NIH Public Access
Author Manuscript
Criminology. Author manuscript; available in PMC 2015 August 01.

NIH-PA Author Manuscript

Published in final edited form as:
Criminology. 2014 August ; 52(3): 371–398. doi:10.1111/1745-9125.12040.

EXPLAINING THE ASSOCIATION BETWEEN INCARCERATION
AND DIVORCE*
Sonja E. Siennick1, Eric A. Stewart1, and Jeremy Staff2
1College

of Criminology and Criminal Justice, Florida State University

2Department

of Criminology and Sociology, The Pennsylvania State University

Abstract

NIH-PA Author Manuscript

Recent studies have suggested that incarceration dramatically increases the odds of divorce, but
we know little about the mechanisms that explain the association. This study uses prospective
longitudinal data from a subset of married young adults in the National Longitudinal Study of
Adolescent Health (N = 1,919) to examine whether incarceration is associated with divorce
indirectly via low marital love, economic strain, relationship violence, and extramarital sex. The
findings confirmed that incarcerations occurring during, but not before, a marriage were
associated with an increased hazard of divorce. Incarcerations occurring during marriage also were
associated with less marital love, more relationship violence, more economic strain, and greater
odds of extramarital sex. Above-average levels of economic strain were visible among
respondents observed preincarceration, but only respondents observed postincarceration showed
less marital love, more relationship violence, and higher odds of extramarital sex than did
respondents who were not incarcerated during marriage. These relationship problems explained
approximately 40 percent of the association between incarceration and marital dissolution. These
findings are consistent with theoretical predictions that a spouse’s incarceration alters the rewards
and costs of the marriage and the relative attractiveness of alternative partners.

Keywords

NIH-PA Author Manuscript

incarceration; divorce; relationship quality; marital problems
Spouses can be important sources of support for former inmates. They provide emotional
and material support after incarceration, they serve as bridges to extended kinship networks,
and they may play a role in preventing or reducing recidivism (Laub and Sampson, 2003;
Sampson and Laub, 1993; Uggen, Manza, and Behrens, 2004). Yet former inmates’
marriages seem to be quite fragile. Specifically, recent studies have suggested that
incarceration dramatically increases the odds of divorce (Apel et al., 2010; Lopoo and

*Additional supporting information can be found in the listing for this article in the Wiley OnlineLibrary at http://
onlinelibrary.wiley.com/doi/10.1111/crim.2011.52.issue-3/issuetoc.
Direct correspondence to Sonja E. Siennick, College of Criminology and Criminal Justice, Florida State University, 145 Convocation
Way, Tallahassee, FL 32306, Phone: 850-645-9265, (ssiennick@fsu.edu)..
Co-Author Information
Eric A. Stewart: estewart2@fsu.edu
Jeremy Staff: jus25@psu.edu

Siennick et al.

Page 2

NIH-PA Author Manuscript

Western, 2005; Massoglia, Remster, and King, 2011). If we wish to help inmates preserve
these potentially beneficial partnerships, to counteract the corrosive effects of imprisonment
on family well-being (Giordano, 2010; Wildeman, 2010; Wildeman, Schnittker, and Turney,
2012), or to improve our understanding of this apparent collateral consequence of
incarceration, we must first determine why incarceration and divorce are associated.
Although the studies on this topic are few in number, their findings are very consistent.
First, only incarcerations occurring during, versus before, marriage lead to divorce (Lopoo
and Western, 2005; Massoglia, Remster, and King, 2011). Second, inmates’ marriages
continue to be at risk of dissolving even after the spell of incarceration (Apel et al., 2010;
Lopoo and Western, 2005; Massoglia, Remster, and King, 2011; Western, 2006). Third, the
effect is large, with studies reporting up to a .20 increase in the probability of, or a doubling
or more of the odds of, divorce among the formerly incarcerated (Apel et al., 2010;
Massoglia, Remster, and King, 2011). Fourth, the effect increases with incarceration length,
with each additional year behind bars increasing the odds of divorce by 32 percent
(Massoglia, Remster, and King, 2011).

NIH-PA Author Manuscript

The mechanisms behind the incarceration-divorce association have remained relatively
unstudied. One of the only studies to address mechanisms directly focused on which aspect
of the incarceration, the stigma of being an excon or the period of separation from one’s
spouse, is most relevant for divorce (Massoglia, Remster, and King, 2011). Theoretically,
the stigma of a prison record or—as Massoglia and colleagues (2011) found was more likely
—the imposed period of separation would harm specific qualities of marriages, which in
turn would raise the odds of divorce. With respect to these marital qualities, scholars have
speculated that incarceration may change the way that spouses interact and reduce spouses’
ability to support the shared household (Apel et al., 2010; Massoglia, Remster, and King,
2011; Wildeman, Schnittker, and Turney, 2012). Research to date has not yet tested whether
such qualities do mediate the incarceration-divorce association, probably because few data
sources include detailed information on both incarceration and marital quality. Such data
limitations also mean that research has not tested whether inmates had more than their share
of marital problems even before their incarcerations. If they did, then part of the
incarceration-divorce association could be spurious to preexisting relationship risk factors
for divorce.

NIH-PA Author Manuscript

This study contributes to this literature in two ways. First, drawing on theories of marital
instability, we present a conceptualization of incarceration as an experience that alters the
key factors behind marital cohesion: the rewards and costs of the marriage, the barriers to
leaving the marriage, and the relative appeal of alternatives to being in the marriage. We
describe how the conceptualization of these factors in the marriage literature overlaps with
themes emerging from recent studies of incarceration’s impact. Second, we present findings
from a partial test of this perspective. To do this, we capitalize on detailed relationship
information from a subset of married young adults interviewed as part of a larger national
panel study. The data allow us to 1) examine the association between incarceration and the
duration-dependent risk of divorce; 2) test whether incarceration is associated with lower
emotional and economic rewards of the marriage, higher physical costs of being in the
marriage, and increased odds that spouses will turn to alternative romantic partners; and 3)

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 3

NIH-PA Author Manuscript

examine whether these rewards, costs, and alternatives mediate the incarceration-divorce
association. The data also allow us to compare these marital qualities among couples who
have already experienced an incarceration and couples who soon will experience an
incarceration. This comparison sheds light on whether incarceration precedes marital
problems, which would be consistent with mediation, or whether marital problems are likely
to emerge before incarceration, which would be consistent with spuriousness.

A THEORY OF MARITAL DISSOLUTION

NIH-PA Author Manuscript

A prominent perspective on marital dissolution focuses on how the relative attractiveness of
the relationship and the extent of moral and structural commitments to the marriage combine
to influence marital stability. Levinger’s (1965) classic statement of this perspective drew on
social exchange theory (Thibaut and Kelley, 1959) and proposed that divorce is a function of
inducements to remain in the marriage, specifically the attractions of the marriage and
barriers to leaving, and inducements to leave the marriage, specifically the attractiveness of
the alternatives to staying. The attractions of the marriage are conceptualized as the ratio of
the relationship’s rewards to its costs. Rewards are positive aspects such as love, happiness,
respect, trust, sex, companionship, and socioeconomic resources; costs are negative aspects
such as relationship violence (Amato and Hohmann-Marriott, 2007; Previti and Amato,
2003). Barriers to leaving the marriage are personal and social commitments such as
children, religious beliefs, pressure from relatives, and community stigma (Previti and
Amato, 2003; White and Booth, 1991). The relative attractiveness of alternatives to the
marriage is the relative appeal of alternative sources of rewards such as affection, sex, and
socioeconomic resources (Udry, 1981; White and Booth, 1991). The joint influence of these
factors means that unhappy marriages are not always ended (e.g., if the barriers to leaving
are high) and that some happy marriages end (e.g., if the alternatives are more attractive;
Lewis and Spanier, 1979).

NIH-PA Author Manuscript

Elements of this social exchange perspective underlie theories of various aspects of marital
functioning (Collett, 2010), and research has suggested that attractions, barriers, and
alternatives do influence marital dissolution. Married couples report that love, friendship,
communication, commitment, respect, compatibility, trust, and children—indicators of
attractions and barriers—keep them together (Previti and Amato, 2003). Unhappily married
couples who hold pro-marriage values and have few alternative sources of support—both
barriers to exit—report that it is unlikely they will divorce (Heaton and Albrecht, 1991).
Divorced people say that factors such as extramarital affairs, incompatibility, and a lack of
closeness or communication—indicating low attractions and viable alternatives—led to the
divorce (Amato and Previti, 2003). Longitudinally, indicators of low attractions and favored
alternatives, such as low or declining levels of closeness and relationship satisfaction,
household economic hardship, high levels of relationship violence, and involvement in
infidelity, predict later divorce (Amato and Rogers, 1997; Conger et al., 1990; DeMaris,
2013; Kurdek 2002; Rodrigues, Hall, and Fincham, 2006; White and Rogers, 2000).
Although this perspective is “one of the best explanations of marital stability” (White, 2013:
28), an important criticism is that it does not address the sources of changes in marriages,
that is, what might cause shifts in the rewards, costs, and attractiveness of alternatives to a

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 4

NIH-PA Author Manuscript

relationship (Karney and Bradbury, 1995). We suggest that spouses can have life events and
experiences that, by altering these factors, cause their stable marriages to become unstable.
Incarceration may be one such experience that indirectly leads to divorce via its harmful
effects on inducements to remain in the marriage and on inducements to leave it.

APPLYING THE THEORY TO COUPLES WHO HAVE EXPERIENCED
INCARCERATION

NIH-PA Author Manuscript

Past works have described in detail the personal, financial, and social difficulties that
individuals and their spouses experience during a spell of incarceration. Incarceration
physically separates couples, curtailing their closeness and intimacy (Comfort, 2008;
Harman, Smith, and Egan, 2007; Karakurt et al., 2013). The removal of a partner and earner
from the household increases parenting-related stress and family economic instability
(Wildeman, Schnittker, and Turney, 2012). In addition, both inmates and their spouses
worry that their partners have or will cultivate relationships with other romantic interests
(Comfort, 2008; Fishman, 1990). Despite these stresses, inmates report optimism about their
postrelease relationships (Benson et al., 2011). Inmates and their partners often anticipate
that their postrelease relationships will be similar to or better than their preincarceration
relationships (Fishman, 1990; Travis and Waul, 2003).
The elevated rate of postincarceration divorce indicates that these hopes are not always
realized. For instance, Massoglia and colleagues (2011) found that more than 40 percent of
divorces among ever-incarcerated males from a general population sample occurred
postrelease. Apel and colleagues (2010) found that the impact of incarceration on divorce
among Dutch men actually grew stronger over the 10 years after release. The timing of these
divorces suggests that they could be the culminating events in longer term processes of
marital erosion. The timing also is consistent with scholars’ suggestion that any
“honeymoon” period of optimism after incarceration is likely to fade over time as families’
high expectations go unfulfilled (Fishman, 1990; Nurse, 2002).

NIH-PA Author Manuscript

Research on families and reentry has indicated that problems consistent with the elements of
social exchange theory may continue to affect couples postrelease. First, postincarceration
marriages may carry relatively low rewards and high costs. One spouse has been “marked”
as dishonest and unreliable, which damages both spouses’ reputations and interferes with the
former inmate’s ability to obtain and maintain employment and contribute financially
(Arditti and Parkman, 2011; Braman, 2004; Western, 2002). Consistent with this, declines in
mothers’ trust in previously incarcerated fathers partly explain why those fathers are less
involved with their children (Turney and Wildeman, 2013). At the same time that
incarceration undermines partners’ respect and trust, it also creates the threat that partners
will report parole violations if releasees do not contribute enough, stay out too late, or
violate other domestic expectations (Goffman, 2009; Nurse, 2002). As control over money,
household resources, and even freedom shifts away from the previously incarcerated spouse,
power differentials are created or exacerbated (Braman, 2004; Harman, Smith, and Egan,
2007; Oliver and Hairston, 2008). These differentials may amplify the effects of already
poor listening, communication, and conflict resolution skills to increase relationship
violence (Harman, Smith, and Egan, 2007; White et al., 2002). It thus is possible that the
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 5

NIH-PA Author Manuscript

incarceration-divorce association operates in part through lower levels of love, respect, trust,
and relationship satisfaction and through higher levels of shame, embarrassment, family
economic hardship, and relationship violence.
Second, a spouse’s “ex-con” status may lower or remove barriers to divorce, in particular,
external pressures that would have been imposed by relatives, friends, and the wider
community. Under average circumstances, the stigma and social disapproval of divorce
might keep some unhappy marriages intact (Levinger, 1965). The incarceration of a spouse
formally identifies that spouse as a criminal offender, creating its own stigma and social
disapproval (Braman, 2004). Communities attach that stigma and disapproval not only to
inmates but also to inmates’ relatives and associates (Arditti, 2012; Comfort, 2008). These
social penalties narrow social support networks, sometimes for years after the incarceration
(Braman, 2004; Turanovic, Rodriguez, and Pratt, 2012; Turney, Schnittker, and Wildeman,
2012). Partners’ families also may withdraw their support for the relationship in light of the
criminal justice system involvement (Nurse, 2002). If kin, friends, and community members
judge affiliation with a former inmate more harshly than they do divorce, or even encourage
divorce from criminal spouses, then couples experiencing incarceration might be socially
freed to divorce.

NIH-PA Author Manuscript

Third, postincarceration couples may see higher relative appeal in alternatives to remaining
in the marriage, such as alternative partners. For these couples, extramarital partners may
offer rewards and fulfill needs that their marriages do not. For the former inmate, additional
partners provide additional sources of emotional and material support and opportunities to
make up for the lack of sexual contact while incarcerated (Braman, 2004; Harman, Smith,
and Egan, 2007; Thomas et al., 2008). For the nonincarcerated spouse, they provide support
and serve as “backups” in case their marriages fail (Braman, 2004; Harman, Smith, and
Egan, 2007; Thomas et al., 2008). The incarceration of a spouse thus could create conditions
of low reward, high cost, low barriers to exit, or good alternatives to existing marriages—
any one of which could prompt divorce under exchange theories of marital instability.

NIH-PA Author Manuscript

Proposing that marital problems and eventual divorce are collateral consequences of
incarceration presupposes that the elevated marital problems follow the incarceration. Yet it
also is possible that couples who go on to experience incarceration have troubled marriages
to begin with. Many divorces are foreshadowed by relationship problems visible when
couples are still newlyweds (Lavner, Bradbury, and Karney, 2012). On its surface,
incarceration and its aftermath may provide a plausible explanation for divorce, but this
could obscure the fact that preexisting weak bonds, financial problems, and violence might
have increased these couples’ odds of divorce apart from the incarceration (Goffman, 2009;
Nurse, 2002). Most studies of this topic have been unable to examine the sequencing of
incarceration and marital problems, but this sequencing is important for our understanding
of divorce among former inmates.

CURRENT STUDY
This study examines the association of incarceration with marital problems and divorce over
a 6-year period among a sample of married young adults interviewed as part of a larger

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 6

NIH-PA Author Manuscript

national panel study. These respondents were observed during the peak age span of early
spells of incarceration (Bonczar, 2003). Although our data cover an important age range for
criminal justice system involvement, they cover relatively early marriages. Early marriages
like our respondents’ are not rare—one fifth of young people marry by their early 20s
(Uecker and Stokes, 2008)—but they tend to be less stable than later marriages. Our
findings may not be generalizable to couples who married at older ages. We return to this
issue in the discussion.

NIH-PA Author Manuscript

Our analyses examine whether incarceration 1) is associated with an increased risk of
marital dissolution among married young adults, 2) is associated with marital problems
among these young adults, and 3) is indirectly associated with marital dissolution via its
associations with marital problems. The marital problems we examine are indicators of
rewards of, costs of, and alternatives to respondents’ marriages, specifically couples’ levels
of love, their ability to make ends meet, their levels of relationship violence, and their
involvement in extramarital sex. As noted, these factors are associated with incarceration
and are known predictors of divorce from the marriage literature. Our data do not include
information on external barriers to divorce (e.g., social disapproval), but the removal of such
barriers also could explain part of the incarceration-divorce association. We leave it to future
studies to determine whether they do.

NIH-PA Author Manuscript

Examinations of this topic must account for the possibility that young adults who are
incarcerated may be poor relationship partners to begin with, or that low-quality marriages
led to incarceration. We address this by subdividing respondents who were incarcerated
during marriage into two groups: one that completed interviews about their marital
characteristics before they were incarcerated, and one that completed interviews about their
marital characteristics after they were incarcerated. If incarceration leads to divorce because
it triggers marital problems, then marital problems should be worse among the latter group
and should explain the divorces of only that group. As an added safeguard, we include
covariates that were used in past work, namely, race and ethnicity, parenthood, educational
attainment, employment status, religiosity, substance use, involvement in nonfamily
violence, incarceration prior to marriage, and the length of the marriage (Lopoo and
Western, 2005; Massoglia, Remster, and King, 2011). We also include other known
correlates of marital problems and dissolution, namely, age at marriage, gender, and whether
the couple lived together before marrying (Amato, 2010), and two other potentially relevant
factors, problem gambling and prior domestic violence convictions.
Couples who divorce often report that their marriage had had more than one problem
(Amato and Rogers, 1997). Similarly, couples affected by imprisonment often have multiple
interdependent relationship problems (Harman, Smith, and Egan, 2007). It thus is likely that
the four mediators we examine will best explain the incarceration-divorce association when
considered together. Still, it is important to determine whether any one marital problem in
particular is an especially powerful explanation of the association because such information
would be useful to families, clinicians, and interventionists. We have no a priori
expectations about the relative importance of the four factors, so we examine the extent to
which they jointly and individually mediate any observed effect of incarceration on marital
dissolution.

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 7

METHOD
NIH-PA Author Manuscript

DATA

NIH-PA Author Manuscript

The data are from the National Longitudinal Study of Adolescent Health (Add Health). Add
Health is a good source of data for this project because it features detailed longitudinal
information on respondents’ romantic partnerships and contacts with the criminal justice
system. Add Health drew on a nationally representative sample of adolescents who were in
grades 7-12 during the 1994-1995 school year. Participants were selected via a two-stage
stratified sampling design. First, 132 schools were randomly selected from a national
sampling frame stratified by region, urbanicity, school size, school type, and racial
composition. Then, students in each school were stratified by grade and gender, and a
nationally representative probability sample of nearly 19,000 adolescents was selected for
the longitudinal in-home component of the study. To date, in-home respondents have
completed four in-person survey interviews. The key measures for this study come from the
wave 3 interviews (conducted in 2001-2002) when the now-adult (18-28 years of age)
respondents were first asked detailed questions about their marriages, and from the wave 4
interviews conducted six years later (2007-2008) when respondents (now 24-34 years of
age) provided information about their incarceration histories and the status of their
marriages. We also include some demographic information from wave 1 (1994-1995).
At wave 3, respondents provided details about their romantic relationship histories. To
create the analytical sample, from the total pool of 15,197 wave 3 respondents, we selected
those who were married at the time of the wave 3 interview (n = 2,222). From that subset,
we selected respondents who participated at wave 4 (n = 1,930). We omitted 11 respondents
whose spouses were deceased by wave 4.

NIH-PA Author Manuscript

At wave 4, respondents again provided details about their romantic relationship histories.
We used respondents’ wave 3 and wave 4 reports of their wedding dates and the number of
times they had ever married to identify the wave 4 relationship reports that corresponded to
respondents’ marriages at the time of the wave 3 interview. Our analyses of time to marital
dissolution used data from all 1,919 respondents who participated at wave 4 and whose
spouses were not deceased by wave 4. Our analyses of whether incarceration predicted
relationship characteristics used data from the 1,847 (96 percent of) respondents whose
longitudinal relationship reports could be matched with confidence; unlike the analyses of
time to marital dissolution, these models required observed marriage end dates. Following
Kreager and colleagues (2013), matches were cases where respondents’ wave 3 and wave 4
reports of the marriage start month and year were identical, or where their wave 3 and wave
4 reports of their marriage start dates differed by less than 2 years and they reported only
one marriage during that time span.1 We did not count as matched 6 of the 1,853 marriages
that met these criteria because these 6 respondents’ wave 4 reports of both the beginning and

1Restricting the latter criterion to a difference of less than 1 year resulted in a loss of 65 cases and did not change the study’s
substantive findings. Even without this time span restriction, most of the 72 unmatched cases still would have gone unmatched for
reasons including wave 4 reports of wedding dates that followed the date of the wave 3 interview, and multiple wave 4 relationships
that could have been matched to the wave 3 marriage.
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 8

end dates of the marriage preceded their wave 3 reports of the marriage start date; we could
not be certain that these reports referred to the same marriages.

NIH-PA Author Manuscript

FOCAL OUTCOME VARIABLE
Our main dependent variable is a dichotomous measure of marital relationship dissolution
by the time of the wave 4 interview (0 = no, 1 = yes). Relationships were counted as
dissolved if the respondent reported an end date for the marriage, reported that the marriage
was no longer ongoing, or reported that they were not currently in a relationship with their
spouse from wave 3.
FOCAL PREDICTORS

NIH-PA Author Manuscript

At wave 4, respondents who spent time in a jail, prison, or other correctional facility
reported how old they were when this first happened, as well as when this most recently
happened for those with multiple incarcerations. They also reported the total amount of time
they had ever spent incarcerated during adulthood (mean = 3.9 months). Approximately 63
percent of ever-incarcerated married respondents served a total of 1 month or less, 20
percent served 2 to 5 months, 8 percent served 6 to 11 months, and 9 percent served 1 year
or more. These lengths of time served are lower than those in the study by Massoglia and
colleagues (2011), but they are comparable with those in the study by Apel and colleagues
(2011). Nationally, 80 percent of jail inmates serve less than 1 month, the average state
prison inmate serves approximately 16 months, and the average federal prison inmate serves
slightly less than 3 years (Bonczar, 2011; Noonan, 2010; Pew Charitable Trusts, 2012). It is
likely that our findings speak mainly to the effects of jail sentences, whereas the findings of
Massoglia and colleagues speak more to the effects of longer prison sentences. Given these
differences in data sources and samples, it is noteworthy that the effects of incarceration on
divorce are highly consistent across studies. In ancillary analyses, we observed that even
lifetime incarceration stays totaling less than a month were associated with marital
disruption. This association became more pronounced as total time incarcerated increased.

NIH-PA Author Manuscript

Using information on respondents’ ages at incarceration, birthdates, wedding dates, and ages
at the time of each interview, we created a dichotomous2 indicator of whether respondents
had been incarcerated during marriage. To help establish the timing of incarceration and
relationship dynamics better, we also subdivided this indicator into separate dichotomous
indicators of whether respondents had been incarcerated during marriage and by wave 3
when relationship dynamics were measured, and whether respondents had been incarcerated
during marriage but only after wave 3. These two indicators are mutually exclusive; the
reference category is never incarcerated during marriage. We examine these indicators as
part of a proposed sequence leading from incarceration to divorce via relationship dynamics.
If this chain is accurate, then the indicator of incarceration during marriage but after wave 3
should not predict relationship characteristics measured at wave 3 (prior to the
incarceration), and relationship characteristics thus should not mediate any association
between this indicator and divorce.
2For seven respondents, our information on the timing of incarceration was not fine-grained enough for us to tell definitively whether
these respondents were more accurately categorized as incarcerated before marriage or as incarcerated during marriage. Because of
this uncertainty, we imputed the timing of incarceration for these cases.
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 9

RELATIONSHIP CHARACTERISTICS

NIH-PA Author Manuscript
NIH-PA Author Manuscript

We examine four marital characteristics, all measured at wave 3, that may explain the
incarceration-divorce association. Low marital love (α = .74) is an item response theory
(IRT) scale of two items assessing how much respondents loved their spouse (0 = a lot, 3 =
not at all) and how much they thought their spouse loved them (0 = a lot, 3 = not at all).
Preliminary analyses revealed that 95 percent of respondents who were never incarcerated
during marriage, 76 percent of respondents who were incarcerated during marriage and
before wave 3, and 93 percent of respondents who were incarcerated during marriage and
after wave 3 loved their partners a lot; the corresponding percentages for reports of partners’
love were 93 percent, 79 percent, and 86 percent, respectively. We created this scale and the
other IRT (for scales based on ordinal indicators) and Rasch (for scales based on
dichotomous indicators) scales used in this study using Thissen et al.’s (2003) MULTILOG
7.0 program. IRT and Rasch scaling techniques use measurement models to estimate
respondents’ latent “true” scores on the construct of interest, based on the observed
indicators (Raudenbush, Johnson, and Sampson, 2003). The resulting scores have desirable
statistical properties: They are approximately normally distributed and, unlike summative
scales, can accommodate items with different numbers of response choices, are not
dominated by the most commonly endorsed items, and are not dependent on the number of
items included.
Economic strain (α = .68) is a Rasch scale of seven items assessing whether in the past 12
months respondents or their households did not pay the full amount of the rent or mortgage
for lack of money; were evicted from a house or apartment for not paying the rent or
mortgage; did not pay the full amount of a gas, electricity, or oil bill for lack of money; had
gas, electric, or oil service turned off because payments were not made; went without
telephone service; did not receive needed medical care because they could not afford it; and
did not receive needed dental care because they could not afford it (0 = no, 1 = yes for each
item). Thirty-six percent of respondents who were never incarcerated during marriage, 66
percent of respondents who were incarcerated during marriage and before wave 3, and 61
percent of respondents who were incarcerated during marriage and after wave 3 experienced
at least one of these forms of economic strain.

NIH-PA Author Manuscript

The IRT scale of relationship violence (α = .81) includes eight items assessing how often in
the past 12 months respondents had threatened, pushed or shoved, or thrown something at
their spouse; slapped, hit, or kicked their spouse; insisted on or made their spouse have
sexual relations when the spouse did not want to; or injured their spouse during a fight, and
how often their spouse had done each of these things to them (0 = never, 4 = 6 or more
times). Preliminary analyses revealed that respondents incarcerated during marriage and
before wave 3 had the highest unadjusted base rates of each form of violence, with 55
percent reporting threats, shoves, or thrown objects; 48 percent reporting slaps, hits, or
kicks; 26 percent reporting forced sexual relations; and 19 percent reporting injurious
violence. The unadjusted base rates for respondents incarcerated during marriage and after
wave 3 were lower (45 percent, 37 percent, 15 percent, and 16 percent, respectively), and
those for respondents who were never incarcerated during marriage were lower still (27
percent, 21 percent, 9 percent, and 7 percent, respectively). Relationship violence was most

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 10

often bidirectional; only 18 percent, 16 percent, and 15 percent of the three groups,
respectively, reported unilateral violence (committed by only one spouse) in the marriage.

NIH-PA Author Manuscript

Finally, extramarital sex is a dichotomous indicator of whether respondents reported being
in a current sexual relationship with someone besides their spouse, had been married more
than 1 year and reported having more than one past-year sexual partner, or reported that
their spouse had had other sexual partners during their relationship (0 = no, 1 = yes). Unlike
the items used to assess the other marital problems, the items assessing extramarital sex are
not well suited to distinguishing among levels of intensity of this problem, so we measure it
as a dichotomy.
CONTROL VARIABLES

NIH-PA Author Manuscript
NIH-PA Author Manuscript

We include measures of respondents’ male gender (0 = no, 1 = yes) and of their race and
ethnicity, measured as a set of dummy variables indicating Black (0 = no, 1 = yes), Hispanic
(0 = no, 1 = yes), or other non-White race (0 = no, 1 = yes); White is the omitted reference
category. Co-resident child indicates whether the respondent had a son or daughter who
lived in the same household (0 = no, 1 = yes). Non-co-resident child indicates whether the
respondent had a child who did not live in the same household (0 = no, 1 = yes). Educational
attainment is measured as the highest level of education respondents had completed (1 =
eighth grade or less, 5 = some graduate school). We include a dichotomous indicator of
whether the respondent was employed full-time at wave 3 (0 = no, 1 = yes). Religiosity (α = .
91) is measured as the mean of the Z scores of 10 items assessing respondents’ frequency of
religious service attendance and participation in organized religious activities over the past
year (for both, 0 = never, 6 = more than once a week), how important their religious faith
and spiritual life were to them (for both, 0 = not important, 3 = more important than
anything else), the extent to which they were religious and spiritual (for both, 0 = not at all,
3 = very), the extent of their agreement that they were being “led” spiritually and that they
used their spiritual beliefs as a basis for how to act and live (for both, 1 = strongly disagree,
5 = strongly agree), how often they prayed privately (0 = never, 7 = more than once a day),
and a count of the hours per week they spent in religious activities at home. Hard drug use
indicates whether the respondent had used cocaine, crystal meth, or other hard drugs in the
year prior to wave 3 (0 = no, 1 = yes). Problem drinking is the sum of three ordinal items
assessing the number of past-year school or work, interpersonal, and health or safety
problems the respondent had because of drinking (0 = none, 2 = two or more). Problem
gambling is a dichotomous indicator of whether the respondent spent a lot of time thinking
about or planning gambling, gambled to relieve uncomfortable feelings, gambled to get even
after losses, or had relationship problems as a result of gambling (0 = no, 1 = yes). Nonfamily violence indicates whether the respondent had used a weapon to get something from
someone, taken part in a group fight, used a weapon in a fight, or brought a gun to school or
work in the year prior to wave 3 (0 = no, 1 = yes). Prior domestic violence conviction
indicates whether the respondent reported being convicted of domestic violence before wave
3 (0 = no, 1 = yes); five respondents did. Incarcerated before marriage is a dichotomous
indicator of whether respondents had ever been incarcerated before marrying their wave 3
spouse (0 = no, 1 = yes). Finally, important relationship covariates include the respondent’s
age at marriage, the couple’s number of years married, and whether the couple had

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 11

cohabited before marriage (0 = no, 1 = yes). Information on race and ethnicity came from
wave 1; the other covariates were measured at wave 3.

NIH-PA Author Manuscript

Table 1 shows descriptive statistics for the study variables among respondents with matched
wave 3 and wave 4 relationship reports, separately for respondents who were never
incarcerated during marriage (n = 1,752), those who were incarcerated during marriage and
before wave 3 (n = 33), and those incarcerated during marriage and after wave 3 (n = 62).
Five percent of respondents were incarcerated during marriage. Other studies on
incarceration and divorce using nationally representative data have found that incarceration
among married people is similarly rare. For instance, 49 men had been incarcerated during
marriage in Lopoo and Western’s (2005) analyses of the NLSY 1979 cohort, and our own
ancillary analyses of data from the NLSY 1997 cohort revealed that 2 percent of married
respondents were incarcerated during the marriage.3 As shown in table 1, a higher
percentage of females, Whites, and respondents with higher levels of education were in the
category of never incarcerated during marriage (i.e., column 1) compared with those
incarcerated during marriage (i.e., columns 2 and 3). Respondents who were never
incarcerated during a marriage, compared with those who were, also had lower levels of
problem drinking, problem gambling, non-family violence, and hard drug use.

NIH-PA Author Manuscript

Table S.1 of the online supporting information compares the descriptive statistics of cases
with and without matched longitudinal relationship reports.4 The 72 respondents whose
marriages could not be matched with confidence were more likely to be male, non-White,
and non-resident parents; had less education; married younger, had been married longer, and
reported less marital love and more extramarital sex; were more likely to have histories of
incarceration; and showed more hard drug use, problem gambling, problem drinking, and
violence.
ANALYTIC STRATEGY

NIH-PA Author Manuscript

We estimate a series of models relating incarceration during marriage to relationship
characteristics and to marital dissolution. First, we examine the association of incarceration
with the duration-dependent risk of marital dissolution via life table analyses and a discrete
time event history model. These analyses allow examinations of marital duration in the
presence of censoring and require no assumptions about the shape of the baseline hazard. In
the discrete time model, observations were person-years and time was specified as a set of
dummy variables indicating the number of years since marriage; “married this year” was the
omitted reference category. We treat respondents as at risk for divorce until they either
divorced or were right-censored (i.e., still married) at the time of their wave 4 interviews.
These analyses include the small number of cases (n = 72) whose wave 3 marriages could

3Furthermore, studies that have used nationally representative data to examine incarceration effects on other outcomes (e.g.,
employment) also have reported a similar percentage of incarcerated individuals. Indeed, using the NLSY 1997 data, Apel and
Sweeten (2010: 456) reported that only 5 percent of the sample had been incarcerated. Despite the small number of incarcerated
individuals in our data, our analysis was able to estimate strong and significant incarceration effects during marriage net of an
extensive set of control variables. Still, a larger sample size of incarcerated individuals may have allowed us to detect additional
effects or differences that went undetected in the current study.
4Additional supporting information can be found in the listing for this article in the Wiley Online Library at http://
onlinelibrary.wiley.com/doi/10.1111/crim.2011.52.issue-3/issuetoc.
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 12

not be matched to wave 4 reports of marital outcomes; we treat these cases as right-censored
at the time of their wave 3 interviews.

NIH-PA Author Manuscript

Second, we estimate three linear regressions and one logistic regression predicting marital
characteristics at wave 3 (low marital love, economic strain, relationship violence, and
extramarital sex) from incarceration during marriage and the control variables. These
models will show the degree of association between these potential mediators and
incarcerations that occurred during marriage and by wave 3. They also will show whether
respondents who would be incarcerated during marriage, but had not yet been by wave 3,
already showed marital problems at wave 3. These analyses use data from the 1,847 cases
with matched longitudinal relationship reports because they require information on postwave 3 marital incarcerations, which was not available for respondents with unknown
marriage outcomes and end dates.

NIH-PA Author Manuscript

Third, we estimate discrete time models predicting relationship dissolution from
incarceration before marriage, incarceration during marriage and by wave 3, and
incarceration during marriage but only after wave 3, once without and once with the
measures of relationship characteristics. These analyses include, but treat as censored, the
small number of cases with unmatched marital outcomes. We compare coefficients from
these two models using Karlson, Holm, and Breen’s (2012) test for indirect effects to
determine whether relationship characteristics are statistically significant mediators of the
association between incarceration and relationship dissolution. Because the relationship
characteristics were measured at wave 3, if they are mediators rather than confounds, we
would expect their inclusion in the model to reduce only the coefficient for incarceration
during marriage and by wave 3. The results of several robustness checks also are noted in
the Results section.

NIH-PA Author Manuscript

Finally, we repeat our main analyses on data from the subset of our sample that was male.
Males have a much higher incarceration rate than do females—11 percent of males in our
sample, versus 2 percent of females, were incarcerated during marriage—and most past
studies of this topic focused on males (e.g., Apel et al., 2010; Massoglia, Remster, and King,
2011). Although there were too few incarcerated females in our sample to allow tests of
gender differences, the findings for males provide a closer point of comparison for the
findings of past studies as well as some insight into the potential for gender differences in
the observed associations.
Add Health used a stratified and clustered sampling strategy. This study focuses only on
married respondents and does not aim to provide estimates for the total population, so we
present unweighted analyses. Although tests revealed that in nearly all of our models the
clustered nature of the sample did not violate the assumption of independence of residuals,
residual levels of economic hardship did vary systematically across sampling units. To
adjust for this dependence, we include a fixed effect (i.e., a set of dummy variables) for
sampling unit in the model predicting economic hardship. We used multiple imputation to
reduce potential bias from item-missing data (Carlin, Galati, and Royston, 2008; Royston,
2005). More specifically, we created ten complete data sets featuring imputed values for
missing cases and combined estimates across the ten following Rubin’s (1987) rules.

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 13

RESULTS
NIH-PA Author Manuscript

RELATING INCARCERATION TO MARITAL DURATION AND DISSOLUTION
The first phase of our analysis established the magnitude and the timing of the association
between incarceration and marital dissolution in this sample. As an initial step, we examined
survival functions for marriage among respondents who were and were not incarcerated at
some point during the marriage. A plot of the survival function created via the actuarial
method (see figure S.1 in the online supporting information) revealed a widening gap
between these groups in the probability of remaining married, beginning between the fourth
and fifth years of marriage and increasingly favoring the nonincarcerated group over time (p
< .05 for group difference). This finding indicates that the marriages of respondents who
experienced an incarceration during marriage lasted fewer months than did the marriages of
other respondents. Additional exploration of the data indicated that the point of divergence
of the curves approximately corresponded to the average number of years into the marriage
at which the marital incarcerations occurred.

NIH-PA Author Manuscript

We next examined whether incarceration was associated specifically with subsequent
marital dissolution net of controls for demographic, behavioral, and relationship history
factors. Table 2 shows the results of a discrete time model predicting the risk of marital
dissolution in a given year from a dichotomous time-varying indicator of whether the
respondent had been incarcerated by that year and from the control variables. Even when
factors such as religiosity, substance use, and age at marriage were accounted for,
incarcerations occurring during a marriage were associated with an increased risk of that
marriage dissolving (b = .70, p < .001). Exponentiating the coefficient to obtain the odds
ratio indicated that incarceration during marriage was associated with 102 percent higher
odds of marital dissolution. In contrast, incarcerations occurring before a marriage did not
significantly increase the odds of that marriage dissolving (b = .19, p > .05).

NIH-PA Author Manuscript

To examine time to divorce after an incarceration, we estimated survivor and hazard
functions for postincarceration marital duration among the subset of 107 respondents who
were incarcerated during marriage. Figure 1 shows that the probability that such a
respondent remained married declined steadily across the years after the incarceration before
leveling off approximately 6 years later. The hazard function (not shown) indicated that the
rate of marital dissolution among these respondents hovered around .15 for the first six time
intervals, and then it declined to nearly zero. As noted earlier, some respondents experienced
the incarceration by the wave 3 interview, and others after. The survival curves for these two
groups were comparable. However, by design, the uncensored postincarceration observation
period tended to be longer for respondents who were incarcerated by wave 3 (more than 6
years on average) than for respondents incarcerated only after wave 3 (3 years on average).
The estimated survival functions imply that in the absence of censoring, we might have
observed more marital dissolutions among the latter group.

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 14

RELATIONSHIP PROBLEMS AMONG RESPONDENTS INCARCERATED DURING
MARRIAGE

NIH-PA Author Manuscript
NIH-PA Author Manuscript

We next examined whether incarceration during marriage predicted the hypothesized
mediators of the association between incarceration and divorce. These models compared the
associations of pre-versus post-wave 3 incarcerations with relationship characteristics
measured at the wave 3 interview. The linear and logistic regression coefficients shown in
table 3 indicate that relative to respondents who were not incarcerated during marriage,
respondents who were incarcerated during marriage and by wave 3 showed less marital love
(b = .17, p < .001), more economic strain (b = .48, p < .001), and more relationship violence
(b = .35, p < .001), as well as higher log-odds of extramarital sex (b = .83, p < .05) at wave
3. In contrast, respondents who were incarcerated during marriage but only after wave 3 did
not show less marital love (b = .02, p > .05), more relationship violence (b = .10, p > .05), or
higher log-odds of extramarital sex (b = −.17, p > .05) at wave 3, prior to their
incarcerations. For all three of these outcomes, the nonsignificant coefficients for post-wave
3 incarceration were significantly different from the significant coefficients for pre-wave 3
incarceration (all ps for differences < .05). Post-wave 3 incarceration was associated with
wave 3 economic strain, however (b = .26, p < .01). Although its coefficient was 46 percent
smaller than the coefficient for pre-wave 3 incarceration, the two coefficients were not
statistically distinguishable (p for difference > .05).5
USING RELATIONSHIP PROBLEMS TO EXPLAIN THE INCARCERATION-DIVORCE
ASSOCIATION

NIH-PA Author Manuscript

We next estimated two discrete time models predicting relationship dissolution from
incarceration before marriage, a time-varying indicator of incarceration during marriage and
by wave 3, and a time-varying indicator of incarceration during marriage and after wave 3.
Model 1 estimated the effect of incarceration on the risk of divorce net of the controls;
model 2 added the relationship characteristics hypothesized to mediate the association.
Table 4 shows the results. Model 1 shows that incarcerations occurring during marriage and
by wave 3 positively predicted marital dissolution (b = .68, p < .001). Exponentiating this
coefficient revealed that these respondents had nearly twice the odds of marital dissolution
by wave 4 as did respondents not incarcerated during marriage (e.68 = 1.97). Incarcerations
occurring during marriage and after wave 3 had similar associations with marital dissolution
(b = .71, p < .001, odds ratio [OR] = 2.03).
Model 2 of table 4 shows that as expected, relationship dissolution was significantly
predicted by low marital love, economic strain, relationship violence, and extramarital sex.
Our main interest is in the effect of these characteristics on the coefficients for incarceration.
The significance tests for indirect effects reveal that wave 3 relationship characteristics
significantly reduced the effect of prior incarceration during marriage on relationship
dissolution (p for reduction in coefficient < .05). Specifically, the remaining direct effect
was 40 percent smaller than the original coefficient; it also was no longer statistically
5In multiple sets of ancillary analyses, we included in these models cases with unmatched longitudinal relationship reports by making
various assumptions about these cases’ post-wave 3 marriage end dates, which were needed to code these cases’ post-wave 3 marital
incarcerations. The results of these models were similar and sometimes identical to the results presented here, although in some cases
the difference in coefficients for pre- and post-wave 3 incarceration in predicting relationship violence was not statistically significant.
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 15

NIH-PA Author Manuscript

significant. When added individually to model 1 of table 4, each of the four characteristics
produced a decline in the incarceration coefficient, but low marital love and extramarital sex
produced larger declines (of 19 percent and 25 percent, respectively, p for both reductions
< .05) than did economic strain (4 percent, p < .05) and relationship violence (7 percent, p
< .10). In addition, only low marital love and extramarital sex had unique mediating effects,
producing declines in the incarceration coefficient even net of the other three characteristics
and the controls (p < .05). Because table 3 suggests that economic strain had uncertain
sequencing with incarceration, we examined the joint indirect effect of love, relationship
violence, and extramarital sex while accounting for economic strain and the controls. These
three relationship characteristics reduced the remaining effect of prior incarceration during
marriage by 37 percent (from .65 to .41; p for reduction in coefficient < .05). The addition of
relationship characteristics to the model did not significantly change the effect of subsequent
incarceration during marriage or the effect of incarceration before marriage.
RESULTS FOR THE MALE SUBSAMPLE

NIH-PA Author Manuscript

Finally, we examined the associations of incarceration with marital characteristics and
dissolution among male respondents (see table S.2). Most of the coefficients found for this
subsample were similar to those found for the whole sample, but some significance levels
changed. For instance, the association between low marital love and incarceration remained
positive but was not statistically significant among the male subsample. Similarly, the
positive association between incarceration and economic strain among males was only
significant for pre-wave 3 incarceration. The relationship characteristics explained a
statistically significant 35 percent of the effect of pre-wave 3 incarceration, compared with
40 percent among the full sample. Although comparing coefficients in tables 3 and S.2
revealed some hints of potential gender differences, because we could not test them directly,
we cannot draw firm conclusions about gender differences from these analyses. Nonetheless,
we encourage future researchers to examine the possibilities that the incarceration of males
has a weaker impact on love and a stronger impact on household economic strain than does
that of females.
ROBUSTNESS CHECKS

NIH-PA Author Manuscript

Testing Mediation in Reduced Models—The main analyses adjust for many
covariates, but the sample sizes of the key incarceration groups are modest, raising the
possibility that the models are overly complex. Table S.3 illustrates the consistency of the
mediation findings across different reduced form models. Regardless of whether and which
control variables were included, adjusting for relationship characteristics reduced the
coefficients for incarceration during marriage and by wave 3 by at least 53 percent and to
statistical nonsignificance. In some models, adjusting for relationship characteristics
produced modest but significant reductions in the coefficients for incarceration during
marriage and after wave 3. These models all omitted the behavioral control variables.
Ancillary analyses indicated that adding religiosity, hard drug use, and problem drinking,
and in some cases any one of these covariates, to these models eliminated the “mediating”
effect of relationship characteristics for post-wave 3 incarceration. In 16 additional reduced
form models (results available on request), we confirmed the differential effects of pre- and
post-wave 3 incarceration on wave 3 relationship characteristics, with one exception: In a
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 16

NIH-PA Author Manuscript

model with only relationship history controls, the difference in coefficients predicting
extramarital sex, although large (.82 versus .15), was not statistically significant. We thus
conclude that the findings presented earlier are not simply artifacts of the extensive set of
control variables used in the main analyses.
Logistic Regression Models Predicting Relationship Dissolution—We also
verified our results for relationship dissolution via logistic regression models predicting
marital dissolution among cases with matched longitudinal relationship reports. The bottom
panel of table S.3 shows results for full and reduced models. These models confirmed that
incarcerations occurring during marriage and by wave 3 were associated with doubled odds
of marital dissolution by wave 4 (e.g., in the full model, e.85 = 2.34). In contrast, net of the
controls, incarcerations occurring during marriage but after wave 3 did not significantly
predict marital dissolution, although the coefficients were in the expected direction. The
survival curves presented in figure 1 suggest that a longer follow-up period might have
allowed us to observe more divorces among this group, so censoring could explain the
difference between these results and the discrete time results presented earlier.

NIH-PA Author Manuscript

Together, our findings indicate that young adults who are incarcerated while they are
married remain at increased risk for divorce for several years after their release. This
increased risk seems to be in large part because incarceration is associated with lower
emotional and economic rewards derived from the marriage and higher physical risks of
staying in the marriage, and these relationship problems in turn are associated with divorce.
It is possible that these marital problems are confounders rather than mediators of the
incarceration effect. Yet although preincarceration couples do show increased levels of
economic strain, unlike postincarceration couples, they do not seem to love each other less,
engage in more relationship violence, or have higher odds of extramarital sex than do
couples who do not experience an incarceration during marriage. This finding provides
suggestive evidence that marital problems follow from spells of incarceration and thus
mediate the incarceration-divorce association.

DISCUSSION

NIH-PA Author Manuscript

This study used recent data from a contemporary national sample to confirm an important
past finding: that incarceration occurring during marriage, but not before, is associated with
an increased risk of marital dissolution for several years postrelease (Apel et al., 2010;
Lopoo and Western, 2005; Massoglia, Remster, and King, 2011). When combined with our
existing knowledge of the impacts of incarceration, social exchange perspectives suggest
that incarceration disrupts marriages by reducing inducements to stay in them and raising
inducements to leave them (Levinger, 1965). Our findings go beyond past work by
demonstrating that incarcerations occurring during marriage are associated with less love
between spouses, more marital violence, and greater odds of extramarital sex, and that these
factors in turn are associated with increased odds of divorce. It is possible that the
incarceration-divorce association is spurious to, rather than mediated by, these factors. Yet
postincarceration couples, but not preincarceration couples, were the couples that showed
elevated levels of these marital problems. Our results do suggest that preincarceration
couples experience more than their share of economic strain, which also seems to play some

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 17

NIH-PA Author Manuscript

role in the incarceration-divorce association. Still, overall the results are consistent with the
idea that some forms of marital problems are intermediate steps in a sequence leading from
incarceration to marital dissolution.
These marital problems, which also are strong predictors of divorce in the family literature
(Amato, 2010), explained 40 percent of the incarceration-divorce association. This finding
implies that former inmates’ marriages may fail for some of the same reasons that other
people’s marriages fail, and they may fail at a higher rate because couples that experience
incarceration experience more of these marital problems. One criticism of the social
exchange perspective on divorce is that the perspective is silent on why rewards, barriers,
and alternatives might change during the course of a marriage (Karney and Bradbury, 1995).
Our findings suggest that one source of such changes could be spouses’ life events and
experiences with other social institutions, such as the penal system. This possibility
underscores the potential for criminological and family scholarship to inform each other.

NIH-PA Author Manuscript

Many scholarly predictions about the reduced respect, trust, and loyalty that may follow
incarcerations were developed from research on lengthy spells of imprisonment (e.g.,
Braman, 2004; Harman, Smith, and Egan, 2007; Nurse, 2002). In addition, many past
studies linking incarceration with divorce examined spells longer than those examined here
(e.g., Lopoo and Western, 2005; Massoglia, Remster, and King, 2011). Our study found a
comparably large link between incarcerations of only a few months or less and later divorce.
For our interpretation of our findings to be correct, these brief spells would need to be
sufficient to damage couples’ love, fidelity, and conflict resolution, and possibly their
financial well-being. We would need longitudinal data on pre- and postincarceration marital
characteristics to test these ideas directly. In the absence of such data, the possibility remains
that these couples already had marital problems before experiencing incarceration. If true,
than rather than pointing to mechanisms of the incarceration effect, our results point to
important relational confounds that should be accounted for in future studies of incarceration
and relationship dissolution. If our results do capture mechanisms, we can say only that
these mechanisms follow incarcerations of modest length; they may not explain the effects
of longer prison stays.

NIH-PA Author Manuscript

This study thus also relates to important theoretical debates about the aspects of
incarceration that undermine former inmates’ social integration. Broadly, incarceration
could affect the costs, rewards, and alternatives of a marriage via stigma, that is, by
tarnishing inmates’ reputations and causing people to want to disassociate from them.
Alternatively, it could affect these things via incapacitation, that is, by making it physically
difficult for inmates to participate in and contribute to their relationships (Apel et al., 2010;
Massoglia, Remster, and King, 2011). Although the effects of incarceration may grow
stronger as incarceration length increases (Massoglia, Remster, and King, 2011), our
findings suggest that incarcerations need not be lengthy to produce divorce. The brevity of
these apparently impactful incarcerations may imply that stigma does play a role in shaping
marital love, conflict, openness to outside relationships, and ultimately divorce. We do not
think this possibility is incompatible with the idea that longer incarcerations may be more
detrimental to marital quality or that they may trigger additional destabilizing mechanisms,
and indeed findings to that effect would bridge many of the differences between this study

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 18

and past works. Future research should examine whether our findings generalize to longer
prison stays.

NIH-PA Author Manuscript

Our attempt to identify intervening processes may tell us something meaningful about
incarceration’s unintended consequences across a wide number of conventional domains.
One of the many domains disrupted by incarceration is marriage. Although several studies
have reported that incarceration increases marital instability (Lopoo and Western, 2005;
Massoglia, Remster, and King, 2011; Western, 2006), the mechanisms that link
incarceration to marital dissolution have remained elusive. By examining potential
mechanisms, this study provides suggestive evidence on why incarceration might affect
marital stability, namely, by adversely affecting important relationship dynamics that are
strongly tied to marital cohesion. To the extent that this is the case, the patterns we observed
indicate that incarceration may be implicated in the production of social inequalities (e.g.,
deficits in love, domestic violence, and infidelity) within marital relationships that increase
the odds of marital instability.

NIH-PA Author Manuscript
NIH-PA Author Manuscript

More broadly, our results are in line with many studies that have highlighted the harmful
effects of incarceration on positive life outcomes (Apel and Sweeten, 2010; Hagan and
Dinovitzer, 1999; Patillo, Weiman, and Western, 2004; Petersilia, 2003; Pettit and Western,
2004; Uggen, Manza, and Thompson, 2006). This body of research has provided substantial
evidence that incarceration inhibits prosocial life-course transitions that can lead to
cumulative and compounded disadvantages as ex-inmates return home (Laub and Sampson,
2003; Pettit, 2012; Uggen, 2000; Wakefield and Uggen, 2010; Western, 2006). Indeed,
incarceration sanctions are intended to reduce or disrupt an individual’s offending trajectory
and simultaneously deter crime. However, research has continued to show that incarceration
has unintended consequences by disrupting conventional achievement prospects that have
been shown to lower criminal offending (e.g., employment, marriage; Hagan and
Dinovitzer, 1999; Huebner, 2005; Lopoo and Western, 2005; Sampson and Laub, 1993;
Uggen, 2000). For example, stable employment and quality marriages have been linked to
desistance by serving as a source of informal social control for criminal offenders (Sampson
and Laub, 1993). On the other hand, persistent offending is more likely among those who
fail to obtain prosocial stakes in conformity. Thus, this broad line of evidence suggests that
incarceration and existing criminal justice penal policies have several unintended
consequences that lead to future offending or other negative life outcomes (Patillo, Weiman,
and Western, 2004; Petersilia, 2003).
Our findings have important implications for policy and intervention. Most directly, they
indicate that efforts to build marital closeness, strengthen marital commitment, and promote
nonviolent conflict resolution among couples who have experienced incarceration could
help preserve current and former inmates’ marriages. Efforts to improve couples’ financial
health could have similar effects. Multipronged interventions that address marital quality,
fidelity, domestic violence, and family economic well-being together could be most
effective for three reasons. First, these relationship characteristics best explained the
incarceration-divorce association when considered jointly (cf. Amato and Rogers, 1997;
Harman, Smith, and Egan, 2007). Second, the link between incarceration and domestic
violence indicates that efforts to promote marital stability without comprehensively

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 19

NIH-PA Author Manuscript

addressing these couples’ problems could inadvertently put some spouses at risk. Third,
preserving poor quality marriages likely would do little to prevent recidivism (Sampson and
Laub, 1993) and might worsen marital conflict (Levinger, 1965). We thus suggest that the
goal of intervention should not be the simple preservation of all of these couples’ marriages,
but instead it should be the amelioration of the specific relational stresses caused by
incarceration.

NIH-PA Author Manuscript

Our data covered people who were married before their late 20s and followed those
marriages prospectively for 6 years. At wave 4, it was still too soon to know what would
become of all of these marriages, especially those of respondents incarcerated after wave 3.
In addition, although our sample was observed at a key age range for incarceration, studying
this age range necessarily means that we were studying relatively early marriages. People
who marry by their mid-20s are more likely to divorce when the marriage turns bad,
suggesting that they may be subject to fewer constraints against divorce or less resilient to
marital stressors (Glenn, Uecker, and Love, 2010). Yet people who are incarcerated at later
ages may be more persistent or serious offenders who are at increased risk for a variety of
negative outcomes. Not enough research exists on the overlap between the incarcerated and
married populations for us to predict with confidence how our results might differ in an
older sample. It is noteworthy that we observe the same general incarceration-divorce
relationship found in data capturing a broader age range. Still, future research should
examine whether our findings generalize to other marital contexts such as marriages
occurring later in life, marriages of longer duration, and remarriages.

NIH-PA Author Manuscript

This study has some additional limitations. First, like most studies of this topic, our data
only captured the incarceration experiences of one spouse (the respondent). Although only
one partner’s negative marital experience may be enough to end the marriage, it is possible
that respondents’ spouses had different views of the marriage. In addition, our reference (no
incarceration) groups may have included some respondents whose spouses had been
incarcerated. This possibility probably makes our estimates conservative tests of the
differences between couples who are and are not affected by incarceration, but information
on both partners’ experiences would be desirable. In addition, despite the relative rarity of
incarceration among married people, we detected strong effects of incarceration on marital
processes, but our incarcerated samples were too small for us to examine variations in
effects across demographic subgroups. Future studies should examine whether these
processes differ across genders, racial and ethnic groups, parental status, and other important
dimensions. Despite these limitations, this study offers the advantages of prospective
information on marriages and incarcerations, and a rare look at the potential mechanisms
behind former inmates’ marital instability.
In conclusion, our study adds to a growing body of research documenting the negative
impact of incarceration on family well-being (Giordano, 2010; Wildeman, 2010; Wildeman,
Schnittker, and Turney, 2012). Although our findings contribute important information
about why former inmates’ marriages fail, there is more work to be done. We explained
some of the incarceration effect using a partial list of marital costs, rewards, and alternatives,
but other indicators as well as barriers to marriage dissolution are theoretically relevant
(Previti and Amato, 2003). We must determine the relevance of these factors not only to

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 20

NIH-PA Author Manuscript

improve our understanding of inmates’ marriages but also because many of these couples
have children who may keep unhappy, conflict-ridden marriages intact (Previti and Amato,
2003). In addition, small-scale implementations of marital interventions in correctional
facilities have reported beneficial effects on inmates’ relationship skills (Einhorn et al.,
2008; Shamblen et al., 2012). Our findings suggest that such interventions could have longterm benefits for marital cohesion and stability, and this possibility should be tested. Finally,
although marital dissolution could be an intermediate link between incarceration and other
important outcomes, we know little about the consequences of former inmates’ divorces.
Most of the available evidence suggests that family disruption harms inmates, spouses, and
children, but some families could benefit when an offender is removed from the household
(Johnson and Easterling, 2012). A better understanding of incarceration’s effects on marital
process could shed light not only on marital outcomes but also on broader individual and
family adjustment and development.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.

NIH-PA Author Manuscript

Acknowledgments
We thank Dan Mears for providing national statistics on sentence length, Wayne Osgood for providing statistical
advice, Alex Widdowson for providing tabulations of NLSY data, and editor Rosemary Gartner and the anonymous
reviewers for their thoughtful and helpful feedback on the manuscript. Funding for this research and resulting
publication was provided in part by the Florida State University Council on Research & Creativity through a
Committee on Faculty Research Support award (to the first author). This research uses data from Add Health, a
program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and
Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by grant P01-HD31921 from
the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding
from 23 other federal agencies and foundations. Special acknowledgment is due Ronald R. Rindfuss and Barbara
Entwisle for assistance in the original design. Information on how to obtain the Add Health data files is available on
the Add Health website (http://www.cpc.unc.edu/addhealth). No direct support was received from grant P01HD31921 for this analysis.
Grant Number: R24 HD041025

Biographies
NIH-PA Author Manuscript

Sonja E. Siennick is an assistant professor in the College of Criminology and Criminal
Justice at Florida State University. Her research examines the interpersonal causes and
consequences of crime and deviance over the life course, with recent emphasis on family
relationships and on incarceration.
Eric A. Stewart is a professor in the College of Criminology and Criminal Justice at Florida
State University. He is a member of the Racial Democracy, Crime and Justice Network. His
research interests include racial inequality and criminal outcomes, crime over the life course,
and contextual processes and microprocesses that affect adolescent development.
Jeremy Staff is an associate professor of Criminology and Sociology at The Pennsylvania
State University. His research and teaching interests include criminology, stratification, and
the life course. He is currently studying how family, school, and work transitions are

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 21

associated with fluctuations in alcohol use and misuse, as well as the consequences of heavy
drinking with respect to midlife socioeconomic attainment, health, and mortality.

NIH-PA Author Manuscript

REFERENCES

NIH-PA Author Manuscript
NIH-PA Author Manuscript

Amato, Paul R. Research on divorce: Continuing trends and new developments. Journal of Marriage
and Family. 2010; 72:650–66.
Amato, Paul R.; Bryndl, Hohmann-Marriott. A comparison of high and low-distress marriages that end
in divorce. Journal of Marriage and Family. 2007; 69:621–38.
Amato, Paul R.; Denise, Previti. People’s reasons for divorcing: Gender, social class, the life course,
and adjustment. Journal of Family Issues. 2003; 24:602–26.
Amato, Paul R.; Rogers, Stacy J. A longitudinal study of marital problems and subsequent divorce.
Journal of Marriage and the Family. 1997; 59:612–24.
Apel, Robert; Blokland, Arjan A. J.; Paul, Nieuwbeerta; Marieke van, Schellen. The impact of
imprisonment on marriage and divorce: A risk set matching approach. Journal of Quantitative
Criminology. 2010; 26:269–300.
Apel, Robert; Gary, Sweeten. The impact of incarceration on employment during the transition to
adulthood. Social Problems. 2010; 57:448–79.
Arditti, Joyce A. Parental Incarceration and the Family: Psychological and Social Effects of
Imprisonment on Children, Parents, and Caregivers. NYU Press; New York: 2012.
Arditti, Joyce A.; Tiffaney, Parkman. Young men’s reentry after incarceration: A developmental
paradox. Family Relations. 2011; 60:205–20.
Benson, Michael L.; Leanne Fiftal, Alarid; Burton, Velmer S.; Cullen, Francis T. Reintegration or
stigmatization? Offenders’ expectations of community re-entry. Journal of Criminal Justice. 2011;
39:385–93.
Bonczar, Thomas P. Prevalence of Imprisonment in the U.S. Population, 1974-2001. U.S. Department
of Justice, Office of Justice Programs, Bureau of Justice Statistics; Washington, DC: 2003.
Bonczar, Thomas P. State Prison Releases, 2008: Time Served in Prison, by Offense and Release
Type. U.S. Department of Justice, Office of Justice Programs, Bureau of Justice Statistics;
Washington, DC: 2011.
Braman, Donald. Doing Time on the Outside: Incarceration and Family Life in Urban America.
University of Michigan Press; Ann Arbor: 2004.
Carlin, Josh B.; Galati, John C.; Patrick, Royston. A new framework for managing and analyzing
multiply imputed data in Stata. The Stata Journal. 2008; 8:49–67.
Collett, Jessica L. Integrating theory, enhancing understanding: The potential contributions of recent
experimental research in social exchange for studying intimate relationships. Journal of Family
Theory & Review. 2010; 2:280–98.
Comfort, Megan. Doing Time Together: Love and Family in the Shadow of the Prison. University of
Chicago Press; Chicago, IL: 2008.
Conger, Rand D.; Elder, Glen H., Jr.; Lorenz, Frederick O.; Conger, Katherine J.; Simons, Ronald L.;
Whitbeck, Les B.; Shirley, Huck; Melby, Janet N. Linking economic hardship to marital quality
and instability. Journal of Marriage and the Family. 1990; 52:643–56.
DeMaris, Al. Burning the candle at both ends: Extramarital sex as a precursor of marital disruption.
Journal of Family Issues. 2013; 34:1474–99. [PubMed: 24634559]
Einhorn, Lindsey; Tamra, Williams; Scott, Stanley; Nicole, Wunderlin; Howard, Markman; Joanne,
Eason. PREP inside and out: Marriage education for inmates. Family Process. 2008; 47:341–56.
[PubMed: 18831311]
Fishman, Laura T. Women at the Wall: A Study of Prisoners’ Wives Doing Time on the Outside.
SUNY Press; Albany, NY: 1990.
Giordano, Peggy C. Legacies of Crime: A Follow-Up of the Children of Highly Delinquent Girls and
Boys. Cambridge University Press; New York: 2010.
Glenn, Norval D.; Uecker, Jeremy E.; Love, Robert W. B. Later first marriage and marital success.
Social Science Research. 2010; 39:787–800. [PubMed: 22948068]

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 22

NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Goffman, Alice. On the run: Wanted men in a Philadelphia ghetto. American Sociological Review.
2009; 74:339–57.
Hagan, John; Ronit, Dinovitzer. Collateral consequences of imprisonment for children, communities,
and prisoners. In: Tonry, Michael H.; Joan, Petersilia, editors. Crime and Justice: A Review of
Research. Vol. 26. University of Chicago Press; Chicago, IL: 1999.
Harman, Jennifer J.; Smith, Vernon E.; Egan, Louisa C. The impact of incarceration on intimate
relationships. Criminal Justice and Behavior. 2007; 34:794–815.
Heaton, Tim B.; Albrecht, Stan L. Stable unhappy marriages. Journal of Marriage and the Family.
1991; 53:747–58.
Huebner, Beth M. The effect of incarceration on marriage and work over the life course. Justice
Quarterly. 2005; 22:281–303.
Johnson, Elizabeth I.; Beth, Easterling. Understanding unique effects of parental incarceration on
children: Challenges, progress, and recommendations. Journal of Marriage and the Family. 2012;
74:342–56.
Karakurt, Gunnur; Christiansen, Abigail T.; Wadsworth, Shelley M.; Weiss, Howard M. Romantic
relationships following wartime deployment. Journal of Family Issues. 2013; 34:1427–51.
Karlson, Kristian B.; Anders, Holm; Richard, Breen. Comparing regression coefficients between samesample nested models using logit and probit: A new method. Sociological Methodology. 2012;
42:286–313.
Karney, Benjamin R.; Bradbury, Thomas N. The longitudinal course of marital quality and stability: A
review of theory, methods, and research. Psychological Bulletin. 1995; 118:3–34. [PubMed:
7644604]
Kreager, Derek A.; Felson, Richard B.; Cody, Warner; Wenger, Marin R. Women’s education, marital
violence, and divorce: A social exchange perspective. Journal of Marriage and the Family. 2013;
75:565–81. [PubMed: 24357879]
Kurdek, Lawrence A. Predicting the timing of separation and marital satisfaction: An eight-year
prospective longitudinal study. Journal of Marriage and the Family. 2002; 64:163–79.
Laub, John H.; Sampson, Robert J. Shared Beginnings, Divergent Lives: Delinquent Boys to Age 70.
Harvard University Press; Cambridge, MA: 2003.
Lavner, Justin A.; Bradbury, Thomas N.; Karney, Benjamin R. Incremental change or initial
differences? Testing two models of marital deterioration. Journal of Family Psychology. 2012;
26:606–16. [PubMed: 22709260]
Levinger, George. Marital cohesiveness and dissolution: An integrative review. Journal of Marriage
and the Family. 1965; 27:19–28.
Lewis, Robert A.; Spanier, Graham B. Theorizing about the quality and stability of marriage. In: Burr,
Wesly R.; Reuben, Hill; Nye, F. Ivan; Reiss, Ira L., editors. Contemporary Theories about the
Family. Vol. 2. The Free Press; New York: 1979.
Lopoo, Leonard M.; Bruce, Western. Incarceration and the formation and stability of marital unions.
Journal of Marriage and Family. 2005; 67:721–34.
Massoglia, Michael; Brianna, Remster; King, Ryan D. Stigma or separation? Understanding the
incarceration-divorce relationship. Social Forces. 2011; 90:133–55.
Noonan, Margaret. Mortality in Local Jails, 2000-2007. U.S. Department of Justice, Office of Justice
Programs, Bureau of Justice Statistics; Washington, DC: 2010.
Nurse, Anne. Fatherhood Arrested: Parenting from Within the Juvenile Justice System. Nashville, TN;
Vanderbilt University Press: 2002.
Oliver, William; Hairston, Creasie F. Intimate partner violence during the transition from prison to the
community: Perspectives of incarcerated African American men. Journal of Aggression,
Maltreatment and Trauma. 2008; 16:258–76.
Patillo, Mary; David, Weiman; Bruce, Western, editors. Imprisoning America: The Social Effects of
Mass Incarceration. Russell Sage Foundation; New York: 2004.
Petersilia, Joan. When Prisoners Come Home: Parole and Prisoner Reentry. Oxford University Press;
New York: 2003.

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 23

NIH-PA Author Manuscript
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Pettit, Becky; Bruce, Western. Mass imprisonment and the life course: Race and class inequality in
U.S. incarceration. American Sociological Review. 2004; 69:151–69.
Pettit, Becky. Invisible Men: Mass Incarceration and the Myth of Black Progress. Russell Sage
Foundation; New York: 2012.
Pew Charitable Trusts. Time Served: The High Cost, Low Return of Longer Prison Terms. Pew
Charitable Trusts; Philadelphia, PA: 2012. http://www.pewtrusts.org/
our_work_report_detail.aspx?id=85899396348
Previti, Denise; Amato, Paul R. Why stay married? Rewards, barriers, and marital stability. Journal of
Marriage and Family. 2003; 65:561–73.
Raudenbush, Stephen W.; Christopher, Johnson; Sampson, Robert J. A multivariate, multilevel Rasch
model with application to self-reported criminal behavior. Sociological Methodology. 2003;
33:169–211.
Rodrigues, Amy E.; Hall, Julie H.; Fincham, Frank D. Divorce and relationship dissolution: Theory,
research and practice. In: Fine, Mark A.; Harvey, John H., editors. Handbook of Divorce and
Relationship Dissolution. Erlbaum; Mahwah, NJ: 2006.
Royston, Patrick. Multiple imputation of missing values: Update. The Stata Journal. 2005; 5:1–14.
Rubin, Donald B. Multiple Imputations for Nonresponse in Surveys. Wiley; New York: 1987.
Sampson, Robert J.; Laub, John H. Crime in the Making: Pathways and Turning Points Through Life.
Harvard University Press; Boston, MA: 1993.
Shamblen, Stephen R.; Arnold, Brooke B.; Patrick, McKiernan; Collins, David A.; Strader, Ted N.
Applying the Creating Lasting Family Connections Marriage Enhancement Program to marriages
affected by prison reentry. Family Process. 2013; 52:477–98. [PubMed: 24033244]
Thibaut, John W.; Kelley, Harold H. The Social Psychology of Groups. Wiley; New York: 1959.
Thomas, James C.; Levandowski, Brooke A.; Isler, Malika R.; Elizabeth, Torrone; George, Wilson.
Incarceration and sexually transmitted infections: A neighborhood perspective. Journal of Urban
Health. 2008; 85:90–9. [PubMed: 18046653]
Travis, Jeremy; Michelle, Waul. The children and families of prisoners. In: Jeremy, Travis; Michelle,
Waul, editors. Prisoners Once Removed: The Impact of Incarceration and Reentry on Children,
Families, and Communities. Urban Institute Press; Washington, DC: 2003.
Turanovic, Jillian J.; Nancy, Rodriguez; Pratt, Travis C. The collateral consequences of incarceration
revisited: A qualitative analysis of the effects on caregivers of children of incarcerated parents.
Criminology. 2012; 50:913–59.
Turney, Kristin; Jason, Schnittker; Christopher, Wildeman. Those they leave behind: Paternal
incarceration and maternal instrumental support. Journal of Marriage and Family. 2012; 74:1149–
65.
Turney, Kristin; Christopher, Wildeman. Redefining relationships: Explaining the countervailing
consequences of paternal incarceration for parenting. American Sociological Review. 2013;
78:949–79.
Richard, Udry, J. Marital alternatives and marital disruption. Journal of Marriage and the Family.
1981; 43:889–97.
Uecker, Jeremy E.; Stokes, Charles E. Early marriage in the United States. Journal of Marriage and the
Family. 2008; 70:835–46. [PubMed: 20305796]
Uggen, Christopher. Work as a turning point in the life course of criminals: A duration model of age,
employment, and recidivism. American Sociological Review. 2000; 65:529–46.
Uggen, Christopher; Jeff, Manza; Angela, Behrens. Less than the average citizen: Stigma, role
transition, and the civic reintegration of convicted felons. In: Shadd, Maruna; Russ, Immarigeon,
editors. After Crime and Punishment: Pathways to Offender Reintegration. Willan; Devon, U.K.:
2004.
Uggen, Christopher; Jeff, Manza; Melissa, Thompson. Citizenship, democracy, and the civic
reintegration of criminal offenders. The ANNALS of the American Academy of Political and
Social Science. 2006; 605:281–310.
Wakefield, Sara; Christopher, Uggen. Incarceration and stratification. Annual Review of Sociology.
2010; 36:387–406.

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 24

NIH-PA Author Manuscript

Western, Bruce. The impact of incarceration on wage mobility and inequality. American Sociological
Review. 2002; 67:526–46.
Western, Bruce. Punishment and Inequality in America. Russell Sage Foundation; New York: 2006.
White, James M. The current status of theorizing about families. In: Peterson, Gary W.; Bush, Kevin
R., editors. Handbook of Marriage and the Family. Springer; New York: 2013.
White, Lynn K.; Alan, Booth. Divorce over the life course: The role of marital happiness. Journal of
Family Issues. 1991; 12:5–21.
White, Lynn K.; Rogers, Stacy J. Economic circumstances and family outcomes: A review of the
1990s. Journal of Marriage and the Family. 2000; 62:1035–51.
White, Robert J.; Gondolf, Edward W.; Robertson, Donald U.; Goodwin, Beverly J.; Caraveo, L.
Eduardo Extent and characteristics of woman batterers among federal inmates. International
Journal of Offender Therapy and Comparative Criminology. 2002; 46:412–26. [PubMed:
12150081]
Wildeman, Christopher. Paternal incarceration and children’s physically aggressive behaviors:
Evidence from the Fragile Families and Child Wellbeing Study. Social Forces. 2010; 89:285–310.
Wildeman, Christopher; Jason, Schnittker; Kristin, Turney. Despair by association? The mental health
of mothers with children by recently incarcerated fathers. American Sociological Review. 2012;
77:216–43.

NIH-PA Author Manuscript
NIH-PA Author Manuscript
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 25

NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 1. Proportion of Respondents Still Married at Each Year Since Incarceration

Source: National Longitudinal Study of Adolescent Health.

NIH-PA Author Manuscript
Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 26

Table 1

Descriptive Statistics for Study Variables, by Respondent Incarceration

NIH-PA Author Manuscript

Variables

Timing of Incarceration
During
Marriage,
by Wave 3

Never During
Marriage
Mean

SE

Mean

SE

Range

During
Marriage,
after
Wave 3
Mean

SE

Focal Outcome Variable
Relationship dissolution by wave 4

.35

.61

.48

0–1

Relationship Characteristics
Low marital love
Economic strain
Relationship violence
Extramarital sex

−.97

(.01)

−.77

(.10)

−.92

(.05)

−1.03–1.03

.15

(.02)

.64

(.14)

.51

(.10)

−.29–2.65

−.02

(.01)

.41

(.13)

.19

(.08)

−.36–1.97

.21

.48

.29

0–1

.33

.73

.70

0–1

.11

.00

.10

0–1

.18

.36

.24

0–1

.07

.03

.06

0–1

Co-resident child

.57

.55

.79

0–1

Non-co-resident child

.05

.15

.15

0–1

Educational attainment

2.92

Demographic Characteristics
Male

NIH-PA Author Manuscript

a

Black

a

Hispanic

Other non-White race/ethnicity

a

Employed full-time

(.01)

2.73

.59

(.08)

.61

2.77

(.06)

.54

1–5
0–1

Behavioral Characteristics
Religiosity

.03

(.02)

−.19

(.14)

Hard drug use

.05

Problem drinking

.18

Problem gambling

.01

.06

.07

0–1

Non-family violence

.04

.21

.21

0–1

Prior domestic violence conviction

.00

.00

.02

0–1

Incarcerated before marriage

.04

.14

.30

0–1

.12
(.02)

.64

−.15

(.08)

.25
(.20)

.62

−1.66–2.46
0–1

(.16)

0–6

NIH-PA Author Manuscript

Relationship History
Age at marriage
Years married at wave 3
Cohabited with spouse before
marriage
n

21.06

(.04)

19.91

(.31)

20.81

(.24)

17–27

1.69

(.04)

2.52

(.25)

1.91

(.22)

0–5

.59

.76

.76

1,752

33

62

0–1

NOTE: Variables measured at wave 3 unless otherwise noted.
ABBREVIATION: SE = standard error (omitted for dichotomous variables).
Source: National Longitudinal Study of Adolescent Health.
a

Measured at wave 1.

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 27

Table 2

NIH-PA Author Manuscript

Discrete Time Survival Model Predicting Duration-Dependent Risk of Relationship Dissolution From
Respondent Incarceration During Marriage (N = 14,644 observations on 1,919 respondents)
b

SE

OR

.70***

(.18)

2.02

Male

−.35***

(.10)

.71

Black

.61***

(.12)

1.85

Predictors
Focal Predictor
Incarcerated during marriage
Demographic Characteristics

Hispanic

.06

(.11)

1.06

Other non-White race/ethnicity

−.11

(.17)

.90

Co-resident child

−.12

(.09)

1.13

Non-co-resident child

.43*

(.15)

1.54

Educational attainment

−.11

(.11)

.89

.07

(.09)

1.07

−.20**

(.06)

.82

.33*

(.17)

1.39

Problem drinking

.04

(.05)

1.04

Problem gambling

.18

(.33)

1.20

Employed full-time

NIH-PA Author Manuscript

Behavioral Characteristics
Religiosity
Hard drug use

Non-family violence

−.05

(.19)

.95

Prior domestic violence conviction

.42

(.59)

1.53

Incarcerated before marriage

.19

(.18)

1.21

Age at marriage

−.10***

(.03)

.90

Years married at wave 3

−.19***

(.04)

.82

.61***

(.10)

1.83

One year since marriage

−.47*

(.19)

Two years since marriage

−.27

(.18)

Three years since marriage

.13

(.17)

Four years since marriage

.42**

(.16)

Five years since marriage

.41*

(.16)

Six years since marriage

.30†

(.17)

Seven years since marriage

.21

(.19)

Eight years since marriage

.80***

(.19)

Nine years since marriage

.54*

(.25)

Ten years since marriage

.83**

(.29)

.26

(.53)

Relationship History

Cohabited with spouse before marriage
Time

NIH-PA Author Manuscript

Eleven years since marriage

Criminology. Author manuscript; available in PMC 2015 August 01.

Siennick et al.

Page 28

b

Predictors

SE

NIH-PA Author Manuscript

Twelve years since marriage

−.36

(1.03)

Thirteen or more years since marriage

1.67

(1.11)

Constant

−.95

(.68)

OR

NOTE: Reference category for time variables was married this year.
ABBREVIATIONS: OR = odds ratio; SE = standard error.
Source: National Longitudinal Study of Adolescent Health.
†

p < .10;

*

p < .05;

**

p < .01;

***

p < .001 (two-tailed).

NIH-PA Author Manuscript
NIH-PA Author Manuscript
Criminology. Author manuscript; available in PMC 2015 August 01.

NIH-PA Author Manuscript

NIH-PA Author Manuscript
(.02)
(.02)

.00
.00
.04†
−.07***

Other non-White race/ethnicity

Co-resident child

Non-co-resident child

Educational attainment

Criminology. Author manuscript; available in PMC 2015 August 01.
.01
.10
−.01

Non-family violence

Prior domestic violence conviction

Incarcerated before marriage

(.00)
(.00)

−.00
.01*

Years married

(.03)

(.12)

(.03)

Age at marriage

Relationship History

.04

Problem gambling

(.05)

(.01)

.04***

Problem drinking

(.01)
(.03)

−.02**

(.01)

(.02)

.02

Hard drug use

Religiosity

Behavioral Characteristics

.00

(.01)

−.00

Hispanic

Employed full-time

(.02)

.11***

Black
(.01)

(.01)

(.04)

(.04)

−.03*

c

.02

.17***

.01

−.01

(.01)

(.01)

(.07)

(.34)

1.16***
.12

(.08)

(.14)

−.11

−.05

(.02)

(.07)

.22**
.09***

(.02)

(.03)

−.08*

.03

(.05)

−.22***

(.07)

(.03)

.18***
.05

(.06)

(.06)

(.06)

(.03)

(.09)

(.12)

SE

.08

.07

.12†

−.14***

.26**

.48***

b

b
SE

Economic
a
Strain

Low Marital
a
Love

Male

Demographic Characteristics

Incarcerated during marriage, after wave 3

Incarcerated during marriage, by wave 3

Focal Predictors

Predictors

(.05)

.10*

.01

−.03**

.01

.99***

.01

−.01

.11***

.10†

−.04*

−.01

−.04

.11*

(.01)

(.01)

(.06)

(.24)

(.06)

(.11)

(.02)

(.06)

(.02)

(.03)

(.04)

(.05)

(.03)

(.03)

.10**

.01

(.04)

(.03)

(.07)

(.09)

SE

.23***

−.11***

c

.10

.35***

b

Relationship
a
Violence

.09†

−.10*

.34

2.07†

.20

−.05

.28***

.77**

−.19*

−.10

.05

.27

.30*

.21

.25

1.02***

−.16

c

−.17

.83*

b

Extramarital
b
Sex

(.05)

(.04)

(.28)

(1.15)

(.27)

(.51)

(.08)

(.24)

(.09)

(.13)

(.17)

(.24)

(.13)

(.22)

(.15)

(.18)

(.14)

(.33)

(.38)

SE

Linear and Logistic Regression Coefficients Predicting Relationship Characteristics at Wave 3 From Respondent Incarceration (N = 1,847)

NIH-PA Author Manuscript

Table 3
Siennick et al.
Page 29

NIH-PA Author Manuscript
(.09)

.91**
(.32)

.15***

−.71***

SE
(.03)

b

SE
(.01)

b
−.02
.54**

.07*

b

Relationship
a
Violence

Logistic coefficients shown.

p < .001 (two-tailed).

p < .01;

***

**

p < .05;

*

p < .10;

p < .05 for difference from coefficient for incarcerated during marriage, by wave 3.

†

c

b

Linear coefficients shown.

a

Source: National Longitudinal Study of Adolescent Health

ABBREVIATION: SE = standard error.

NOTE: Model predicting economic strain included fixed effects for primary sampling unit (omitted from table).

Constant

Cohabited with spouse before marriage

Economic
a
Strain

(.21)

(.03)

SE

NIH-PA Author Manuscript
Low Marital
a
Love

−.01

.26†

b

Extramarital
b
Sex

(.98)

(.13)

SE

NIH-PA Author Manuscript

Predictors

Siennick et al.
Page 30

Criminology. Author manuscript; available in PMC 2015 August 01.

NIH-PA Author Manuscript
Table 4

NIH-PA Author Manuscript

NIH-PA Author Manuscript

Incarcerated before marriage

(.08)
(.09)

.26**
.63***

Relationship violence

Extramarital sex

n.s.

n.s.

***

KHB Test for
Change in
Coefficient

Criminology. Author manuscript; available in PMC 2015 August 01.

p < .001 (two-tailed).

***

p < .01;

**

p < .05;

p < .10;

*

†

Source: National Longitudinal Study of Adolescent Health.

ABBREVIATIONS: KHB = Karlson, Holm, and Breen; n.s. = not significant; SE = standard error.

NOTES: Control variables were respondent male gender, race and ethnicity, co-resident child, non-co-resident child, educational attainment, full-time employment, religiosity, hard drug use, problem
drinking, problem gambling, non-family violence, prior domestic violence conviction, age at marriage, years married at wave 3, whether the couple cohabited before marriage, and dummy variables for time
since marriage (omitted from table).

(.06)

.13*

Economic strain

(.18)

(.24)

(.26)

(.14)

.15

.72**

.41

.88***

(.18)

(.24)

(.25)

Low marital love

Relationship Characteristics

.19

.71**

Incarcerated during marriage, after wave 3

Prior Incarceration

.68**

SE

Model 2
b

SE

Model 1
b

Incarcerated during marriage, by wave 3

Focal Predictors

Predictors

Discrete Time Survival Models Predicting Duration-Dependent Risk of Relationship Dissolution From Respondent Incarceration and
Relationship Characteristics (N = 14,644 observations on 1,919 respondents)

Siennick et al.
Page 31