Abstract
Objectives:
This study tests theorized mechanisms of desistance, and whether the process of desistance is conditioned by social structural position.
Methods:
We investigate how marriage promotes desistance from crime among urban African American males raised in the Woodlawn community, a disadvantaged neighborhood in Chicago. Using hierarchical linear modeling, we test the resiliency of the marriage effect by observing offending trajectories following marital dissolution; is the marriage effect conditional upon staying married, indicating situational effects? or does the effect persist when marriage is taken away, indicating enduring effects? Further, we test if the process of desistance is conditional upon contextual disadvantage.
Results:
While initial findings show an increase in violent and property offending upon divorce, further analysis shows evidence that this effect differs by neighborhood structural context; the increase in offending upon divorce is apparent only for African American men who experience continued disadvantage across the life course. Those who moved to relatively more advantaged areas by adulthood show no increase in offending upon marital dissolution.
Conclusions:
How marriage matters for desistance is partially influenced by social structural position; context matters. These findings invigorate criminological research on the mechanisms driving the marriage effect and provide insight into the interactive nature of person and context.
Much has been written about the transformative potential of marriage for fostering desistance from crime; similar to the general health literature (Waite and Gallagher 2000), the majority of empirical evidence points to a beneficial effect of marriage such that involvement in crime diminishes alongside the transition to and state of being married (Bersani and Doherty 2013; Craig, Diamond, and Piquero 2014; Laub and Sampson 2001; Skardhamar et al. 2015). Accumulating research testing the relevance of the marriage effect—if marriage is associated with desistance—across diverse samples suggests that marriage is a robust correlate of desistance across racial, ethnic, and gender groups (see Bersani and Doherty 2013). While the evidence regarding race is not universally consistent, marriage is generally associated with reduced states of offending across race. Piquero and colleagues (2002) found benefits of marriage for desistance among a sample of parolees released nearly 40 years ago, yet the benefits for Black parolees were contingent on the type of crime being studied; variation by crime type was not found among the White parolees. More recently, although there are exceptions (e.g., Craig 2015), research has found a marriage–desistance relationship for African Americans, males in particular; Doherty and Ensminger (2013) report a strong and consistent effect of marriage on desistance among African Americans and Bersani and DiPietro (2014) find a stronger effect of marriage on criminal offending for minority men (African Americans and Hispanics) compared to White men.
While research on understanding desistance among racially diverse samples has informed the field, the sole focus has been on if event transitions result in changes in criminal behavior for racially and ethnically diverse samples. Somewhat lost in the empirical literature has been a focus on how marriage matters, and whether the way in which marriage matters differs across groups from varying social structural positions (e.g., gender, race, and class). 1 Specifically, less attention has been aimed at understanding why salient life events often occur alongside behavioral changes in the life course, and whether the way in which life events matter differs across demographically diverse samples.
Recent work by Bersani and Doherty (2013) suggests that the mechanisms of change fostering desistance lie on a continuum with enduring mechanisms at one end of the continuum and situational mechanisms at the other end. Enduring effects focus on individual changes (i.e., bonding, identity), whereas situational effects focus on contextual or structural changes (i.e., criminogenic networks, supervision, and daily routines). While Bersani and Doherty’s (2013) work was instrumental in shifting the focus from if marriage matters to how marriage matters, the sample for the study focused largely on individuals who were White (81 percent) with few defined as disadvantaged (7 percent) and was limited in the extent to which it captured high-rate offenders. As a result, the findings that lend support to situational mechanisms of desistance potentially mask differences in the desistance process by social structural position. Laub and Sampson (2001:55) hypothesize that the mechanisms of desistance should be more general than specific, but they leave open the potential for important variation in the process of desistance across key subgroups in the population dictated by social structural position.
To elucidate the possible variation in the process of change associated with marriage by social position, we utilize the strategy introduced by Bersani and Doherty (2013) of testing the durability of the marriage effect by examining changes (or the lack thereof) in offending trajectories upon the dissolution of marriage. By considering whether someone’s offending increases upon the removal of marriage (i.e., marital dissolution) or whether the reduction in offending endures even when the marriage is discontinued, we gain insight into whether the process of change is situational or enduring in nature. We then use this strategy to examine whether the process of change is contingent upon social structural position.
Theorizing Desistance from Crime
A variety of theoretical explanations have been offered to explain how individuals desist from offending in the presence of life events, such as marriage. While research into the factors that facilitate or hinder the process of desistance from crime has a long history and highlights the importance of social bonds to conventional life (see, e.g., Glaser 1969; Irwin 1970; Meisenhelder 1977; Reitzes 1955), Sampson and Laub’s (1993) age-graded theory of informal social control was the first to theorize explicitly the connection between social bonds and desistance. Using extensive information from in-depth qualitative interviews with White males, Laub and Sampson (2003) extend their original theory and conclude that among those identified as “desisters,” salient life events, most notably marriage, functioned to facilitate this process through several mechanisms.
Although social bonds take center stage for Laub and Sampson (2003), others have placed greater weight on internal changes to identity or external constraints on opportunities to offend. The collective body of literature theorizing pathways to desistance suggests that marriage affects desistance by fostering social bonds to and investments in conventional society, introducing a source of direct social control and monitoring, changing routine activities, providing the impetus for residential change, and aligning with a change in one’s sense of self and identity (for a review, see Bersani and Doherty 2013). Although the various theoretical accounts of the marriage–desistance link correspondingly posit a beneficial effect of marriage, each differs in terms of the role (e.g., mechanism) of marriage in maintaining diminished criminal involvement. Notably, the extent to which these mechanisms are shaped by one’s social position remains understudied. The prevailing focus in the empirical literature on person-level constructs leaves unknown not only the extent to which individuals are shaped by their situational contexts but also how the influence of life experiences—like marriage—may be altered by contextual constraints and opportunities (see, e.g., Farrall and Bowling 1999; Sampson and Laub 1993).
Social Structural Position and the Mechanisms of Desistance
Despite a lengthy literature documenting the race-related challenges faced by African Americans that impact trends in marriage, two patterns relevant to desistance are evident. First, like their racial and ethnic counterparts, marriage is associated with a host of benefits, at least among African American men, including the movement away from crime (Bersani and DiPietro 2014; Doherty and Ensminger 2013; Mudar, Kearns, and Leonard 2002; Nielsen 1999; Piquero, MacDonald, and Parker 2002). Second, for the majority of African American young adults, traditional middle-class norms of marriage and family formation remain important goals (Edin and Nelson 2013; Wood, Avellar, and Goesling 2008), although perhaps less firmly entrenched by African American men than women (Anderson 1999; South 1993). These “positive attitudes toward marriage, coupled with a decrease in the number of African American marriages, suggests that there may be something unique as to how African Americans define the institution of marriage” (Curran, Utley, and Muraco 2010:347). If the meaning of marriage is rooted in culture (i.e., norms and expectations), African American marriages may mean something different from marriages among other groups (Orbuch and Fine 2003). The extent to which structural constraints influence the availability of marriage (Wilson 1987) and the portrayal of marital roles may also differentially impact African American men’s interpretation of marriage. In short, if there are differences in the ways of “doing marriage” based on social structural position, then perhaps there are differences in the processes of “doing desistance” as well.
Enduring mechanisms
The notion of endurance suggests a lasting change that is resistant to weakening in the face of challenge. Two core explanations of the desistant potential of marriage infer an enduring mechanism of change: the (re)building of attachments and identity transformation. First, marriage may initiate (re)investment in conventional society through the marital bond (Sampson and Laub 1993). In addition to signaling a responsibility for another person, marriage links a couple to the larger community (Calverley 2013) strengthening attachments to conventional institutions and changing the way others interact and view the paired unit (Waite and Gallagher 2000). Involvement in deviant, criminal, antisocial behavior declines as bonds to conventional society build and strengthen.
In line with this mechanism, the extant literature largely finds that, among married African Americans, marriage connotes a lifelong commitment involving love, trust, and partnerships (Curran et al. 2010; Hurt 2013). Although the family stress literature would argue that the financial and social barriers disproportionately faced by African Americans would lead to marital hardship, qualitative findings indicate that “sharing life’s journey together, overcoming challenges together, and benefitting from a couple identity” strengthen one’s commitment to the marriage (Hurt 2013:876; see also Orbuch, Veroff, and Holmberg 1993). Marks and colleagues (2008) find that African American couples face a number of challenges including the burden of managing both work and family and providing care for both older and younger generations simultaneously, including nonbiological children; in turn, these challenges are successfully overcome by relying on the committed spouse. Therefore, although African Americans are less likely to marry, the relatively small percentage who do marry evidence a strong attachment and a couple identity. These patterns suggest that enduring mechanisms may be more salient in understanding desistance among African Americans. This prospect of creating an enduring bond with conventional society, however, may depend on marriage acting as a segue to a multitude of prosocial relationships beyond the marriage itself.
Second, marriage may provide offenders with an opportunity for an identity transformation (Giordano, Cernkovich, and Rudolph 2002; Maruna 2001) such that the offender identity becomes incongruent with the married identity. Marriage may confer upon males a core marker of “manhood” or masculine identity (Nock 1998) signaling that he is ready to settle down and commit to meeting societal expectations about what it means to be a husband (Edin and Nelson 2013). While marriage may hold the potential to foster a change in identity from a “hell-raiser to a family man” (Hill 1971; Laub and Sampson 2003:147), this transition may not be sufficient to overcome an African American male identity, laden with the history of discrimination and racism (Calverley 2013). Being African American could be the master identity rendering the married versus not married identity less salient in influencing behavior, including desistance. This diminished capacity of identity change may be particularly salient among African American men living in disadvantage, where the “criminalblackman” identity is particularly pervasive (Pager 2007). Therefore, if enduring mechanisms are most salient in the process of change via marriage, offending patterns should be unchanged or continue to decline upon marital dissolution.
Situational mechanisms
Although enduring explanations of change emphasize within-individual changes that occur alongside salient life course transitions such as marriage, the transition to marriage may instead be beneficial because it alters opportunity structures. Situational explanations of the marriage–desistance link suggest that marriage may function to reduce one’s opportunities to offend by changing one’s daily routines (Osgood et al. 1996), knifing-off offenders from criminogenic peers and places (Kirk 2012; Warr 1998), and increasing direct supervision of behavior (Gottfredson 2005). Rather than changing one’s propensity to offend, situational mechanisms insinuate that marriage restricts opportunities to become involved in criminal behavior. If situational mechanisms are driving the observed marriage effect, we would expect offending to resume, or increase, upon the dissolution of marriage.
The extent to which marriage can evoke situational change may be contingent upon the availability of resources as well as prior exposure to norms surrounding the social obligations of husband and wife roles. Prior research aimed at understanding racial differences in marriage trends emphasizes variation in socialization practices and lack of exposure to married role models, particularly among disadvantaged African American men, that may render situational mechanisms less relevant to the desistance process. Stable African American marriages have been found to be more collaborative in nature as opposed to cooperative (as is found with White marriages) indicating the relative importance of maintaining an independent self among African American couples (Brown, Orbuch, and Maharaj 2010; Orbuch et al. 1993). Peer groups are more highly regarded than marriage for urban African American men, in particular (Anderson 1999), suggesting that African American men might be less likely to alter their routine activities upon marriage with respect to spending time with friends (see also Calverley 2013; South 1993). The higher rates of offending and incarceration among African American males living in disadvantaged areas, in particular, have reached the status of an expected stage of the life course (Doherty et al. 2015; Pettit and Western 2004). The prevalence of offending and criminal justice interactions may reduce the stigmatization of offending and interacting with other criminals, supporting a dual legitimate and illegitimate lifestyle. In this sense, a “hell-raiser” identity may not be incompatible with a “family man” identity.
In sum, with respect to situational mechanisms, research revealing differences in the meaning and roles of marriage among African American men may render situational mechanisms less prominent in understanding the desistance process. For African American men, situations may be less likely to change upon marriage, which would suggest that enduring mechanisms may hold greater salience in explaining the relationship between marriage and desistance from offending for these men. However, it may be more complicated than being enduring or situational as one’s multidimensional social structural position (i.e., neighborhood context among African American men) may differentially impact the marriage process.
Present Study
The current study advances knowledge on how marriage matters and provides insight into the interactive nature of person and context by investigating whether structural disadvantage in adulthood moderates the association between relationship status and offending among a community cohort of urban African American men raised in the disadvantaged neighborhood of Woodlawn (an inner-city community in the South Side of Chicago). 2 We utilize Bersani and Doherty’s (2013) process of change framework to examine the resiliency of the marriage effect on offending across varying social structural positions; we focus attention on the durability of change implied by each end of the continuum by examining the effect of marital dissolution on offending trajectories. If enduring mechanisms are driving the process of desistance, then marital dissolution should not impact offending; bonds and family man identities are resistant to change and take time to dissolve. If situational mechanisms are most salient in the marriage process, we would expect offending to increase upon dissolution of the marriage as a major constraint on opportunities to offend would be removed.
Our overriding question is: Do the differences in the meaning of marriage found in prior research translate into differences in the mechanisms of desistance among African American males? In other words, do the ways of doing marriage have implications for the processes of doing desistance? To address this question, we hold stable two dimensions of one’s social position—race and gender—and add variation on a third dimension of one’s social position (i.e., social class) as characterized by neighborhood context. We then examine whether the process of change underpinning desistance from crime for these men is contingent upon variation in contextual (dis)advantage experienced over the life course. Thus, we investigate whether neighborhood context moderates the relationship between marriage and offending among African American men.
Methods
Data
The study population represents 606 males from a community cohort of 1,242 African American male and female first graders in 1966 to 1967 from Woodlawn, IL (only 13 families declined to participate). At the inception of the study in 1966, although there was individual-level economic heterogeneity among the Woodlawn cohort families (due to racial segregation at the time), the Woodlawn neighborhood was a socially disadvantaged one. A comparison of the 1970 census data for Woodlawn and Chicago highlights the disadvantaged nature of the Woodlawn community; for instance, during the study initiation period, Woodlawn was characterized by a higher percentage of (1) African Americans (96 percent vs. 33 percent), (2) families with children younger than 18 living in poverty (32 percent vs. 15 percent), (3) female-headed households with children younger than 18 (46 percent vs. 21 percent), and (4) vacant houses (15 percent vs. 6 percent) than Chicago. Woodlawn was the fifth poorest community in Chicago and had the highest rate of male juvenile delinquents (33.5 per 100 males between the ages of 12 and 16; Council for Community Services 1975).
To date, Woodlawn research team members have followed the cohort three additional times: in adolescence, when they were 16-17 using a group-based self-reported assessment (1976 to 1977); in young adulthood, when the cohort was aged 32-33 (1992 to 1993); and in middle adulthood, when the cohort was aged 42-43 (2002 to 2003) using personal structured interviews. In addition to the self-report information from these four time points, annual criminal history data and mortality data have been collected over the years by Woodlawn researchers. The criminal history information is drawn from the Chicago Police Department and the Federal Bureau of Investigation obtained in 1993, which spanned the age of majority (age 17 in Illinois) to 32, and updated criminal records from the Illinois State Police and the Illinois Criminal Justice Information Authority extending these histories to age 52 (the criminal histories from ages 17 to 42 are used in these analyses to map the interview data). 3 Mortality is assessed through searches of the National Death Index as well as by corroborating reports of deaths from families and friends (n = 132 dead as of 2009, 11 percent of the original cohort).
Measures
Criminal history
We focus on violent offending and property offending separately, defined as the number of official arrest counts for each crime type at each age (17 to 42) for each individual. Examples of violent offending include crimes such as homicide, assault, rape, and robbery; examples of property offenses include crimes such as burglary, larceny, auto theft, fraud, and criminal damage. In lieu of actual days incarcerated, sentencing data from the arrest records were incorporated into the criminal histories such that a person is considered incarcerated in any year where he has zero offenses and is known to have been sentenced to more than one year in prison at that age. 4 Using this strategy, we attempt to reduce the chances of presuming someone has desisted from offending who in fact was incarcerated (see Eggleston, Laub, and Sampson 2004; Piquero et al. 2001). Finally, mortality information was also integrated into the longitudinal criminal histories to safeguard against presuming someone had stopped offending who had instead died (for more information on the Woodlawn criminal history information, see Doherty and Ensminger 2014).
Marital history
We draw upon the self-report interview data from the young adult (age 32) and midlife interviews (age 42) to compile annualized information of marital status at each age. At the young adult and midlife interviews, each person was asked about his current marital status, the number of times he had been married, and the age of marriage for his current marriage. If the person was married more than once, the respondent was also asked the age his first marriage began, the age his first marriage ended, and how that marriage ended. This variable was dummy coded such that a person was coded as single (i.e., not yet married) versus other, married versus other, or divorced/separated versus other in any given year. Thus, individuals can only assume one relationship status in any given year; for instance, if an individual is coded a 1 on the marriage variable, they are coded 0 on the single variable and 0 on the divorced/separated variable. We combine divorced and separated for this study because the majority of the men with a marital dissolution (n = 153) reported divorce only (66.7 percent) with an additional 9.2 percent reporting both separation and divorce; approximately one-quarter reported a separation only (24.2 percent). Finally, once married, we restricted the model to include only one additional change in marital status (marriage to divorce/separation), so that we could focus on the impact of marital dissolution. Thus, once a person is widowed (n = 5) or remarries (n = 50), the criminal and marital histories for that person are removed beginning at the age of remarriage. Of the 50 men who remarry, 50 percent remarry at age 35 or older (mean = 34.44, SD = 5.19, and range = 23 to 42).
Context
To take advantage of the study design where all of the men were raised in a disadvantaged neighborhood, we tap into the continuity of living in a disadvantaged context as opposed to moving to a more advantaged area by young adulthood. Thus, holding early childhood context constant, we operationalize this continuity or discontinuity in context by measuring neighborhood disadvantage in young adulthood. We use an index developed by Ross and Mirowsky (2001), which assigns an objective neighborhood disadvantage score to each respondent using U.S. census data (see also Lo 2010). This neighborhood disadvantage index (NDI) is calculated by adding the (1) percentage of households with incomes below the federal poverty threshold and the (2) percentage of female-headed households with children and subtracting the (3) percentage of adults over the age of 24 with a college degree and the (4) percentage of housing units that are owner-occupied. 5 The sum of the four indicators is then divided by four. The index ranges from −3.35 to 4.55, where lower scores indicate more advantaged neighborhoods (i.e., many adults graduated from college and own their homes; few households are poor or female-headed) and higher scores indicate more disadvantaged neighborhoods (i.e., few adults have a college degree, many rent rather than own their homes; many households are poor and female-headed). For ease of interpretation, this continuous measure is dichotomized such that negative values indicate advantaged neighborhoods (coded “0”) and positive values indicate disadvantaged neighborhoods (coded “1”).
Criminal propensity
To control for criminal propensity, we include a measure, which is a mean scale of eight dichotomous risk factors that constitute three major domains within respondents’ early lives—individual, family, and school. The combined measure was created by taking the mean of the eight dichotomous risk variables when at least six of the eight responses were valid (nonmissing) 6 for each respondent. These eight individual measures are listed in the Appendix. The final continuous measure ranges from 0 to 1 with a mean of 0.53 (SD = 0.24).
Final Sample Size
Of the 606 males in the original cohort, 8 had died before the age of 17, 9 had missing offending information due to discrepancies in the criminal record matching process, and another 95 had missing marital history information. Thus, 494 males had information on arrest history and marriage (81.5 percent of the original male cohort). The neighborhood context measure is drawn from the 1990 census data based on place of residence at age 32. Analyses that include this measure use a reduced sample size of 443 men (73.1 percent of the original male cohort).
Analysis
The multivariate analysis begins with a test of whether divorce 7 impacts offending, controlling for criminal propensity. Specifically, we employ hierarchical linear modeling (HLM), which allows a simultaneous estimate of variations in crime within individuals over time and between-individual differences in offending patterns. To study individual change within the HLM framework, the crime counts for each individual by age are viewed as nested within that individual. There are two levels of analysis. Level 1 estimates each person’s development with a unique individual growth trajectory that depends on a set of parameters. The growth parameters from level 1 become the outcome variables in level 2 of the model and are determined by person-level characteristics. Therefore, level 1 in the model estimates the within-individual change, and level 2 in the model represents the between-individual analysis (Raudenbush and Bryk 2002).
At the within-individual level, crime is modeled as a function of age and relationship status. Specifically, a hierarchical overdispersed Poisson model is used for this count data to accommodate for the abundance of zeros at each age (Raudenbush et al. 2004). Thus, the outcome variable is the rate of violent or property offending per year. The basic elements of this within-person model are:
where i is the index for individuals and t stands for the longitudinal observations. The intercept, π0i , is the estimated rate of offending, when age is set to zero. The π1i and π2i parameters estimate the average rate of change and rate of acceleration or deceleration, respectively (Raudenbush and Bryk 2002:163). To increase the meaningfulness of the interpretation of the intercept term, researchers use the technique of centering their data. Centering at the mean provides a more meaningful anchor to better understand the variation around the mean. Here, we center age at the mean of the observed age–person distribution (age 29.5). The π3i and π4i parameters indicate the change in offending due to being not yet married (i.e., single) compared to when married, and divorced compared to when married, respectively. The single and divorce parameters at level 1 have been group mean centered to allow an examination of the deviation in the overall mean level of divorce and its relationship to offending. This allows for an examination of change in offending due to divorce within individuals.
At the between-individual level, the parameters from level 1 are estimated from the level 2 equations. 8 We estimate three models for violent and property crime separately, with all models utilizing the level 1 equation (1) described above. The basic elements of the between-individual model for the first of three models are:
For the second set of models, we add a main effect for the NDI to the level 2 intercept equation (2) as well as a cross-level interaction to the single slope (equation 5; model 2) and then, separately, to the divorce slope (equation 6; model 3). These models allow neighborhood context to moderate the relationship between single (i.e., not yet married as opposed to married) and offending as well as divorced (as opposed to married) and offending.
Results
Descriptives
Close to half of the 494 Woodlawn men in the sample were arrested for at least one violent offense (47.2 percent) and/or a property offense (48.8 percent). Among those with a violent offense, the mean number of violent offenses was 3.85 (SD = 4.26, range = 1 to 26); among those with a property offense, the mean number of offense counts was 5.11 (SD = 6.93, range = 1 to 51). With respect to marriage, Table 1 shows that a majority of the men were married at some point between ages 17 and 42 (n = 284, 57.4 percent) with 53.9 percent of those who married eventually divorcing or separating by age 42. Moreover, those who remained married tended to marry at an older age (28.53 years, on average) than those who divorced or separated (24.48 years on average).
Marriage and Marital Dissolution Descriptives for Ever-married Males.
Note: SD = standard deviation; div. = divorced; sep. = separated.
aAmong those who divorced or separated.
Despite all men growing up in the same disadvantaged neighborhood of Woodlawn, close to half of the males lived in a more advantaged neighborhood context at age 32 (48.0 percent of the 443 males with neighborhood context data). To be clear, the terms advantaged and disadvantaged contexts in this study are relative terms. For instance, neighborhoods categorized as disadvantaged based on the NDI are largely Black (91 percent, on average) with a high percentage of female-headed households and households living in poverty (56 percent and 35 percent, on average, respectively). While neighborhoods characterized as advantaged in this study cannot be considered affluent, they are relatively more advantaged. For instance, these neighborhoods have, on average, fewer female-headed households and fewer residents living in poverty (25 percent and 10 percent, respectively). Moreover, 60 percent of residents in the advantaged neighborhoods own their home (compared to 25 percent in the disadvantaged neighborhoods), and 21 percent are college educated (compared to 10 percent in disadvantaged neighborhoods). At the individual level, the relative advantage for those living in advantaged neighborhoods is also evident with the percentage of men living below the poverty line changing from 45 percent in childhood to 27 percent in young adulthood compared to a corresponding change from 62 percent to 49 percent for those living in disadvantaged neighborhoods in young adulthood.
HLM Results
We begin by investigating the effect of the dissolution of marriage on offending for the full sample of African American males before considering the additional dimension of one’s social structural position—neighborhood structural context. Table 2 presents the HLM results for the effect of being single (i.e., not yet married) and being divorced/separated, compared to when married, on offending over time. Models 1a and 2a, for violent crime and property crime, respectively, indicate a significant and positive effect of both being single and divorced on offending. These coefficients translate into a 39 percent increase in violence ((exp.326) − 1) × 100) and a 62 percent increase in property offending ((exp.483) − 1) × 100) when a man is single compared to when married, and a 34 percent increase in violence ((exp.293) − 1)*100) and a 44 percent increase in property offending ((exp.362) − 1)*100) when a man is divorced compared to when married. These findings of an increase in offending with divorce are consistent with past literature, which has found situational mechanisms underpinning the desistance process. 9
Hierarchical Nonlinear Models of the Effect of Relationship Status on the Rate of Offending.
Note: Coeff. = coefficient; sig. = significance; SE = standard error. We report the population average with robust standard error results.
*p < .05.
**p < .01.
***p < .001.
While divorce increases offending for both violent and property crime, the question of contextual influences (i.e., neighborhood disadvantage) on the marriage effect remains. In other words, are our findings, which support a process of change driven by situational mechanisms, relevant for all African American males or is the process by which African American men’s offending changes alongside marriage dependent upon contextual constraints (or benefits)? To tap into this potential source of variation, we examine whether the marriage effect endures or dissipates with divorce varying one’s level of neighborhood disadvantage in young adulthood.
Models 1b and 2b show a lack of significance in the cross-level interaction effect (disadvantage × single) indicating that the effect of being single (i.e., not yet married) compared to when one is married on offending is similar for those who have persistently lived in disadvantaged areas and those who have moved to more advantaged neighborhoods by young adulthood. This suggests a generality of the beneficial marriage effect for the African American males in this sample, regardless of structural context. Models 1c and 2c of Table 2 present the cross-level interaction (disadvantage × divorce) analyses for divorce and offending for violent and property arrests, respectively. In these models, both of the interaction coefficients are significant, indicating that marital dissolution has differential effects on offending for those who continue to live in disadvantaged areas and those who live in more advantaged contexts by young adulthood.
A visual depiction of the percent change in the rate of offending with each of the relationship transitions by crime type reported in Table 2 (Models 1b, 1c, 2b, and 2c) is plotted in Figure 1. The zero line represents the rate of offending when married, and the bars represent the percent change in the rate of offending for the transition from single (i.e., not yet married) to married and for the transition from married to divorced/separated. First, the increase in the rate of offending for the single state across crime type shows the benefits of marriage regardless of one’s level of neighborhood disadvantage in young adulthood. Looking next at the transition due to marital dissolution across neighborhood context, there is a clear divergent pattern. For African American men who persistently live in areas of disadvantage rates of offending increase upon marital dissolution. Thus, the benefits of marriage appear to be contingent upon staying married for these men—a finding that supports situational factors driving the desistance process. In contrast, for men who move to relatively more advantaged neighborhoods by young adulthood, rates of offending continue to decrease even after the dissolution of marriage. For these men, the process of desistance seems to be a more enduring one.

Percentage change in the rate of offending when single and divorced compared to when married by young adult social context.
Discussion
Using data from a sample of African American men whose early life course was shaped by residence in the same socially disadvantaged Woodlawn neighborhood of Chicago in the mid-1960s, we find two notable findings. First, although we focus attention on the dissolution of marriage, it is important to note that the men in this sample were significantly less criminally active when married compared to when not yet married (i.e., before marriage) and when divorced (i.e., after marriage), which is consistent with previous research (see Bersani and Doherty 2013; Craig et al. 2014). Second, we extend these findings by examining the extent to which social structural position (i.e., neighborhood disadvantage for African American men) influences this relationship. In short, the results reveal that context matters and influences how marriage facilitates the process of desisting from crime.
Though variation in residential mobility in this sample was evident, many men (52 percent) experienced continued exposure to residential disadvantage. The stickiness of place in this sample aligns with Sharkey’s (2013) research documenting the difficulties of moving out of disadvantage for African American men and the detrimental consequences this persistence holds for the intergenerational transmission of inequality. Along this contextual divide, we find evidence of a divergent marriage effect: Among men with a history of disadvantage from childhood into adulthood, we find that divorce is associated with an increase in offending. This suggests that for persistently disadvantaged married African American males, the observed benefits of marriage are situational in nature and appear to dissolve alongside the dissolution of marriage.
Although all the men in this sample spent their childhood in the same disadvantaged neighborhood, a sizable group (48 percent) was able to find a pathway out of poverty and by age 32 were living in noticeably more advantaged, though far from affluent, neighborhoods. Among married men whose life course is characterized by the accumulation of advantage and relative upward mobility, the impact of divorce appears to be negligible at least in terms of offending behavior. For the more advantaged adult African American men in this sample, the beneficial effects of marriage appear to be more enduring in nature surviving beyond the termination of marriage. 10
Exploring Life Course Divergence
Clearly, marriage is not a panacea for offending behavior; while married individuals have a reduced rate of offending compared to when single, the resiliency of this benefit is vulnerable to contextual factors. Men who experienced relative improvements in their level of advantage in adulthood compared to childhood showed a continued decline in offending over the life course. This relative nature of the advantage demonstrates that advancement in advantage need not be dramatic to offer a context ripe for lasting or enduring change and desistance from crime.
We explore further the pattern of life course divergence in the sample by investigating differences across the subsamples of persistently disadvantaged and relatively advantaged men to better understand whether those experiencing enduring effects of marriage were characterized by greater exposure to, or higher dosages of, conventional life events. Looking across a range of key life course domains we find that divorced men living in a more advantaged context were significantly more likely to report having children in the household (disadvantaged divorced men had a higher prevalence of nonresidential children), more likely to be currently employed (although the frequency of unemployment stints did not significantly differ across groups), significantly more likely to have posthigh school education, have higher total income levels, and report involvement in a greater number of community organizations. Thus, the enduring marriage effect observed for relatively more advantaged men in this sample may be functioning as a proxy indicator of the accumulation of advantage across the life course. The fact that the advantaged divorced men experienced more conventional life events may account for the resilient or enduring nature of the marriage effect in that these men experienced multiple ties to conventional society that helped them to continue on their desistant pathway despite the severing of the marital role.
These findings underscore the importance of (1) placing emphasis on the interactive nature of individuals and society and (2) recent calls for using an intersectional perspective, recognizing the salience of social position, to understand differences in the process of desistance (Fader and Traylor 2015). Findings from this research also provide a pathway for future inquiry.
First, these data limit the extent to which we can grapple with the temporal nature of change. Ideally, we would have information on the residential moves for the cohort over their life course, so that we could capture the timing of residential moves along with the timing of marriage and divorce. However, this information is not available in these data. Therefore, we are unable to account for the potential temporal order issue; it may be that residential change as a result of marriage and/or divorce is driving the process of desisting from crime (Kirk 2012). Future research could further investigate the interplay between mobility and relationships as this question is best addressed with data that is able to provide time-varying measures of offending, relationship changes, and residential mobility.
Second, we couch our findings in a modeling strategy akin to an “implication analysis” (Lieberson and Horwich 2008) by evaluating the implications that theories of desistance hold regarding the nature of change (abrupt vs. gradual; see Bersani and Doherty 2013 for a thorough discussion). This framing places emphasis on modeling one facet of the change process inferred by theories of desistance; in short, we examine the expected outcomes derived from theoretical propositions. As a result, less emphasis is placed on the consideration of factors that are not directly related to the theoretical tenets behind the desistance process under study. For instance, divorce can introduce new stressors and strains, both emotional and financial, as well as other negative experiences all of which may hold resonance for understanding the mechanisms of change. Future research should direct attention to the conditions surrounding the dissolution of marriage.
Conclusions
Among men living at the margins of society across their life course, situational mechanisms of the marriage effect appear to be most salient. While marriage may be indicative of a variety of mechanisms that reduce offending (e.g., exposure to greater levels of supervision or monitoring by spouses and extended family, the altering of routine activities to emphasize home life rather than street life reducing these men’s involvement in crime), this change in offending appears to be temporary and upon the dissolution of marriage these men return to criminal pursuits unlike their relatively more advantaged peers. In the words of Maruna and Roy, married individuals may knife-off contact with criminogenic peers (a “street divorce”) but marriage does not sever these ties permanently as “one-time friends and companions might come knocking on the door of the knifer (especially one who chooses to stay in his home community for various reasons)” (2007:108). This pattern is also representative of Sampson and Laub’s (1997) notion of cumulative disadvantage. Marginalized men who continue to occupy “specific positions of interlocking oppression” (Fader and Traylor 2015) may encounter multiple barriers to desistance that dilute the transformative potential of positive life experiences. To be clear, the fact that divorce was related to an increase in offending behavior, compared to when married, among persistently disadvantaged men in this sample does not suggest that persistently disadvantaged men do not desist from crime. Rather, this finding underscores the challenging nature of change in resource-deprived areas as well as the potential differences in the way mechanisms of change function across disparate contexts. In contrast, men in this sample who made relative advancements up the sociostructural ladder, as evidenced by moving to a better neighborhood, appear to be resilient in the face of adversity maintaining a desisting trajectory even though their marriage dissolved. Thus, the empirical data appear to support Waite and Gallagher’s (2000:25) assertion of the transformative power that evolves from the “promise of permanence” embedded in marriage, but only for those not living in disadvantage contexts.
This research continues to advance inquiry into the process of desistance from crime by looking at how marriage matters for those with lives shaped by a particular social structural position. While our findings suggest that even among a sample whose early life course is characterized by exposure to and residence in extreme disadvantage, the benefits of marriage resonate; our findings also suggest that how marriage matters in terms of desistance from crime is at least partially influenced by social structural position. In short, context matters.
Footnotes
Appendix
Individual Indicators of the Criminal Propensity Construct.
| Domain | Specific Domain | Variable Description | Percent (n = 494) |
|---|---|---|---|
| Individual | Difficult child | A two-item additive scale (α = .78), which combines childhood scores for immaturity (e.g., acts too young, cries too much, and seeks too much attention) and inattention (e.g., fidgets, unable to sit still), ranging from “0,” indicating the child was “within minimal limits of acceptable behavior,” to “3,” indicating “severely excessive” difficulty in these areas, as reported by the respondent’s teacher. Once scores for these two items were summed, the combined item (ranging from “0” to “6”) was dichotomized so that scores of “0” remained “0” and all other scores (“1” to “6”) were coded as “1,” indicating difficulty in this area. | 50.4 percent |
| Aggression | Teacher ratings of fighting too much, stealing, lying, resisting authority, damaging property, and being uncooperative, ranging from “0,” indicating no issues with aggression, to “3” indicating “severely excessive” aggressive behaviors. This item was also dichotomized, so that individuals with a score of “0” remained “0” in the binary measure and all other scores (“1” to “3”) were coded as “1,” indicating aggressive behavior. | 38.1 percent | |
| Risky behavior | A combination of dichotomized measures of delinquency (coded as “1” if the respondent was among the top 15 percent of the distribution of self-reported frequency of adolescent delinquent behaviors or among the top 10 percent of the distribution of self-reported frequency of adolescent delinquent behaviors prior to age 15), teen parent (coded “1” if the respondent reported having a child in adolescence), and early onset of drug use (coded “1” if the respondent self-reported using marijuana or alcohol prior to age 15). Risky behavior was coded “1” if respondents indicated any of these three included risks. | 59.5 percent | |
| Family | Poverty | Combined dichotomized measures of family poverty (“1” = family income is below the poverty line) and mother’s education (“1”= mother has less than 12 years of schooling). Those who were at risk on either variable (coded “1”) were coded as “1” in the combined measure, indicating family poverty. | 55.3 percent |
| Poor supervision | The ratio of children to adults within the home. This item was dichotomized so that ratios above 2.5 were coded as “1.” | 51.6 percent | |
| Residential mobility | The number of times that the family had moved since the respondent’s birth. Those who reported moving two or more times were coded as “1” while those who moved only once or not at all were coded as “0.” | 65.0 percent | |
| School | Academic achievement | Mean of first grade math grades, reading grades, and achievement scores for each respondent, with higher scores indicating lower achievement. This item was dichotomized so that those with mean scores of 2.33 and above were coded as “1” to indicate high risk. | 58.1 percent |
| High school dropout | High school dropouts (excluding those who graduated or completed an equivalency degree) were coded as “1” | 26.2 percent |
Acknowledgment
This study uses data from the Woodlawn project, which was designed and executed by Sheppard Kellam and Margaret Ensminger and funded by the National Institute of Drug Abuse and other NIH institutes through the years. We are especially grateful to these original researchers, the Woodlawn study participants, Woodlawn Advisory Board, and all of the researchers who have been instrumental in creating and maintaining this rich data set. We thank Jaclyn Cwick and Shytierra Gaston for their research assistance on this article. Finally, we thank Kerry Green and Margaret Ensminger for their comments on an earlier draft of this article and for their continued collaboration with the Woodlawn Study.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The current research was partially funded by NIDA grant R01 DA033999 and by the Harry Frank Guggenheim Foundation.
