in negative fairness perceptions, lower job
satisfaction, and lower organizational commitment–possibly causing higher turnover
intention or actual turnover (Campbell & Pritchard, 1976; Cropanzano & Greenberg,
1997; Vandenberghe & Tremblay, 2008).

79
Indeed, both equity and discrepancy theories offer considerable insight into how
an employee determines his or her pay satisfaction, and suggests negative outcomes of
pay dissatisfaction are primarily caused by the discrepancy between what pay level he or
she deserves to receive, and what actual pay level he or she obtains. Empirical research
has strongly established the important theoretical link between pay satisfaction and its
organizational outcomes, overall job satisfaction, affective commitment, high sacrifice
commitment, and turnover intention (Heneman & Judge, 2000). First, in the relationship
between pay satisfaction and overall job satisfaction, Miceli, Jung, Near, & Greenberger
(1991) found a significantly positive relationship: as pay satisfaction increases, overall
job satisfaction increases. It is clear to understand the finding since pay satisfaction is
only one facet of overall job satisfaction.
Second, Heneman & Judge (2000) suggested that pay satisfaction has a positive
influence on both affective commitment and continuance commitment. Empirical
research has consistently supported this contention (Dulebohn and Martocchio, 1998;
Huber, Seybolt, & Veneman, 1992; Vandenberghe & Tremblay, 2008). The relationship
of pay satisfaction and affective commitment was analyzed by Dulebohn and Martocchio
(1998). They found that pay satisfaction has a strongly positive correlation with affective
commitment. This finding indicates that higher pay satisfaction binds one with the
organization and thereby enhances his or her affective commitment to the organization. A
most recent confirmed the relationships (Vandenberghe & Tremblay, 2008). Also, they
found a strong positive relationship between pay satisfaction and high sacrifice
commitment: one’s satisfaction with pay enhances the cost of leaving, leading to higher
sacrifice commitment.
Lastly, in the relationship between pay satisfaction and turnover intention, a
higher level of pay satisfaction was found to lessen a higher level of turnover intention:
as pay satisfaction increases, turnover intention decreases (Dailey & Kirk, 1992;
DeConinck & Stilwell, 2004). Moreover, pay satisfaction (Jung, Near, & Greenberger,
1991; Miceli et al., 1991; Motowidlo, 1983) was found to be a significant predictor of
both turnover intention and actual turnover. These findings support Mobely’s (1977)
hypothesis that turnover intention is significantly predicted by pay satisfaction.
Further evidence has indicated that pay satisfaction not only has a direct, but also
an indirect effect on turnover intention, through overall job satisfaction (Lum, Kervin,
Colark, Reid, & Sirola, 1998), and organizational commitment (Lum et al., 1998;
Vandenberghe & Tremblay, 2008). More specifically, Vandenberghe & Tremblay (2008),
in their study of the effects of pay satisfaction and organizational commitment on
turnover intention, found that both affective and high sacrifice commitments were found
to have intervening effects that account for the association between pay satisfaction and
turnover intention (Vandenberghe & Tremblay, 2008). However, the effect of pay
satisfaction on turnover intention was not found to be mediated by either normative
commitments, or a lack of alternative commitments. These empirical findings indicate
that pay satisfaction has both a direct and indirect effect on turnover intention, through
overall job satisfaction, high sacrifice commitment, and affective commitment.

80
In measuring pay satisfaction, however, there has been debate about how to
conceptualize pay satisfaction. Initially, as criticized by Heneman (1985), Adams’ (1963)
equity theory and Lawler’s (1971) discrepancy theory are essentially based on predicting
pay-level satisfaction and explaining its organizational outcomes. Thus, pay satisfaction
has long been conceived as a uni-dimensional construct, where one’s pay-satisfaction
responds only to pay level. In other words, pay satisfaction cannot be explained by pay
level itself. Accordingly, Heneman & Schwab (1985) hypothesized the multi-dimensional
nature of pay satisfaction, and developed four correlated, but distinct dimensions: pay
level, benefits, pay raises, and pay structure/administration.
The multi-dimensionality of pay satisfaction is informative since each dimension
has determinants and organizational outcomes (Judge, 1993). For example, an individual
may be satisfied with benefits while being dissatisfied with his or her level of pay.
Another example is related to different organizational outcomes caused by different
dimensions of pay, and demonstrates that pay-raise satisfaction, one component of the
multidimensional pay satisfaction construct, was found to be a significant predictor of
turnover intention and actual turnover, while pay-level satisfaction was not found to be
significant (Tekleab, Bartol, & Liu, 2005).
Recently, organizational justice has been incorporated into pay satisfaction. This
trend is based on both equity and discrepancy theories and suggests that pay satisfaction
has a strong theoretical association with organizational justice, and that both pay
satisfaction and perceived organizational justice are determined by the discrepancy
between what compensation he or she deserves, and what actual compensation he or she
obtains in comparison with those of a referent other (Campbell & Pritchard, 1976;
Cropanzano & Greenberg, 1997). In other words, fairness perception is the essential tenet
in understanding pay satisfaction and organizational justice. Organizational justice
conceptually includes two types of justice: distributive justice and procedural justice.
Distributive justice is the degree of fairness in distributing rewards (Price & Mueller,
1986), while procedural justice is the degree of fairness in the procedures used for
distribution (Folger & Greenberg, 1985).
More recently, Heneman and Judge (2000) suggested that pay satisfaction and
organizational justice are related, but distinct constructs: one may be satisfied with pay
level but may not feel fairly treated in the policies and procedures by which pay is
administered, possibly leading to lower pay satisfaction. Other theoretical and empirical
research has consistently supported the distinct constructs between compensation-level
satisfaction (pay and benefits), and satisfaction with the compensation
structure/administration (Judge, 1993; Miceli and Lane, 1991; Williams, Malos and
Palmer, 2002). Specifically, as proposed by Heneman and Judge (2000), some
dimensions of pay satisfaction, such as pay-level satisfaction, and benefit satisfaction, are
related to perceived degrees of fairness in distributing rewards (distributive justice),
whereas pay structure/administration is associated with perceived degrees of fairness in
the procedures used for distribution (procedural justice). However, Williams et al. (2006),
in their meta-analysis, found that the causal links between pay satisfaction and

81
organizational justice are not still clear: yet organizational justice is assumed to influence
pay satisfaction and vice versa.
According to organizational justice theory, employees decide whether they have
been treated fairly after comparing what actual compensation they have received with
those of a referent other. Similar to the relationship of pay satisfaction and its
organizational outcomes, if organizational injustice is perceived, one feels relative
deprivation or a feeling of discontent, which in turn will lead to a range of attitudinal and
behavioral effects, such as higher stress, lower job dissatisfaction, lower organizational
commitment, and higher turnover intention or actual turnover (Campbell & Pritchard,
1976; Hendrix et al., 1981).
Empirical research has supported the important theoretical link between
organizational justice and its organizational outcomes. Specifically, overall job
satisfaction (Dailey & Kirk, 1992; Hendrix et al., 1999; McFarlin & Sweeney, 1992;
Miceli et al., 1991), organizational commitment (Folger & Konovsky, 1989; Hendrix et
al., 1999; Konovsky & Cropanzano, 1991; Martin & Bennett, 1996), and turnover
intention (Acquino et al., 1997; Hendrix et al., 1999) are aspects of motivation that were
found to be influenced by employee judgments regarding the fairness of outcomes and
the fairness of the procedures. Taken together, organizational justice and pay satisfaction
are distinct constructs but conceptually related, and thereby the relationship of
organizational justice and its organizational outcomes is similar to that of pay satisfaction
and its organizational outcomes. These findings suggest that the incorporation of
organizational justice into pay satisfaction provides a better understanding of the nature
and realm of pay satisfaction, and enables the incorporated model to better understand
pay satisfaction’s influence on its organizational outcomes.
Causal Link between Overall Job Satisfaction, Organizational Commitment, and
Turnover Intention
As described earlier, job satisfaction and organizational commitment are based on
an employee’s emotional and psychological state. For this report, job satisfaction was
defined as a linkage between an employee and his or her job, resulting from the appraisal
of his or her job and job experiences (Locke, 1976). On the other hand, organizational
commitment was defined as a linkage between an employee and his or her organization,
referring to the strength of his or her identification with and involvement in his or her
organization (Meyer & Allen, 1997).
As for job satisfaction, there is a substantial body of literature that has reported
that job satisfaction is negatively related to turnover intention, and has a direct effect on
turnover intention (e.g., Griffeth et al., 2000; Hom and Griffeth, 1991; Tett and Meyer,
1993). As for the indirect effect of overall job satisfaction, Tett and Meyer (1993)
reported that the relationship between job satisfaction and turnover intention is not
completely mediated by organizational commitment, reflecting the direct effect of job
satisfaction on turnover intention. However, according to Griffeth et al. (2000), in their
updated meta-analysis of antecedents and correlates of employee turnover, findings from

82
a growing body of recent, empirical research support the notion that organizational
commitment is a better predictor of turnover than job satisfaction, and that organizational
commitment mediates a causal link between job satisfaction and employee turnover.
These findings suggest that job satisfaction has a direct effect on both organizational
commitment and turnover intention, as well as an indirect effect on turnover intention
through organizational commitment.
In a causal link between job satisfaction and organizational commitment, the
dominant theoretical view has assumed that an employee’s emotional state and attitude
toward a specific job necessarily precedes their psychological state and attitude towards
the organization (Mowday et al, 1982; Mueller, Boyer, Price, & Iverson, 1994). This
assumption implies that overall job satisfaction causally precedes organizational
commitment. Some research (e.g., Currivan, 1999; Vandenberg & Lance, 1992) has
found an opposite causal sequence and supported the causal ordering from organizational
commitment to overall job satisfaction. Nonetheless, many empirical studies (e.g.,
Mowday et al, 1982; Mueller et al, 1994; Vandenberg & Scarpello, 1990); Williams &
Hazer, 1986) have analyzed the causal ordering from overall job satisfaction and
organization commitment. Although not always, they have generally confirmed the
causal precedence of job satisfaction over organizational commitment. These findings
indicate that organizational commitment may be a more immediate influence on turnover
intention than job satisfaction.
In a causal ordering from organizational commitment and turnover intention,
Meyer and Allen (1997) have reported that organizational commitment is negatively
related to turnover intention, and is also an antecedent to turnover intention. As
mentioned before, pay satisfaction has been found to be positively related to high
sacrifice and affective commitment while unrelated to a lack of alternative and normative
commitment (Dulebohn and Martocchio, 1998; Huber, Seybolt, & Veneman, 1992;
Vandenberghe & Tremblay, 2008). Focusing on high sacrifice and affective commitment,
McGee and Ford (1987) and Meyer, Allen, and Gellatly (1990) provided a theoretical
explanation suggesting that an employee’s awareness of the costs associated with leaving
the organization leads to a higher desire to continue to work, which in turn, may lead to a
greater degree of emotional attachment to, identification with, and involvement in the
organization. Despite a lack of empirical research to test the causal link, intuitively it
appears to manifest through examination of the causal precedence of high sacrifice
commitment over affective commitment.
Given the accumulated theoretical explanation and empirical findings, Figure 12
presents a hypothetical model to examine the causal relationship of both compensation
satisfaction and organizational justice with overall satisfaction, high sacrifice
commitment, affective commitment, and turnover intention. Note that one curved,
double-headed arrow in the figure indicates correlation between compensation
satisfaction and organizational justice, whereas the other straight, single-headed arrows
represents causal relations between two variables. Extending the previous literature into
the report, the following six specific hypotheses were developed:

83
H1: Compensation satisfaction (pay and benefits satisfaction) and organizational
justice (distributive and procedural justice) are positively correlated;
H2: Each of compensation satisfaction and organizational justice has a direct
effect on overall job satisfaction, high sacrifice commitment, affective
commitment and turnover intention.
H3: Each of compensation satisfaction and organizational justice has an indirect
effect on turnover intention through overall job satisfaction, high sacrifice
commitment, and affective commitment.
H4: Overall job satisfaction has a direct effect on high sacrifice commitment,
affective commitment and turnover intention, and also has an indirect effect on
turnover intention through high sacrifice commitment and affective
commitment.
H5: High sacrifice commitment has a direct effect on affective commitment and
turnover intention, and also has an indirect effect on turnover intention through
affective commitment.
H6: Affective commitment has a direct effect on turnover intention.

Procedural Justice

Distributive Justice

Benefits Satisfaction

Pay Satisfaction

Organizational
Justice

Compensation
Satisfaction

Negative relationship

Positive relationship

Figure 12. Hypothetical Model.

Overall Job
Satisfaction

High Sacrifice
Commitment

Affective
Commitment

Turnover
Intention

84

85
Exploratory Factor Analysis and Confirmatory Factor Analysis
As discussed earlier, pay satisfaction and organizational justice are distinct but
related constructs. For this reason, incorporation of organizational justice into pay
satisfaction may provide a better understanding of the nature and domain of pay
satisfaction. This insight should enable the incorporated model to better predict pay
satisfaction’s influence on its organizational outcomes. As seen in Figure 12, the first
latent construct of compensation satisfaction was hypothesized to combine pay
satisfaction and fringe benefits satisfaction, found by the previous theoretical ground and
empirical findings to be correlated. Note that this report adopted the five items of pay
satisfaction developed by Dunham and Smith (1979), and classified by Williams et al.
(2002) as multi-dimensional pay satisfaction, rather than uni-dimensional, pay-level
satisfaction. The second latent construct of organizational justice was hypothesized to
bind distributive justice and procedural justice, found by the previous theoretical ground
and empirical findings to be correlated. Moreover, two latent constructs were
hypothesized to be correlated, but distinct.
First, the exploratory factor analysis using a Varimax rotation in Table 21 was
undertaken to examine whether all items in pay satisfaction, benefits satisfaction,
distributive justice, and procedural justice can be explained by the two latent constructs—
compensation satisfaction and organizational justice. The 20 items of the four variables
(pay satisfaction, fringe benefits satisfaction, distributive justice, and procedural justice)
were factor analyzed. A Varimax rotation yielded four factors, together accounting for
64.24% of the total variance among the variables with eigenvalues greater than 1.0, and
Kaiser-Meyer-Ohlin Measure of Sampling Adequacy (KMO) as 0.915.
As seen in Figure 21, five items of distributive justice were loaded on the first
factor, scoring 0.75 or more. Five items of pay satisfaction were loaded on the second
factor, scoring 0.68 or more. Six items of procedural justice were loaded on the third
factor, scoring 0.61 or more. In addition, four items of fringe benefits were loaded on the
final factor, scoring 0.59 or more. Similar to the interpretation of correlation coefficients,
these factor loading scores satisfy the 0.50 cut-off point, and suggest substantial loadings
(Comrey & Lee, 1992). Results from the exploratory factor analysis demonstrate that all
items were loaded on their original measure, indicating that the four-factor model would
be better than the hypothesized, two-factor model (compensation satisfaction and
organizational justice).

86
Table 21. Exploratory Factor Analysis (N = 3,216)
Rotated Factor Loadings
Item

Factor 1

Factor 2

Factor 3

Factor 4

1. Pay satisfaction #1

0.22

0.81

0.08

0.15

2. Pay satisfaction #2

0.09

0.77

0.11

0.24

3. Pay satisfaction #3

0.16

0.83

0.07

0.18

4. Pay satisfaction #4

0.26

0.69

0.23

0.13

5. Pay satisfaction #5

0.21

0.68

0.14

0.10

6. Benefits satisfaction #1

0.10

0.16

0.09

0.67

7. Benefits satisfaction #2

0.10

0.15

0.07

0.83

8. Benefits satisfaction #3

0.12

0.15

0.07

0.84

9. Benefits satisfaction #4

0.11

0.15

0.14

0.59

10. Distributive justice #1

0.83

0.21

0.30

0.14

11. Distributive justice #2

0.86

0.21

0.25

0.13

12. Distributive justice #3

0.82

0.21

0.24

0.14

13. Distributive justice #4

0.75

0.22

0.18

0.12

14. Distributive justice #5

0.84

0.21

0.29

0.14

15. Procedural justice #1

0.12

0.10

0.70

0.10

16. Procedural justice #2

0.17

0.13

0.75

0.19

17. Procedural justice #3

0.26

0.11

0.66

0.09

18. Procedural justice #4

0.33

0.11

0.65

0.07

19. Procedural justice #5

0.29

0.10

0.64

0.06

20. Procedural justice #6

0.06

0.07

0.61

0.02

Eigenvalue
7.570
Explanation of Variance
19.61
Kaiser-Meyer-Olkin (KMO) = 0.915

2.281
16.32

1.622
15.85

1.374
12.46

Notes : Responses to each item are made on a 5-point scale; Principal components factor analysis with a
varimax rotation.

However, as noted by Hair et al. (2006), “Exploratory factor analysis can be
conducted without knowing how many factors really exist or which variable belong with
which constructs” (p. 773). For this reason, confirmatory factor analysis is the more
appropriate way to cross-validate the factor structure developed by exploratory factor
analysis (Jöreskog & Sörbom, 1999). Therefore, the result from the exploratory factor
analysis should be tested by confirmatory factor analysis, to examine whether the fourfactor model may be proven empirically. Accordingly, confirmatory factor analysis was
also conducted to confirm whether the four-factor model is better than the hypothesized

87
two-factor model. In additional, any alternative factor models, such as one-factor and
three-factor models, were tested by confirmatory factor analysis.
As demonstrated in Table 22, the fit of the model to the data was evaluated by the
following six indices: χ2 Ratio, GIF, RMSEA, NFL, CFI and TLI (Hair et al., 2006). The
χ2 ratio, one of three absolute-fit indices, is 8.74 (χ2 = 8.74/df = 1), well exceeding a ratio
of 2, usually used as a rough, rule of thumb for good-fit. Therefore, the result of the χ2
ratio test doesn’t seem to support the absolute fit of the hypothesized two-factor model to
the data. However, since the χ2 ratio test is very sensitive to the large sample size, the χ2
ratio test itself should not be considered as a best-test of the model’s absolute fit (Hair et
al., 2006; Hu & Bentler, 1995). In contrast, the other absolute fit (GFI = 0.99, RMEAS =
0.49) indices, well exceeding the recommended cut-off values, indicate that the
hypothetical two-factor model, compared to alternative factor models, provided best fit to
the data.
In addition to the absolute fit indices, the incremental fit indices (NFL, CFI and
TLI) demonstrate that relative to alternative factor models, the two-factor model provided
a significant improvement. For example, the three incremental fit indices were better for
the two-factor model (NFI = 0.99, CFI = 0.99, TLI = 0.98) than for the four-factor model
(NFI = 0.94, CFI = 0.94, TLI = 0.83). The results of confirmatory factor analysis do not
support the four-factor model developed by exploratory factor analysis. Instead they
confirm the hypothesis: there are two distinct constructs—compensation satisfaction and
organizational justice—wherein pay satisfaction and fringe-benefits satisfaction
measured compensation satisfaction, while distributive and procedural justice measured
organizational justice. Therefore, the results from the confirmatory factor analysis
support the good discriminant validity of the two constructs (compensation satisfaction
and organizational justice).
Specifically, the factor-loading scores of both pay satisfaction and fringe-benefits
satisfaction well exceed the 0.50 cut-off, suggesting substantial loadings (Comrey & Lee,
1992). The factor-loading score for pay satisfaction is 0.79, which is considered excellent,
while the factor-loading score of fringe-benefits satisfaction is 0.53, which is considered
good. Similarly, the factor-loading scores of both distributive justice and procedural
justice also suggest substantial loadings. The factor-loading score for distributive justice
is 0.88, which is considered excellent, while the factor-loading score of procedural justice
is 0.64, which is considered very good. These findings indicate that pay satisfaction has a
1.49 times (0.79/0.53) higher association with compensation satisfaction than with fringebenefits satisfaction. Also, distributive justice has a 1.38 times (0.88/0.64) higher
association with organizational justice than with procedural justice. These factor-loading
scores are demonstrated in Figure 13.

88
Table 22. A Description and Standard of Selected Fit Indices

Fit Measure

Absolute Fit
Indices

Incremental Fit
Indices

Index

Cutoff Standard

χ 2 Ratio*

below 2

Goodness of Fit Index (GFI)

above 0.9

Root Mean Square Error Approximation (RMSEA)

below 0.05

Normed Fit Index (NFL)

above 0.9

Comparative Fit Index (CFI)

above 0.9

Tucker Lewis Index (TLI)

above 0.9

* χ 2 Ratio is calculated by dividing the χ 2 value by the degrees of freedom.

Results
Table 23 summarizes the key descriptive statistics for the two constructs
(compensation satisfaction and organizational justice), and the other variables. The
average of individual variables included here were measured using the 1-5 Likert scale,
with a rating of 1 indicating “strongly disagree” and a rating of 5 indicating “strongly
agree.” Only a high level of overall job satisfaction was reported with an average mean of
3.52. In contrast, the average of the other individual variables is considered mixed,
neither agreeing nor disagreeing, and thereby not supporting any one particular view.
Table 23. Means, Standard Deviations, and Zero-order Correlations of the Variables of Interest
(N = 3,216)

Variable

Mean

SD

No. of
items

2.

3.

4.

5.

6.

1. Compensation satisfaction*

2.63

0.69

9

0.51

0.29

0.29

0.37

-0.45

2. Organizational justice*

2.72

0.80

11

1.00

0.30

0.20

0.50

-0.44

3. Overall job satisfaction

3.52

0.82

5

1.00

0.20

0.48

-0.53

4. High sacrifice commitment

3.22

1.05

3

1.00

0.25

-0.47

5. Affective commitment

3.20

0.94

5

1.00

-0.63

6. Turnover intention

2.71

0.96

4

1.00

* Composite scores in this factor are calculated by averaging items representing that measure.
Note : Responses range from 1 to 5. Higher scores indicate favorable responses. All correlation coefficients are
signficant at the 0.01 level.

89
Pearson’s zero-correlation analysis was conducted to assess the strength and
direction of the relationship between variables. The correlation matrix in Table 23
demonstrates that compensation satisfaction, organizational justice, overall job
satisfaction, high sacrifice commitment, and affective commitment were all significant at
p < 0.01, and all variables were negatively correlated with turnover intention. Their
negative correlation coefficients for turnover intention ranged between -0.44 and
-0.63, and exceeded -0.40, approaching the cut-off point of -0.50. Therefore, they have
relatively substantial strengths in association with turnover intention. In addition,
affective commitment was found to be the strongest in association with turnover intention.
These findings were consistent with both existing literature and the results from
the previous zero-order correlation tests for both community supervision officers and
direct-care staff. However, the findings from the Pearson’s zero-correlation analysis
cannot be used to evaluate the hypotheses, since correlation relations ignore the influence
of the other variables (Hair, et al., 2006). Structural equation modeling based upon the
hypothetical model (Figure 12) is allowed to simultaneously assess all the predicted
relations between compensation satisfaction, organizational justice, overall job
satisfaction, high sacrifice commitment, affective commitment, and turnover intention.
Final Model
Table 24 presents the finalized structural model’s statistics. The final model
provided a better fit than the hypothesized model in Figure 12. In the hypothesized model,
the χ2 ratio is 3.39 (χ2 = 16.99/df = 5), well exceeding a ratio of 2 usually used as a rough,
‘rule of thumb’ for good fit. This ratio doesn’t seem to support the absolute fit of the
hypothesized model to the data. However, the other absolute fit indices (GFI = 0.99,
RMEAS = 0.27), well exceeded the recommended cut-off values, indicating that the
hypothetical model provided an adequate fit to the data. In the hypothetical model,
however, organizational justice was not a significant predictor of overall job satisfaction
(p = 0.80), high sacrifice commitment (p = 0.17) and turnover intention (p = 0.48). Hence,
the three paths (organizational justice Æ overall job satisfaction; organizational justice Æ
high sacrifice commitment; and, organizational justice Æ turnover intention) were
eliminated and the original model was reanalyzed into the final version.

90
Table 24. Standardized Structural (path) Coefficients for the Final Model
Dependent variable
Individual variable
Compensation satisfaction
Organizational justice

Overall job
satisfaction

High sacrifice
commitment

0.359 ***

0.318 ***

-

-

0.284 ***

-

0.780 ***

0.354 ***

-0.203 ***

0.076 ***

-0.231 ***

Overall job satisfaction
High sacrifice commitment

Affective
commitment
0.072 *

Affective commitment
R -square =

Turnover
intention
-0.297 ***

-0.348 ***
0.129

0.125

0.342

0.595

*. p < 0.05; **. p < 0.001

Compared to the χ2 ratio of the hypothetical model, the χ2 ratio, in the final model
is 0.49 (χ2 = 4.46/df = 5), well below a ratio of 2 as a rough, ‘rule of thumb’ for good fit.
Along with this χ2 ratio, the other absolute fit indices (GFI = 0.99, RMEAS = 0.01) fully
support the absolute best-fit of the final model to the data. Moreover, the incremental fit
indices (NFL, CFI and TLI) demonstrate that compared to the hypothetical model, the
final model provided a slight improvement. The three incremental fit indices were better
for the final model (NFI = 0.999, CFI = 0.999, TLI = 0.998) than for the hypothetical
model (NFI = 0.996, CFI = 0.996, TLI = 0.991). These results indicate that the
hypothesized model fits the data very well but the final model, after leaving out three
insignificant paths (organizational justice Æ overall job satisfaction, high sacrifice
commitment and turnover intention, respectively), best fits the data. In Table 24, 12.9%,
12.5%, 34.2%, and 59.5% of variance in overall job satisfaction, high sacrifice
commitment, affective commitment, and turnover intention were explained respectively
by the final model. Figure 13 presents the significant paths of the final structural model.

91

92
As hypothesized (H1), the effects of compensation satisfaction and organizational
justice are positively correlated at 0.73. This finding indicates no causal order between
the two constructs. Instead, compensation satisfaction and organizational justice are
distinct, but correlated constructs: as organizational justice increases, compensation
satisfaction increases, and vice versa. As predicted, compensation satisfaction was found
to have its significant direct effect on overall job satisfaction (standardized path
coefficient = 0.36), high sacrifice commitment (standardized path coefficient = 0.32),
affective commitment (standardized path coefficient = 0.08), and turnover intention
(standardized path coefficient = -0.30). However, organizational justice was found to
have its significant direct influence on only affective commitment while having an
insignificant direct impact on overall job satisfaction, high sacrifice commitment, and
turnover intention. This finding suggests that when an employee believes that he or she is
fairly treated by the organization, he or she is more likely to have a greater degree of
emotional attachment to, identification with, and involvement in the department.
However, the perceived fairness cannot directly lead to higher levels of overall job
satisfaction and high sacrifice commitment, and lower levels of turnover intention. Hence,
the hypothesis (H2) is only partially supported.
Table 25 summarizes structural equation modeling estimations of indirect, direct,
and total effects of each independent variable on overall job satisfaction, high sacrifice
commitment, affective commitment, and turnover intention. As hypothesized (H3),
compensation satisfaction had its indirect effect on turnover intention through overall job
satisfaction, high sacrifice commitment and affective commitment. Specifically,
compensation satisfaction was found to have an indirect or mediated influence on highsacrifice commitment through overall job satisfaction (0.03); on affective commitment
through overall job satisfaction and high sacrifice commitment (0.15); and on turnover
intention through overall job satisfaction, high sacrifice commitment, and affective
commitment (-0.23). However, organizational justice was found to have its indirect or
mediated effect on turnover intention only through affective commitment. Therefore, the
hypothesis (H3) is only partially supported.
As predicted, overall job satisfaction had a direct effect on high sacrifice
commitment, affective commitment, and turnover intention. Also, it had an indirect effect
on turnover intention through high sacrifice commitment, and affective commitment.
Likewise, high sacrifice commitment had a direct effect on affective commitment, and
turnover intention; and, had its indirect effect on turnover intention through affective
commitment. The total effect of compensation satisfaction was found to have a much
larger influence than that of overall pay satisfaction. This finding indicates that
compensation satisfaction is a stronger predictor of high sacrifice commitment than
overall satisfaction: an employee’s high satisfaction with compensation causes his or her
strong, perceived awareness of the costs associated with leaving the organization,
eventually leading to a strong desire to continue to work. Moreover, affective
commitment had a direct effect only on turnover intention; but had the strongest direct
effect (-0.34), followed by compensation satisfaction (-0.30), high sacrifice commitment
(-0.23) and overall job satisfaction (-0.20). These findings suggest that the hypotheses
(H4, H5, and H6) are fully supported.

0.03

IE*

* Indirect Effect; ** Direct Effect; *** Total Effect.

Affective commitment

High sacrifice commitment

-

0.36

TE***

Overall job satisfaction

0.36

DE**

-

-

IE*

0.08

-

0.32

DE**

0.08

-

0.35

TE***

High Sacrifice Commitment

Organizational justice

Compensation satisfaction

Independent Variable

Overall Job Satisfaction

-

0.01

-

0.15

IE*

0.08

0.35

0.28

0.07

DE**

0.08

0.36

0.28

0.23

TE***

Affective Commitment

Dependent Variable

Table 25. Indirect, Direct, and Total Effects of the Variables of Interests (N = 3216)

-

-0.03

-0.14

-0.10

-0.23

IE*

-0.35

-0.23

-0.20

-

-0.30

DE**

-0.35

-0.26

-0.34

-0.10

-0.53

TE***

Turnover Intention

93

94
Of particular interest in Table 25 is an indirect, direct, and total effect of
compensation satisfaction, organizational justice, overall job satisfaction, high sacrifice
commitment, and affective commitment on turnover intention. Compensation satisfaction
was found to have the largest total effect (indirect and direct) on turnover intention
(-0.53), more than a half of which (56.6%) is due to a relatively large direct effect (-0.30).
The indirect effect of compensation satisfaction on turnover intention (-0.23) is mostly
(86.9%) through the combined variables of high sacrifice and affective commitment.
Followed by compensation satisfaction, affective commitment had the second largest
total effect (only direct) on turnover intention (-0.35), closely followed by overall job
satisfaction.
The total effect of overall job satisfaction is -0.34, having the relatively larger
direct effect (58.8%) of the total effect. Most of the indirect effect of overall job
satisfaction (92.8%) on turnover intention is through affective effect, reflecting the
substantially strong mediating effect of the relationship between overall job satisfaction
and turnover intention. Lastly, 88.4% of the total effect of high sacrifice commitment on
turnover intention is primarily direct (-0.23), whereas the total effect of organizational
justice (-0.10) had only its weak indirect effect on turnover intention, and is less
important than that of the other variable. These findings indicate that compensation
satisfaction is a pivotal organizational influence on turnover intention, followed by
affective commitment, overall job satisfaction, and high sacrifice commitment.
Two additional findings relevant to Table 25 are worth mentioning. First, in
comparing the direct effects of compensation satisfaction, organizational justice, overall
job satisfaction, and high sacrifice commitment on affective commitment, overall job
satisfaction was found to have the largest direct effect (0.35), followed by organizational
justice (0.28). Also, compensation satisfaction (0.07) and high sacrifice commitment
(0.08), were found to have negligible direct effects on affective commitment. Despite the
negligible direct effect of compensation satisfaction, the total effect of compensation
satisfaction (0.23), due to its relatively large indirect effect (0.15), appears to be
important, following affective satisfaction (0.36), and organizational justice (0.28). These
findings suggest that overall job satisfaction is a key influence on affective commitment,
followed by organizational justice, and compensation satisfaction.
Secondly, in comparing the total effects of compensation satisfaction,
organizational justice, and overall job satisfaction on high sacrifice commitment, we find
that compensation satisfaction (0.35) is a key influence on high sacrifice commitment,
followed by overall job satisfaction (0.08). No total effect of organizational justice was
found, suggesting that organizational justice did not have any direct or indirect effects on
high sacrifice commitment. Finally, comparing the total effects of compensation
satisfaction and organizational justice on overall job satisfaction, compensation
satisfaction did have a substantial total effect (0.36) on overall job satisfaction, but
organizational justice did not have a total effect on overall job satisfaction at all.

95
Summary
Taken together, the structural equation modeling analysis supports four out of the
six hypotheses. The two unsupported hypotheses (partially H2 and H3) are related to
organizational justice and its organizational outcomes. Inconsistent with previous
literature, organizational justice was found to have only its direct influence on affective
commitment, not overall job satisfaction, high sacrifice commitment, and even turnover
intention. Also, organizational justice was found to have only its indirect effect on
turnover intention. These findings seem to indicate the lack of the substantial impact of
organizational justice in the final model.
Most importantly, while affective commitment had the strongest direct effect on
turnover intention, the total effect (indirect and direct) of compensation satisfaction on
turnover intention was found to be substantially greater than the total effect of affective
commitment. These findings suggest that the total influence of compensation satisfaction
(pay and fringe-benefits satisfaction) is much more important than that of affective
commitment in reducing high levels of turnover intention in Texas probation.

96

Section 7.
Conclusion & General Policy
Implications

97
A review of the literature suggests that present probation systems fail to resolve
high levels of employee turnover rates, leading to additional direct and indirect costs.
Direct costs include expenditures necessary for recruitment and training. Indirect costs,
although more difficult to measure, include low morale among the remaining staff,
contributing to lower standards of job-related service and productivity. Clearly, direct and
indirect costs stemming from high turnover rates create unnecessary burdens for an
organization, and contribute to a poor working environment. Above all, these negative
consequences could lead to a failure in promoting public safety, the ultimate mission of
the Texas probation system. Therefore, reducing high levels of staff turnover should be a
top priority for Texas probation administrators, faced with tightening administration
budgets and expanding public expectations.
Until now, no detailed data has been collected and reported, and no readily
available, cost-effective mechanism has been implemented to fully and empirically
analyze actual, voluntary turnover rates of Texas probation. Unlike retirement and
termination, voluntary turnover can be preventable by identifying underlying reasons, and
addressing identified causes of voluntary turnover. In response, this report
comprehensively examined (1) the determinant factors (both personal and organizational)
that shape turnover intention among 3,234 line probation officers and direct care staff;
and, (2) pay satisfaction’s influence on organizational outcomes, such as overall job
satisfaction, organizational commitment, and turnover intention.
The data collected from line probation officers and direct-care staff reveals high
levels of inclinations to leave. For example, 41.3 percent reported their turnover
intention: 30.3 percent were having serious thought about leaving in the near future and
another 11 percent were actively looking to leave. Among all organizational variables,
pay and promotion were found to be the most negatively perceived work-related areas in
Texas probation. Another important finding is that the average mean of organizational
commitment was found to be lower than that of overall job satisfaction, suggesting the
employees in Texas probation have a stronger psychological and emotional attachment to
their job and job experience, than to their department.
Results from bivariate and multivariate regression analyses consistently indicate
that organizational factors, rather than individual status factors, have a much greater
association with turnover intention, and made a much greater contribution to predicting
employees’ inclinations to leave Texas probation. This suggests that rather than an
employee, the department has influence of the underlying causes for turnover intention.
Specifically, affective commitment, high sacrifice commitment, overall job satisfaction,
and pay satisfaction were found to substantially contribute to predicting turnover
intention among the line probation officers. Similarly, affective commitment, high
sacrifice commitment, overall job satisfaction, pay, promotion, job stress, role conflict
and operating procedures were found to be main predictors of inclinations to leave among
the direct-care staff. There are four common predictors in both the line probation officer
and direct-care staff groups: affective commitment, high sacrifice commitment, overall
job satisfaction and pay. In both groups, affective commitment, among these four

98
common factors, was found to be the strongest predictor of turnover intention, suggesting
affective commitment is the most immediate precursor of turnover intention.
Multivariate regression analyses are limited in providing only direct effect of an
independent variable on turnover intention. Therefore, structural equation modeling
analysis was employed to compare total effects (direct and indirect) of compensation
satisfaction (pay and fringe benefits), overall job satisfaction, lack of alternatives, high
sacrifice and affective commitment on turnover intention. Results from the structural
equation modeling indicate that the total effect (indirect and direct) of compensation
satisfaction on turnover intention was substantially greater than the total effect of
affective commitment. While affective commitment had the strongest direct effect on
turnover intention, the total influence of compensation satisfaction, especially pay
satisfaction, is more important than affective commitment in reducing high levels of
turnover intention and subsequent voluntary turnover. Taken together, it can be
concluded that pay satisfaction is the strongest underlying cause of high turnover
intention in Texas probation.
Based on the main findings, general recommendations to policy-makers are
provided. Most importantly, Texas probation administrators should be acutely made
aware of the transition from individual to organization factors, especially the significance
of pay satisfaction and affective commitment, as underlying causes leading to high
voluntary turnover rate. As for pay satisfaction, only small portions of the line probation
officers and direct-care staff sampled were satisfied. For example, only 10.3 percent felt
their pay level was good; only 13.5 percent indicated their pay level was either adequate
or more than adequate given the cost of living in their area; and only 15.4 percent
reported that their pay level had a favorable influence on their overall attitude toward
their job. These statistics indicate high levels of pay dissatisfaction. Therefore, probation
administrators should recognize chronic, negative organizational outcomes caused by
inadequate salary, and should be united front to increase more compensation for the
employees in Texas probation. Also, a concerted effort should be made to convince the
Texas Legislature to significantly increase probation funding. Inherent traps in the
vicious cycle of low pay satisfaction, high turnover intention and high voluntary turnover,
may lead to the possible diminished promotion of public safety, comprising the definitive
mission of the Texas probation system.
Increasing compensation is important, but on its own does not necessarily
guarantee an employee’s long-term commitment, especially affective commitment, to the
mission of Texas probation. Affective commitment is the most immediate precursor of
turnover intention; employees with strong affective commitment to the organization are
more valuable employees. However, 3,234 respondents reported the main reason that
they are committed to their department is an awareness of the costs associated with
leaving—such as their personal accumulated investments and limited employment
opportunities—rather than their strong emotional attachment to, identification with, and
involvement in their department.

99
In recognizing existing low levels of affective commitment, probation
administrators should identify its underlying reasons, and develop strategies which
increase employee’s emotional attachment to, identification with, and involvement in
their department. An employee who doesn’t have an emotional connection to the
organization’s mission may start thinking about leaving. Therefore, every department
should have a clearly articulated mission, vision and values that are supported and
reinforced by management. The strategies should be embodied as integral processes in
the strategic plans of the ever evolving Texas probation.
Younger personnel and those with fewer years of service are more likely to feel
inclined to leave their job with Texas probation. Among all individual variables, age was
found to make the most significant contribution to the line probation officers’ turnover
intention; while length of tenure made the most substantial contribution to the direct-care
staff’s turnover intention. Among the nine age groups, high turnover intention was most
prevalent among line probation officers whose age ranged from 20 to 34 years.
Surprisingly, this age range group accounts for 42.8 percent of the line probation officers
sampled. Likewise, high turnover intention was most prevalent among direct-care staff
whose tenure range was somewhere between 0-3 years. This tenure group accounts for
45.6 percent of the direct-care staff sampled.
Given the highest turnover intention among the younger age and tenure groups, it
is highly recommended that probation administrators recognize the unique characteristics
of the new generation of employee and should devote considerable attention and
resources to this new generation, who have a much lower affective commitment and
much higher turnover intention than other groups. Inevitably, the role of probation
managers is extremely important in providing organizational stimulus for this new
generation of employees to encourage their feelings of belonging and to establish their
emotional attachment to, identification with, and involvement in their department.
Specifically, there needs to be a concerned focus by management on developing
mentoring relationships with new employees. Also, shift in supervisory and managerial
roles and styles should be made from directing and controlling the new generation in a
traditional, autocratic organizational climate to one of facilitating, coaching, and
consulting with them. To fulfill these important managerial roles, Texas probation
departments should devote considerable attention and resources to the selection,
development, and training of managers.
Lastly, in the not too distant past, probation administrators did not experience the
need to actively recruit staff. It was not uncommon to have a number of highly qualified
applicants for each available position. This is no longer the case, and probation
departments find themselves in competition with other social service and law
enforcement agencies for prospective employees from a dwindling labor pool. Probation
administrators should become less passive and more active in the recruitment of new
employees by attending job fairs at colleges and universities, developing close
relationships with faculty members of criminal justice programs, and mentoring senior
level students in area high schools with the hope of having them return to the community
after college and seeking employment as a probation officer.

100
References
Adams, J. S. (1963). Toward an understanding of inequity. Journal of Abnormal and
Social Psychology, 67, 422-436.
Acquino, K., Griffeth, R. W., Allen, D. G., & Hom, P.W. (1997). Integrating justice
constructs into the turnover process: a test of a referent cognitions model.
Academy of Management Journal, 40, 1208-1227.
Allen, N. J., & Meyer, J. P. (1990). The measurement and antecedents of affective,
continuance, and normative commitment to the organization. Journal of
Occupational and Organizational Psychology, 63, 1-18.
Begley, T. M., & Czajka, J. M. (1993). Panel analysis of the moderating effects of
commitment on job satisfaction, intent to quit, and health following organizational
change. Journal of Applied Psychology, 78, 552-556.
Brayfield, A., & Roth, H. F. (1951). An index of job satisfaction. Journal of Applied
Psychology, 35, 307-311.
Byrd, T. G., Cochran, J. K., Silverman, I. J., & Blount, W. R. (2000). Behind bars: An
assessment of the effects of job satisfaction, job-related stress, and anxiety on jail
employees’ inclination to quit. Journal of Crime and Justice, 23(2), 69-93.
Campbell, J. P., & Pritchard, R. D. (1976). Motivation theory in industries and
organizational psychology. In M. D. Dunnette (Ed.), Handbook of Industrial and
Organizational Psychology (pp. 63-130). Chicago: Rand-McNally.
Cherniss, C. (1980). Staff Burnout: Job stress in the Human Services. Beverly Hills, CA:
Sage Publications.
Chesney, M., & Rosenman, R. (1980). Type A behavior in the work setting. In C. L.
Cooper & R. Payne (Eds.), Current concerns in occupational stress (pp. 187-212).
New York: John Wiley & Sons.
Cotton, J. L., & Tuttle, J. M. (1986). Employee turnover: A meta-analysis and review and
implications for future research. Academy of Management Review, 11, 55-70.
Tuttle, 1986
Comrey, L. A., & Lee, H. B. (1992). A first course in factor analysis (2nd ed.). Hillsdale,
NJ: Lawrence Erlbaum.
Crank, J. P., Regoli, R., Hewitt, J. D., & Culbertson, R. G. (1995). Institutional and
organizational antecedents of role stress, work alienation, and anomie among
police executives. Criminal Justice and Behavior, 22, 152-171.

101
Cranny, C. J., Smith, C. P., & Stone, E. F. (1992). Job satisfaction: How people feel
about their jobs and how it affects their performance. New York: Lexington
Books.
Crozier, M. (1964). The bureaucratic phenomenon. Chicago: University of Chicago Press.
Cropanzano, R., & Greenberg, J. (1997). Progress in organizational justice: Tunneling
through the maze. In C. L. Cooper & I. T. Roberston (Eds.), International Review
of Industrial and Organizational Psychology (pp. 317-372). New York: John
Wiley & Sons.
Cullen, F. T., Link, B. G., Cullen, J. B., & Wolfe, N. T. (1989). How satisfying is prison
work? A comparative occupational approach. Journal of Offender Counseling,
Services & Rehabilitation, 14, 89-108.
Cullen, F. T., Link, B. G., Wolfe, N. T., & Frank, J. (1985). The social dimensions of
correctional officer stress. Justice Quarterly, 2(4), 505-533.
Dailey, R. C., & Kirk, D. J. (1992). Distributive and procedural justice as antecedents of
job dissatisfaction and intent to turnover. Human Relations, 45, 305-317.
Davis, J. A. (1971). Elementary survey analysis, Englewood Cliffs, NJ: Prentice-Hall
Methods of Social Science.
DeConinck, J. B., & Stilwell, C. D. (2004). Incorporating organizational justice, role
states, pay satisfaction and supervisor satisfaction in a model of turnover
intentions. Journal of Business Research, 57, 225-231.
Dulebohn, J. H., and Martocchio J. J. (1998). Employee perceptions of the fairness of
work group incentive pay plans. Journal of Management, 24, 469-488.
Dunham, R. B., & Smith, F. J. (1979). Organizational surveys: An internal assessment of
organizational health. Glenview, IL: Scott Foresman & Company.
Eby, L. T., Freeman, D. M., Rush, M. C., & Lance, C. E. (1999). Motivational bases of
affective organizational commitment: A partial test of an integrative theoretical
model. Journal of Occupational and Organizational Psychology, 72, 463-483.
Etzion, D., Eden, D., & Lapidot, Y. (1998). Relief from job stressors and burnout:
Reserve service as a respite. Journal of Applied Psychology, 83, 577-585.
Finn, P. (1999). Correctional officer stress: A cause for concern and additional help.
Federal Probation, 62, 65-74.

102
Folger, R., & Greenberg, J. (1985). Procedural justice: An interpretive analysis of
personnel systems. Research in Personnel and Human Resources Management, 3,
141-183).
Folger, R., & Konovsky, M.A. (1989). Effects of procedural and distributive justice on
reactions to pay raise decisions. Academy of Management Journal, 32, 115-130.
Fox, J. A. (1981). Methods in quantitative criminology. New York: Academic Press.
Gardner, R. (1981). Guard stress. Corrections Magazine, 7, 6-14.
Gerstein, L. M., Topp, C. G., & Correll, G. (1987). The role of the environment and
person when predicting burnout among correctional personnel. Criminal Justice
and Behavior, 14, 352-369.
Gist, M., & Mitchell, T. N. (1992). Self-efficacy: A theoretical analysis of its
determinants and malleability. Academy of Management Review, 17, 183-211.
Gore, A. (1995). The Gore report on reinventing government. New York: Random House.
Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A meta-analysis of antecedents and
correlates of employee turnover: Update, moderator tests, and research
implications for the next millennium. Journal of Management, 26, 463-488.
Grossi, E. L, & Berg, B. L. (1991). Stress and job satisfaction among correctional
officers: An unexpected finding. International Journal of Offender Therapy and
Comparative Criminology, 35(1), 73-91.
Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. L., & Tatham, R. L. (2006).
Multivariate data analysis (6th ed.). Upper Saddle River, NJ: Prentice-Hall Inc.
Hammer, T. H., Landau, J., & Stern, R. N. (1981). Absenteeism when workers have a
novice: The case of employee ownership. Journal of Applied Psychology, 66,
561-573.
Hendrix, W. H., Robbins, T., Miller, J., & Summers, T. P. (1998). Effects of procedural
and distributive justice on factors predictive of turnover. Journal of Social
Behavior & Personality, 13, 611-633.
Hamilton, L. C. (1990). Modern data analysis: A first course in applied statistics. Pacific
Grove, CA: Brooks/Cole.
Heneman, H. G. III (1985). Pay satisfaction. In M. Rowland & R. Ferris (Eds.), Research
in Personnel and Human Resources Management (pp. 115-140). Greenwich, CT:
JAI Press.

103
Heneman, H. G. III, & Judge, T. A. (2000). Compensation attitudes. In S. L. Rynes & B.
Gerhart (Eds.), Compensation in Organization: Current Research and Practice
(pp. 61-203). San Francisco: Jossey-Bass.
Heneman, H. G. III & Schwab, D. P. (1985). Pay satisfaction: Its multidimensional nature
and measurement. International Journal of Psychology, 20, 129-142.
Hom, P. W., & Griffeth, R. W. (1995). Employee turnover. Cincinnati, OH: South
Western.
Hom, P. W., & Griffeth, R. W. (1991). Structural equations modeling test of a turnover
theory: cross-sectional and longitudinal analyses. Journal of Applied Psychology,
76, 350-366.
Huber, V. L., Seybolt, P. M., & Veneman, K. (1992). The relationship between individual
inputs, perceptions, and multidimensional pay satisfaction. Journal of Applied
Social Psychology, 22, 1356-1373.
Huselid, M., & Day, N. (1991). Organizational commitment, job involvement, and
turnover: A substantive and methodological analysis. Journal of Applied
Psychology, 76, 380-391.
Hy, R., Feig, D., & Regoli, R. (1983). Research method and statistics: A primer for
criminal justice & related sciences, Cincinnati, OH: Anderson Publishing.
Jackson, S. E. (1983). Participation in decision-making as a strategy for reducing jobrelated strain. Journal of Applied Psychology, 68, 3-19.
Jaros, S. J., Jermier, J. M., Koehler, J. W., & Sincich, T. (1993). Effects of continuance,
affective and moral commitment on the withdrawal process: An evaluation of
eight structural equation models. Journal of Applied Psychology, 36, 951-995.
Jöreskog, K. G., & Sörbom, D. (1999). LISREL 8: User’s Reference Guide. Lincolnwood,
IL: Scientific Software International. .
Judge, T. A. (1993). Validity of the dimensions of the pay satisfaction questionnaire:
Evidence of differential prediction. Personnel Psychology, 46, 331-355.
Jurik, N., & Halemba, G. (1984). Gender, working conditions and the job satisfaction of
women in a non-traditional occupation: female correctional officers in men’s
prisons. Sociological Quarterly, 25, 551-566.
Ko, J., Price, J. L., & Mueller, C. W. (1997). Assessment of Meyer and Allen’s threecomponent model of organizational commitment in South Korea. Journal of
Applied Psychology, 82, 961-973.

104
Koberg, C. S., Boss, R. W., Senjem, J. C., & Goodman, E. A. (1999). Antecedents and
outcomes of empowerment: Empirical evidence from the health care industry.
Group and Organizational Management, 24, 71-91.
Kohn, M. (1976). Occupational structure and alienation. American Journal of Sociology,
82, 111-130.
Konovsky, M. A., & Cropanzano, R. (1991). Perceived fairness of employee drug testing
as a predictor of employee attitudes and job performance. Journal of Applied
Psychology, 76, 698-707.
Lambert, E. G., Hogan, N. L., & Griffin, M. L. (2007). The impact of distributive and
procedural justice on correctional staff job stress, job satisfaction, and
organizational commitment. Journal of Criminal Justice, 35, 644-656.
Lambert, E. G., Hogan, N. L., Paoline, E. A. III, & Clarke, A. (2005). The impact of role
stressors on job stress, job satisfaction, and organizational commitment among
private prison staff. Security Journal, 18, 33-50.
Lawler, E. E. III (1971). Pay and organizational effectiveness. New York: McGraw-Hill.
Lee, W., & Beto, D. R. (2008, Spring). Factors associated with adult probation officer
turnover in Texas. Executive Exchange, 3-11.
Leonardi, T., & Frew, D. (1991). Applying job characteristics theory to adult probation.
Criminal Justice Policy Review, 5, 17-28.
Lindquist, C., & Whitehead, J. T. (1986a). Burnout, job stress, and job satisfaction among
southern correctional officers: Perceptions and causal factors. Journal of Offender
Counseling Services and Rehabilitation, 10(4), 5-26.
Locke, E. (1976). The nature and causes of job satisfaction. In M. Dunnette (Ed.),
Handbook of Industrial and Organizational Psychology (pp. 1297-1349).
Chicago: Rand McNally.
Lum, L., Kervin, J., Colark, C., Reid, F., & Sirola, W. (1998). Explaining nursing
turnover intent: Job satisfaction, pay satisfaction or organizational commitment?
Journal of Organizational Behavior, 19, 305-320.
Maslach, C. (1982). Burnout: The cost of caring. Englewood Cliffs, NJ: Prentice-Hall.
Maslach, C., & Jackson, S. E. (1981). The measurement of experienced burnout. Journal
of Occupational Behavior, 2, 99-113.

105
Martin, J. (1981). Relative deprivation: A theory of distributive justice for an era of
shrinking resources. In L. L. Cummings & B. M. Staw (Eds.). Research in
Organizational Behavior (vol. 3: pp. 53-108). Greenwich, CT: JAI Press.
Martin, C. L., & Bennett, N. (1996). The role of justice judgments in explaining the
relationships between job satisfaction and organizational commitment. Group and
Organizational Management, 21, 84-104.
Mathieu, J. E., & Zajac, D. M. (1990), A review and meta-analysis of the antecedents,
correlates and consequences of organizational commitment, Psychological
Bulletin, 2, 171-194.
Maxfield, M., & Babbie, E. (2005). Research methods for criminal justice and
criminology (4th ed.). Belmont, CA: Thomson/Wadsworth Publishing.
McGee, G. W., & Ford, R. C. (1987). Two or more dimensions of organizational
commitment: Reexamination of the affective and continuance commitment scales.
Journal of Applied Psychology, 72, 638-641.
McKean, J., & Byers, B. (2000). Data analysis for criminal justice and criminology:
Practice and applications, Needham Heights, MA: Allyn and Bacon, Inc.
McFarlin, D. B., & Sweeney, P. D., (1992). Distributive and procedural justice as
predictors of satisfaction with personal and organizational outcomes. Academy of
Management Journal, 35, 626-637.
Mertler, C. A., & Vannatta, R. A. (2005). Advanced and multivariate statistical methods:
Practical application and interpretation (3rd ed.). Glendale, CA: Pyrczak
Publishing.
Meyer, J. P., & Allen, N. J. (1997). Commitment in the workplace: theory, research, and
application. Thousand Oaks, CA: Sage Publications.
Meyer, J. P., Allen, N. J., & Gellatly, I. R. (1990). Affective and continuance
commitment to the organization: Evaluation of measures and analysis of
concurrent and time-lagged relations. Journal of Applied Psychology, 75, 710-720.
Meyer, J. P., & Herscovitch, L. (2001). Commitment in the workplace: Toward a general
model. Human Resource Management Review, 11, 299-326.
Meyer, J. P., Stanley, D. J., Herscovitch, L., & Topolnytsky, L. (2002). Affective,
continuance, and normative commitment to the organization: A meta-analysis of
antecedents, correlates, and consequences. Journal of Vocational Behavior, 61,
20-52.

106
Miceli, M. P., Jung, I., Near, J. P., & Greenberger, D. B. (1991). Predictor and outcomes
of reactions to pay-for-performance plans. Journal of Applied Psychology, 76,
508-521.
Miceli, M. P., & Lane, M. C. (1991). Antecedents of pay satisfaction: A review and
extension. In K. Rowland & G. R. Ferris (Eds.), Research in Personnel and
Human Resources Management (pp. 235-309). Greenwich, CT: JAI Press.
Mitchell, O., MacKenzie, D. L., Styve, G. .J., & Gover, A. R. (2000). The impact of
individual, organizational, and environmental attributes on voluntary turnover
among juvenile correctional staff members. Justice Quarterly, 17, 333-357.
Mobely, W. H. (1977). Intermediate linkages in the relationships between job satisfaction
and employee turnover. Journal of Applied Psychology, 62, 237-240.
Motowidlo, S. J. (1983). Predicting sales turnover from pay satisfaction and expectation.
Journal of Applied Psychology, 68, 484-489.
Mouzelis, N. (1968). Organization and bureaucracy. Chicago: Aldine.
Mowday, R. T., Porter, L. W., & Steers, R. M. (1982). Employee-organization linkage:
The psychology of commitment, absenteeism, and turnover. New York: Academic
Press.
Moynihan, D. P., & Landuyt, N. (2008). Explaining Turnover Intention in State
Government. Review of Public Personnel Administration, 28, 120-143.
Mueller, C. W., Boyer, E. M., Price, J. L., & Iverson, R. D. (1994). Employee attachment
and noncoercive conditions of work: the case of dental hygienists. Work and
Occupations, 21, 179-212.
National Institute of Corrections. (1994). Recruitment and retention in community
corrections: Report from National Institute of Corrections conference. Federal
Probation, 58, 16-19.
Nunnally, J. (1978). Psychometric theory (2nd ed.). New York: McGraw Hill.
Peterson, M. F., Smith, P. B., Akande, A., Ayestaran, S., Bochner, S., Callan, V., Cho, N.
G., Jesuino, J. C., D'Amorim, M., Francois, P., Hofmann, K., Koopman, P. L.,
Leung, K., Lim, T. K., Mortazavi, S., Munene, J., Radford, M., Ropo, A., Savage,
G., Setiadi, B., Sinha, T. N., Sorenson, R., & Viedge, C. (1995). Role conflict,
ambiguity, and overload: A 21-nation study. Academy of Management Journal,
38, 429-452.
Poole, E., & Regoli, R. (1980). Work relations and cynicism among prison guards.
Criminology, 7, 303-314.

107
Powell, D. M., & Meyer, J. P. (2004). Side-bet theory and the three-component model of
organizational commitment. Journal of Vocational Behavior, 65, 157-177.
Price, J. L., & Mueller, C. W. (1986). Handbook of Organizational Measurement.
Marshfield, MA: Pitman.
Robbins, S. P., Chatterjee, P., & Canda, E. R. (1998). Contemporary Human Behavior
Theory. Boston: Allyn & Bacon.
Selye, H. (1956). The stress of life. New York: McGraw-Hill.
Sheeley, B. (2008). What do the officers say? Understanding job safety and satisfaction
from the officers’ standpoint. Perspectives, 32, 47-52.
Shore, L. & Martin, H. (1989). Performance and turnover intentions, job satisfaction, and
organizational commitment in relations to work. Human Relations, 42: 325-638.
Simmons, C., Cochran, J. K., & Blount, W. R. (1997). The effects of job-related stress
and job satisfaction on probation officers’ inclinations to quit. American Journal
of Criminal Justice, 21, 213-229.
Slate, R. N., & Vogel, R. E. (1997). Participative management and correctional
personnel: A study of the perceived atmosphere for participation in correctional
decision making and its impact on employee stress and thoughts about quitting.
Journal of Criminal Justice, 25, 397-408.
Slate, R. N., Vogel, R. E., & Johnson, W. (2001). To quit or not to quit: Perceptions of
participation in correctional decision making and the impact of organizational
stress. Corrections Management Quarterly, 5, 68-78.
Spector, P. E. (1997). Job satisfaction: Application, assessment, causes, and
consequences. Thousand Oaks, CA: Sage.
Spreitzer, G. M. (1995). Psychological empowerment in the workplace: Dimensions,
measurement, and validation. Academy of Management Journal, 38, 1442-1465.
Spreitzer, G. M. (1996). Social structural characteristics of psychological empowerment.
Academy of Management Journal, 39, 483-504.
Spreitzer, G. M., & Mishra, A. K. (2002). To stay or to go: Voluntary survivor turnover
following an organizational downsizing. Journal of Organizational Behavior, 23,
707-729.
State Auditor’s Office. (2007). An annual report on classified employee turnover for
fiscal year 2007. Austin, TX: State Auditor’s Office.

108
Steel, R. P., & Ovalle, N. K. (1984). A review and meta-analysis of research on the
relationship between behavioral intentions and employee turnover. Journal of
Applied Psychology, 69, 673-686.
Stevens, J. (1992). Applied multivariate statistics for the social science (2nd ed.).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Tabachnick, B. G., & Fiddell, L. S. (1996). Using multivariate statistics (3rd ed.). New
York: Harper Collins.
Tekleab, A. G., Bartol, K. M., & Liu, W. (2005). Is it pay levels or pay raises that matter
to fairness and turnover? Journal of Organizational behavior, 26, 899-921.
Tett, R. P., & Meyer, J. P. (1993). Job satisfaction, organizational commitment, turnover
intention, and turnover: path analyses based on meta-analytic findings. Personnel
Psychology, 46, 259–293.
Texas Juvenile Probation Commission (2000). A Study of Salaries and Turnover Rates of
Juvenile Probation Department Personnel in Texas. Retrieved March 1, 2008,
from the Texas Juvenile Probation Commission Website:
http://www.tjpc.state.tx.us/publications/Reports/RPTOTH200001.pdf
Thomas, R. L. (1987). Stress perception among select federal probation and pretrial
services officers and their supervisors. Federal Probation, 52(3), 45-58.
Trevor, C., Gerhart, B, & Boudreau, J. W. (1997). Voluntary turnover and job
performance: Curvilinearity and the moderating influences of salary growth and
promotion. Journal of Applied Psychology, 82, 44-61.
Vandenberg, R. J., & Lance, C. E. (1992). Examining the causal order of job Satisfaction
and organizational Commitment, Journal of Management, 18, 153-167.
Vandenberghe, C., & Tremblay, M. (2008). The role of pay satisfaction and
organizational commitment in turnover intentions: A two-sample study. Journal
of Business and Psychology, 22, 275-286.
Vandenberg, R. J., & Scarpello, V. (1990). The matching model: An examination of the
process underlying realistic job previews. Journal of Applied Psychology, 75, 6067.
Veneziano, C. (1984). Occupational stress and the line correctional officer. Southern
Journal of Criminal Justice, 8, 214-231.
Vernon, M., & Byrd, J. (1996). Leadership in the 21st century: New roles for federal
probation and pretrial service chiefs. Federal Probation, 60, 21-29.

109
Wagner, J. A. (1994). Participation’s effect on performance and satisfaction: A
reconsideration of research evidence. Academy of Management Review, 19, 312330.
Waters, J. E. (1999). Impact of work resources on job stress among correctional treatment
and staff. Journal of Addictions and Offender Counseling, 20, 26-34.
Whitehead, J. T. (1985). Job burnout in probation and parole: Its extent and intervention
implications. Criminal Justice and Behavior, 12, 91-110.
Whitehead, J. T. (1987). Probation officer job burnout: A test of two theories. Journal of
Criminal Justice, 15, 1-16.
Whitehead, J. T., & Lindquist, C. (1996). Correctional officer job burnout: A path model.
Journal of Research in Crime and Delinquency, 23(1), 23-42.
Williams, L. J., & Hazer, J. T. (1986). Antecedents and consequence of satisfaction and
commitment in turnover models: A reanalysis using latent variable structural
equation models. Journal of Applied Psychology, 71, 219-231.
Williams, M. L., Malos, S. B., & Palmer, D. K. (2002). Benefit system and benefit level
satisfaction: An expanded model of antecedents and consequences. Journal of
Management, 28, 195-215.
Williams, M. L., McDaniel, M. A., & Nguyen, N. T. (2006). A meta-analysis of the
antecedents and consequences of pay level satisfaction. Journal of Applied
Psychology, 91, 392-413.
Wu, V., & Short, P. M. (1996). The relationship of empowerment to teacher job
commitment and job satisfaction. Journal of Instructional Psychology, 23, 85-89.

110
Appendix. Comparison between Total and Usable Responses
(A special thank to all survey participants in the following 120 departments)

Total
Dpt #

CSCD

CSO

Usable Data
DCS

CSO

DCS

Total
Dpt #

CSCD

CSO

Usable Data
DCS

CSO

DCS

1.

Anderson

6

0

6

0

61.

Jim Wells

4

0

4

0

2.

Andrews

3

0

3

0

62.

Johnson

39

4

39

4

3.

Angelina

18

4

18

4

63.

Jones

3

0

3

0

4.

Atascosa

17

11

17

11

64.

Kaufman

8

2

8

2

5.

Bailey

3

0

3

0

65.

Kendall

2

0

2

0

6.

Bastrop

17

0

17

0

66.

Kerr

11

2

11

2

7.

Baylor

1

0

1

0

67.

Kleberg

2

0

2

0

8.

Bell

14

3

14

3

68.

Lamar

10

0

10

0

9.

Bexar

251

43

250

43

69.

Lamb

1

1

1

1

10.

Bowie

14

18

14

18

70.

Lavaca

11

0

11

0

11.

Brazoria

33

0

33

0

71.

Liberty

9

2

9

2

12.

Brazos

31

0

31

0

72.

Limestone

7

0

7

0

13.

Brown

3

0

3

0

73.

Lubbock

61

33

60

33

14.

Burnet

16

16

16

16

74.

Matagorda

4

0

4

0

15.

Caldwell

46

8

46

8

75.

Maverick

1

0

1

0

16.

Cameron

34

3

34

3

76.

McCulloch

2

0

2

0

17.

Cass

11

13

11

13

77.

McLennan

40

6

40

6

18.

Cherokee

4

1

4

1

78.

Midland

11

8

11

8

19.

Childress

4

0

4

0

79.

Milam

2

0

2

0

20.

Collin

43

2

43

2

80.

Montague

3

1

3

1

21.

Comanche

3

0

3

0

81.

Montgomery

41

22

41

22

22.

Cooke

4

2

4

2

82.

Moore

5

1

5

1

23.

Coryell

5

0

5

0

83.

Morris

7

0

7

0

24.

Crockett

25.

Dallas

2

0

2

0

84.

Nacogdoches

7

0

7

0

334

73

334

73

85.

Navarro

1

0

1

0

26.

Dawson

7

0

7

0

86.

Nolan

2

1

2

1

27.

Deaf Smith

6

2

6

2

87.

Nueces

35

31

35

31

28.

Denton

24

2

24

2

88.

Orange

6

2

6

2

29.

Eastland

4

1

4

1

89.

Palo Pinto

2

1

2

1

30.

Ector

7

0

7

0

90.

Panola

3

1

3

1

31.

EI Paso

74

13

74

13

91.

Parker

14

6

14

6

32.

Ellis

6

2

6

2

92.

Pecos

7

0

7

0

33.

Erath

4

0

4

0

93.

Polk

9

0

9

0

34.

Falls

1

2

1

2

94.

Potter

38

0

38

0

35.

Fannin

4

0

4

0

95.

Reeves

2

1

2

1

36.

Fayette

5

2

5

2

96.

Rockwall

2

0

2

0

37.

Floyd

2

0

2

0

97.

Rusk

3

2

3

2
9

38.

Fort Bend

28

3

28

3

98.

San Patricio

18

9

18

39.

Galveston

12

0

12

0

99.

Scurry

2

0

2

0

40.

Gray

3

0

3

0

100.

Smith

12

0

12

0

41.

Grayson

1

0

1

0

101.

Starr

8

0

8

0

42.

Gregg

10

0

10

0

102.

Tarrant

185

1

183

1

43.

Guadalupe

7

0

7

0

103.

Taylor

26

14

26

14

44.

Hale

4

0

4

0

104.

Terry

2

0

2

0

Table continued…

111
Appendix continued
Total
Dpt #

CSCD

CSO

Usable Data
DCS

CSO

Total

DCS

Dpt #

CSCD

CSO

Usable Data
DCS

CSO

DCS

45.

Hardin

10

0

10

0

105.

Tom Green

43

49

43

48

46.

Harris

367

98

366

98

106.

Travis

115

10

115

10

47.

Harrison

7

0

7

0

107.

Tyler

3

1

3

1

48.

Haskell

1

0

1

0

108.

Upshur

10

3

10

3

49.

Henderson

50.

Hidalgo

8

0

8

0

109.

Uvalde

12

3

12

3

121

20

121

20

110.

Val Verde

1

0

1

51.

0

Hill

6

0

6

0

111.

Van Zandt

4

2

4

52.

2

Hockley

3

1

3

1

112.

Victoria

31

0

31

0

53.

Hood

2

0

2

0

113.

Walker

11

3

11

3

54.

Hopkins

6

0

5

0

114.

Webb

11

0

11

0

55.

Howard

5

0

5

0

115.

Wheeler

1

0

1

0

56.

Hunt

0

2

0

2

116.

Wichita

13

0

13

0

57.

Hutchinson

4

0

4

0

117.

Wilbarger

2

0

2

0

58.

Jack

0

1

0

1

118.

Williamson

35

3

35

3

59.

Jasper

2

0

2

0

119.

0

1

0

1

60.

Jefferson

30

10

30

10

1

0

1

0

582

2653

Sum
Total Sum

2659
3241

Wood
120. Young

581
3234

Note : Crane and Winkler CSCDS, respectively, were excluded from the data collection for the survey since each department had only one
personnel with both line officer and managerial duties.

112

Supplemental Statistical Information
on Salary and Tenure

113
Line Probation Officers
Figure 1 depicts the frequency distribution of 2,653 line probation officer salaries.
Figure 1. Line Probation Officer Salary on December 31, 2007 (Including 105 Part-time
Officers)
40.0%
35.0%
30.0%
25.0%
20.0%
15.0%
10.0%
5.0%
0.0%

Percent
Number

Less than
$20,000

$20,000$24,999

$25,000$29,999

$30,000$34,999

$35,000$39,999

$40,000$44,999

$45,000$49,999

$50,000 or
higher

1.3%
35

3.8%
102

9.6%
254

34.2%
907

19.8%
524

17.5%
464

10.3%
273

3.5%
94

¾ A total of 391 line probation officers (14.7%) earned less than $30,000 annually.
¾ A total of 1298 line probation officers (48.9%) earned less than $35,000 annually.
¾ Of the 2,653 line probation officer sampled, the mean salary was $35,175 on
December 31, 2007.

114
Figure 2 shows the percentage of line probation officers in various categories based on
years of experience with their current department. A lack of tenured line officers exists
across the state.
Figure 2. Percentage of Line Probation Officers in Each Tenure Category Based on Years
of Experience with Current Department

42.3%
45%
40%
35%
30%
25%
20%

12.9%

15%

11.3%

10.6%
8.0%

7.6%

7.3%

10%
5%
0%
0-3

4-6

7-9

10-12

13-15

16-18

19+

Years

¾ A total of 1,298 line probation officers (48.9%) had 6 years of less of experience.
¾ Only 880 line probation officers (33.5%) had 10 years or more of experience.

115
Figure 3 reflects the mean salary of line probation officers, based on years of experience
with their department.
Figure 3. Mean Salaries of Line Probation Officers Based on Years of Experience with
Current Department

$45,000
$41,330

$41,335

$39,600
$40,000

$38,050
$35,300

$35,000

$30,000

$32,800

$29,200

$25,000
0-3

4-6

7-9

10-12

13-15

16-18

19+ Years

¾ Line probation officers with 0-3 years experience earned a mean salary of
$29,200.
¾ The mean salary of line probation officers with 4-6 years of experience was only
$32,800.
¾ Line probation officers with 7-9 years experience earned only $35,300.

116
Figure 4 compares the mean line probation officer salary for small, medium and large
departmental size categories. Note that the size category is based on the number of direct,
indirect, and pretrial offenders under supervision: Large Size CSCD (N = 10) - over 9,500
direct, indirect, and pretrial offenders under supervision; Medium Size CSCD (N = 23) less than 9,500 but more than 3,500 direct, indirect, and pretrial offenders under
supervision; and Small Size CSCD (N = 89) - less than 3,500 direct, indirect, and pretrial
offenders under supervision.
Figure 4. Line Probation Officer Mean Salary by Small, Medium and Large Size CSCDs

$34,750
$35,000

$34,000
$32,800

$32,750

$33,000

$32,000

$31,000
Small (N=449)

Medium (N=660)

Large (N = 1544)

¾ There was almost no difference between the mean salary of a small size
department ($32,800) and the mean salary of a medium size department ($32,750).
¾ Compared to the mean salaries of a large department ($34,750), the mean salary
of a small size department was $1,950 less and the mean salary of a medium size
department was $2,000 less.

117
Direct-Care Staff
Figure 5 depicts the frequency distribution of 581 direct-care staff salaries.
Figure 5. Direct-Care Staff Salary on December 31, 2007 (Including 37 Part-time Staff)

35%
30%
25%
20%
15%
10%
5%
0%

Percent
Number

Less than
$20,000

$20,000$24,999

$25,000$29,999

$30,000$34,999

$35,000$39,999

$40,000$44,999

$45,000$49,999

$50,000 or
higher

12.4%
72

29.1%
169

17.9%
104

20.3%
118

8.3%
48

7.4%
43

2.4%
14

2.2%
13

¾ A total of 241 direct-care staff (41.5%) earned less than $25,000 annually.
¾ A total of 345 direct-care staff (59.4%) earned less than $30,000 annually.
¾ Of the 581 direct-care staff sampled, the mean salary was $27,200 on December
31, 2007.

118
Figure 6 shows the percentage of direct-care staff in various categories based on years of
experience with their current department. A lack of tenured line officers exists across the
state.
Figure 6. Percentage of Direct-Care Staff in Each Tenure Category Based on Years of
Experience with Current Department

45.4%
50%
45%
40%
35%
30%
25%
20%

13.1%

15%

10.8%
7.2%

9.5%

10%

6.4%

7.6%

5%
0%
0-3

4-6

7-9

10-12

13-15

16-18

19+

¾ A total of 331 direct-care staff (58.5%) had 6 years of less of experience.
¾ Only 174 direct-care staff (30.1%) had 10 years or more of experience.

119
Figure 7 reflects the mean salary of direct-care staff, based on years of experience with
their department.
Figure 7. Mean Salaries of Direct-Care Staff Based on Years of Experience with Current
Department

$36,000

$33,950

$34,000
$31,250

$32,000
$30,000
$28,000

$28,100

$28,150

$28,350

7-9

10-12

13-15

$26,300

$26,000
$24,000

$23,300

$22,000
$20,000
0-3

4-6

16-18

19+ Years

¾ Direct-care staff with 0-3 years experience earned a mean salary of $23,300.
¾ The mean salary of Direct-care staff with 4-6 years of experience was only
$26,300.
¾ Direct-care staff with less than 16 years experience earned less than $30,000.

120
Figure 8 compares the mean direct-care staff salary for small, medium and large
departmental size categories. Note that the size category is based on the number of direct,
indirect, and pretrial offenders under supervision: Large Size CSCD (N =10) - over 9,500
direct, indirect, and pretrial offenders under supervision; Medium Size CSCD (N =23) less than 9,500 but more than 3,500 direct, indirect, and pretrial offenders under
supervision; and Small Size CSCD (N = 89) - less than 3,500 direct, indirect, and pretrial
offenders under supervision.
Figure 8. Direct-Care Staff Mean Salary by Small, Medium and Large Size CSCDs

$27,500

$28,000

$27,000
$25,600
$26,000

$25,150

$25,000

$24,000

$23,000
Small (N=112)

Medium (N=204)

Large (N=265)

¾ Compared to the mean salaries of a small department ($25,600), the mean salary
of a medium size department was $450 less.
¾ Compared to the mean salaries of a large department ($27,500), the mean salary
of a small size department was $1,900 less and the mean salary of a medium size
department was $2,350 less.