Prison as a Turning Point? The Conditional Nature of Incarceration in Women's Offending Trajectories

Abstract

Objectives

This study examines incarceration as a turning point in criminal trajectories. It explores how offending behavior changes following incarceration and whether these changes vary based on prior offending patterns and specific prison experiences.

Methods

Using longitudinal data from 205 incarcerated women, we analyze offending trajectories over a ten-year period—five years before and five years after incarceration. We apply dual group-based trajectory modeling to identify patterns of continuity and change and assess whether transitions differ based on differential prison experiences, including victimization and access to employment. We complement the analysis with C-tests and growth-curve models.

Results

Continuity in offending is the most common pattern, particularly among individuals with low prior involvement. However, transitions—especially toward lower offending—are observed among those with higher preincarceration offending. Access to employment or training and the lack of visits correspond to transitions to less frequent offending.

Conclusions

Results are consistent with incarceration as a contingent turning point, with changes in pre- to postincarceration offending varying by prior trajectories and tending to be gradual. Patterns related to prison conditions appear sensitive to selection processes and sample size, warranting replication.

Keywords

incarceration turning points offending trajectories prison experiences women

From a life-course perspective, incarceration can act as a turning point, redirecting prior trajectories of criminal behavior. The direction of these changes, however, remains debated. While some perspectives emphasize its deterrent function, arguing that the experience of confinement increases the perceived costs of crime and promotes desistance, others highlight its criminogenic consequences, pointing to the ways in which incarceration disrupts social ties, exacerbates disadvantage, and may reinforce involvement in crime (Nagin 2013; Travis, Western, and Redburn 2014; Wakefield and Uggen 2010). Despite extensive empirical attention, most studies find little to no average effect of incarceration on subsequent offending (Loeffler and Nagin 2022; Nagin, Cullen, and Lero Jonson 2009). These largely null findings suggest that the consequences of incarceration may vary across individuals, with the aggregate effect potentially masking significant heterogeneity in how the experience of incarceration alters life trajectories.

Turning points are likely conditional on individuals’ prior offending trajectories and the circumstances in which the event occurs (Wheaton and Gotlib 1997). Recently, Hickert and colleagues (2021) discussed incarceration as a two-stage turning point that requires examining how the prison experience alters mechanisms that shape future offending. This, in turn, has important methodological implications. Rather than relying solely on postrelease outcomes, they argued, researchers must compare changes in offending after incarceration in contrast to offending before incarceration. And although most studies of incarceration and recidivism include controls for prior offending through static indicators of criminal history —partially capturing the “inertial, historicist character of the trajectories” (Abbott 1997:92)—they typically analyze offending after release without assessing changes in behavior relative to the preincarceration patterns (e.g., McCuish, Lussier, and Corrado 2018). This contrast between offending patterns prior to and after incarceration is central to identifying whether and for whom incarceration led to the “substantial change” that defines turning points (Elder, Johnson, and Crosnoe 2003), and—given the opposite theoretical expectations—in which direction.

Moreover, research on incarceration and recidivism has often treated confinement as a monolithic event (Nguyen and Loughran 2018; van Ginneken and Palmen 2023), overlooking the extent to which the experience of imprisonment varies across individuals. The conditions of confinement—including security levels, access to rehabilitative programs, social interactions, and exposure to violence—are likely to affect postrelease outcomes (Audry and Liebling 2020; Gaes and Camp 2009; McCuish et al. 2025; Zweig et al. 2015). The question, then, is not simply whether incarceration serves as a turning point but how the specific conditions of confinement influence shifts in criminal trajectories.

Given the differences in women's pathways into and out of crime, turning points are unlikely to be gender-invariant. Women tend to begin offending later, engage in lower-frequency trajectories, and desist earlier (Broidy and Thompson 2018). Their involvement in the criminal justice system is also more closely tied to economic marginality, substance use, and extensive histories of victimization (Daly 1992; Heimer, Malone, and De Coster 2023). These factors influence women's offending trajectories over the life course and may also condition how imprisonment is experienced. As suggested by growing work on women's imprisonment, these conditions are not experienced uniformly; rather, they interact with women's prior disadvantages and caregiving responsibilities in ways that may heighten or mitigate harm (Bucerius, Haggerty, and Dunford 2021; Crewe, Hulley, and Wright 2017). Although scholarship on other life-course domains has increasingly incorporated gender—analyzing women's responses to key transitions such as marriage, employment, and motherhood (e.g., Broidy and Cauffman 2017; DiPietro, Doherty, and Bersani 2018)—the study of incarceration as a turning point still relies largely on male samples.

This study builds on these insights to examine incarceration as a turning point among women. Using data from the Reintegration, Desistance, and Recidivism Among Female Inmates in Chile Study (RDFC), we analyze changes in offending trajectories from before to after incarceration. Following prior calls to examine pre- and postincarceration change (Hickert et al. 2021), we use a dual-trajectory modeling framework that provides a richer description of continuity and change in offending (Nguyen and Loughran 2018) and supplement these estimates with additional models that capture individual-level variation. We further extend prior work by assessing how specific conditions of confinement relate to changes in offending and by focusing on justice-impacted women, a less studied group with heterogeneous behavioral patterns. The results suggest that, while continuity in offending is the most likely outcome, change is common among women with more substantial involvement in offending prior to incarceration. Access to employment and lack of visits while incarcerated emerge as potential factors associated with transitions to lower offending patterns. Overall, by moving beyond aggregate indicators of reoffending and incorporating differences in the prison experience, this study contributes to a more nuanced understanding of incarceration as a contingent turning point in the life course of justice-impacted women.

Background

Incarceration and Reoffending

The relationship between incarceration and subsequent offending has been the subject of substantial empirical research and theoretical debate. Despite extensive empirical attention, most studies find little to no average effect of imprisonment on subsequent offending, with small positive or negative effects depending on institutional context (Loeffler and Nagin 2022). This pattern suggests that aggregate results may mask significant heterogeneity in the consequences of incarceration,¹ consistent with competing expectations about deterrence versus strain theories.

From a deterrence standpoint, the experience of confinement could shift perceptions of the costs associated with offending, thereby reducing crime as a behavioral response (Nagin 2013). This prediction aligns with Paternoster and Bushway's (2009) identity theory of desistance, which ties movement away from crime to the “crystallization of discontent,” when individuals realize that “the costs of crime and a criminal identity are beginning to become too great” (p. 1121). Incarceration—the more severe form of punishment and arguably an indicator of failure in a criminal trajectory (Hickert et al. 2021)—can become the event that motivates that recognition.

At the same time, prior work documents criminogenic effects associated with incarceration. Compared to other sanctions, confinement can weaken social ties with prosocial institutions that provide informal social control (e.g., Apel 2016), expose individuals to new criminal networks (e.g., Nguyen et al. 2017), and become a source of stigma and rejection (e.g., Pager 2003), overall disrupting access to legal opportunities while simultaneously increasing access to illegal opportunities upon release. These dynamics align with the predictions of general strain theory, which states that exposure to negative experiences, the removal of positive stimuli, and the blockage of opportunities to achieve individuals’ goals generate negative emotions (e.g., anger), which may, in turn, lead to offending when legitimate coping resources and constraints are weak (Agnew 1992; Broidy and Agnew 1997). As the experience of confinement implies the loss of freedom and the exposure to additional negative stimuli—for example, victimization—imprisonment may become a source of strain, increasing the risk of offending as a coping mechanism (Listwan et al. 2013; Slocum, Simpson, and Smith 2005). For example, Listwan et al. (2013) showed that a negative prison environment, that is, where individuals felt afraid of being victimized and were exposed to frequent fights, was associated with an increased risk of rearrest and recommitment to prison.

Taken together, while deterrence and general strain both treat incarceration largely as a negative experience, they differ in their assessment of its behavioral consequences—whether it increases or decreases criminal activity. If, as life-course work suggests, incarceration can be a positive turning point for some but a negative one for others (e.g., Laub and Sampson 2003), then the null effect is plausibly the result of averaging over heterogeneous responses to the same—though diversely experienced–event.

Incarceration as a Contingent Turning Point

Turning points are described as specific events that generate significant changes in the course of someone's long-term patterns of behavior (Abbott 1997; Sampson and Laub 1990). As such, turning points are necessarily recognized “with respect to a previously established trajectory” (Wheaton and Gotlib 1997:5) or “master narrative” (Abbott 1997:93)—the pattern that is altered, abruptly or gradually, after the event. Methodologically, this means that analyzing turning points requires comparing the development of a specific behavior prior to and after the event of interest (see Nguyen and Loughran 2018, for a similar discussion). To illustrate, recidivism—the standard metric for assessing the link between incarceration and reoffending (NAS 2022)—does not necessarily signal a change in trajectory. Even when measured in levels rather than as a binary outcome, high postrelease offending can reflect continuity for high-rate trajectories but escalation for individuals with low or sporadic prior involvement.

In line with this approach, Hickert and colleagues (2021) advance the question of prison as a turning point by analyzing shifts in the number of offenses from before to after confinement and examining the mechanisms behind these shifts. Using a sample of Dutch-born men, they report that over two-thirds maintained stable levels of offending following incarceration, while those who decreased their criminal activity outnumbered those who escalated by nearly three to one. These results align with the only two studies that analyzed changes in offending before and after incarceration (Bottoms and Shapland 2011; Frisch 2018): while continuity prevails, reductions in offending are more common than increases. The results highlight the diverse impact of incarceration experiences across individuals.²

Substantively, because turning points are defined by change relative to what came before, their effects are often conditional rather than universal (Bersani and Doherty 2018; Elder 1985). In particular, changes in offending are especially likely contingent on prior criminal trajectories (Wheaton and Gotlib 1997). As Nguyen and Loughran note, prior life experiences “can themselves condition how an individual might respond to a given life event” (Nguyen and Loughran 2018:350). Using causal inference parlance, they refer to this as treatment effect heterogeneity: exposure to a similar event (i.e., incarceration) can yield different outcomes depending, among other factors, on prior experience with the system (see also Wright et al. 2001). The expectations, however, are not straightforward. Some research suggests incarceration is particularly harmful to individuals with limited prior offending (Nieuwbeerta, Nagin, and Blokland 2009), and to those least likely to experience it (Turney and Wildeman 2015).³ Yet life-history accounts also show both continuity and change among those who churn through the system: for some, new sentences become part of life; for others, a single experience after subsequent terms may be “the icing on the cake,” leading to a process of desistance as they realize that “enough is enough” (Carlsson 2012:9; see also Paternoster and Bushway 2009).

Gender Pathways and Incarceration

Life-course research has increasingly emphasized the role gender plays in access to, the experience of, and responses to particular turning points. A growing literature documents gender-specific mechanisms for well-studied turning points such as partnerships and employment (e.g., DiPietro, Doherty, and Bersani 2018; Leverentz 2014; Rodermond et al. 2023), with many of these works raising doubts about the relative consensus on their benefits, especially among women in marginalized positions, such as those impacted by the criminal justice system (Bersani and Doherty 2018).

Women have felt the expanding reach of the criminal justice system, with the female prison population rising in recent decades (Fair and Walmsley 2024), but have too often been treated as peripheral in analyses of imprisonment and punishment (Crewe, Hulley, and Wright 2017; Kruttschnitt and Gartner 2003). Yet the experience of and response to incarceration are often gendered. Theoretically, although deterrence implicitly treats decision-making as gender-neutral, general strain emphasizes the gendered ways in which strain is experienced and managed, with women facing distinct forms of strain (e.g., victimization) and adopting different emotional and behavioral coping strategies (Broidy and Agnew 1997). Evidence among incarcerated women further underscores that the same event need not yield the same response across women (Slocum, Simpson, and Smith 2005; Slocum et al. 2022).

Importantly, there are gendered pathways into and out of the criminal justice system. Women's life-course offending trajectories are, like men's, heterogeneous (Broidy and Thompson 2018). While they share similarities in temporal patterns, women are overrepresented in low-rate groups and are less likely to populate chronic, high-rate trajectories (Broidy, Payne, and Piquero 2018). It is also well recognized that justice-involved women present multiple and intersecting markers of marginalization—including economic precarity, trauma exposure, mental-health and substance-use needs, and caregiving responsibilities (Heimer, Malone, and De Coster 2023). These factors shape pathways into prison and are likely to define the experience (Bucerius, Haggerty, and Dunford 2021) and pathways out of it (Rodermond et al. 2023).

While scholarship has expanded on how women experience imprisonment, highlighting distinct “pains” and constraints (Bucerius, Haggerty, and Dunford 2021; Crewe, Hulley, and Wright 2017), it has been more limited in analyzing its consequences as a potential turning point. In fact, the three studies that rely on a before-and-after prison comparison do not analyze gender differences, with two using an all-male sample (Bottoms and Shapland 2011; Hickert et al. 2021). As discussed, those results are not necessarily generalizable to incarcerated women; thus, we still do not know whether—and for whom—incarceration becomes a turning point in this case. Accordingly, we ask:

Research Question 1: How do offending patterns change after incarceration, and how does this change differ across preincarceration trajectory groups?

The Heterogeneous Experience of Incarceration

While research on other life-course turning points has shown that the nature of the experience matters—with, for example, job quality conditioning the association between employment and desistance (Apel and Horney 2018; Jaynes 2020), studies on incarceration and reoffending have often treated confinement as a uniform and homogeneous experience (van Ginneken and Palmen 2023). The literature on the pains of imprisonment argues otherwise (Sykes 1958): confinement comprises distinct deprivations (liberty, autonomy, security, goods/services, intimate ties) that shape the meaning of custody and vary across people and places (Crewe, Hulley, and Wright 2017; Haggerty and Bucerius 2020). Consistent with the gendered-pathways argument, women's lived experience of custody is also known to diverge across contexts and life histories, reinforcing the fact that imprisonment is not a monolithic event (Bucerius and Sandberg 2022).

The prison experience varies along multiple, often measurable, dimensions—security level, institutional social climate (e.g., order, legitimacy, safety), and access to interventions and visits (Auty and Liebling 2020; Hyatt 2024; Steiner, Makarios, and Travis 2017; van Ginneken and Palmen 2023). These objective conditions provide different subjective experiences that differ both across and within the facilities. Importantly for this work, variations in prison experiences have been empirically associated with differential chances of future offending. For example, individuals released from higher-security prisons are more likely to be reconvicted than similarly classified individuals released from lower-security institutions (Gaes and Camp 2009). Greater perceived staff support lowers the chances of reconviction (McCuish, Lussier, and Corrado 2018), while victimization is associated with higher chances (Zweig et al. 2015).

Stratification research, on the other hand, while recognizing incarceration as an institution that reproduces inequality through stigma, disrupted careers, and constrained opportunities after release (Kirk and Wakefield 2018; Western 2006), also notes its potential to “compensate for disadvantage” when systems provide credible education, treatment, and reentry supports (Wakefield and Uggen 2010:394). Cross-national analyses indeed suggest that systems that prioritize rehabilitation over punishment are more likely to report crime-reducing effects (Loeffler and Nagin 2022). However, the effect of access to prison education and employment training on recidivism is far from consistent.⁴

In terms of gender, women's facilities tend to differ from men's in regard to distance from home (and corresponding access to visitation), levels of security, and access to interventions (Belknap 2021). While classic accounts of the “pains of imprisonment” apply broadly, many of these pains—such as the loss of family contact, separation from children, lack of privacy, and threats to psychological well-being—are experienced more acutely by women (Crewe, Hulley, and Wright 2017; Owen 1998). At the same time, how women experience prison is closely tied to the challenges and opportunities they encounter outside prison (Bucerius and Sandberg 2022), suggesting variation within gender. While for some women confinement is associated with severe pains and isolation (Crewe, Hulley, and Wright 2017), others describe incarceration as offering temporary refuge from violence and instability (Bucerius, Haggerty, and Dunford 2021).

Overall, incarceration represents a fundamentally heterogeneous experience, one that, as Auty and Liebling (2020:358) note, “may both damage and repair” individuals. In causal inference terms, such variation reflects treatment heterogeneity (Nguyen and Loughran 2018), acknowledging significant differences in the form and context in which a sentence is served. When connected to the theoretical debates between strain and deterrence, this insight suggests that the same event (e.g., incarceration) can serve as a turning point toward desistance or, conversely, lead to an escalation in offending, depending on the context in which confinement occurs. To explore this possibility, we posed two additional questions:

Research Question 2: Does a more punitive prison experience increase the likelihood of moving toward lower or higher criminal activity?

Research Question 3: Does access to rehabilitative opportunities correspond to decreased criminal activity after release?

Given the data limitations detailed below, we see our analysis of the heterogeneity of prison experiences as an exploratory, yet theoretically relevant, step.

Context of the Study

This study examines incarceration as a potential turning point among incarcerated women in Chile, a context that differs in important ways from the settings that dominate the existing literature. Chile's penal system combines comparatively high incarceration rates with limited rehabilitative capacity, placing it closer to punitive models common in Latin America yet without the scale and racialized dynamics characteristic of U.S. mass incarceration (Bergman and Fondevila 2021). At the time of this study, Chile imprisoned roughly 240–245 individuals per 100,000 residents (Fair and Walmsley 2024), one of the highest rates among OECD countries and in Latin America, though substantially below the United States. Women represented about 8%–10% of the prison population, a share that has grown steadily, mirroring trends observed in several countries where female incarceration has increased at a faster rate than male incarceration (Heimer, Malone, and De Coster 2023).

As in other contexts, women incarcerated in Chile come from highly disadvantaged backgrounds. They typically report low educational attainment, unstable labor histories, substance and mental-health problems, and nearly universal motherhood (Morales et al. 2016; Rufs, Larroulet, and Valenzuela 2023). These characteristics align with the literature on gendered pathways into crime, emphasizing cumulative disadvantage and victimization as key predictors of women's criminal involvement in Chile (Larroulet et al. 2020), and determinants of their prison experience (Sanhueza et al. 2019).

In comparative perspective, women in Chile serve longer sentences and under harsher conditions than in Western European systems, but shorter and less institutionally differentiated than in the United States. Chilean prisons provide limited programming and tend to operate under material scarcity, overcrowding, and inconsistent access to basic services. National monitoring bodies document frequent shortages of sanitation, hot water, and essential goods (INDH 2019). Access to work, education, or vocational programs reaches only a minority of incarcerated women, and tends to be highly gendered: sewing, cooking, and other traditionally feminized activities remain far more common than training linked to the formal labor market ( Larroulet et al. 2021a). More intensive interventions based on the risk-need-responsivity model are available only to those serving longer sentences and cover a relatively small fraction of the population (Centro de Estudios Justicia y Sociedad 2016).

These dimensions may impact both the meaning of incarceration and the mechanisms through which it may redirect or reinforce behavioral trajectories. By examining women incarcerated in Chile, this study extends prior research on prison as a turning point to a setting where incarceration is punitive and not embedded in a strong welfare state, and where class-based marginalization primarily structures criminal justice contact. In this context, incarceration may intensify pre-existing disadvantage for some women, reinforcing continuity in offending trajectories (Wakefield and Uggen 2010), while for others it may function as a temporary “refuge” (Bucerius, Haggerty, and Dunford 2021), providing access to employment and interventions otherwise unavailable in the community, thereby facilitating incremental movement toward desistance. Prior evidence among juveniles in Chile suggests that the effects of the system are indeed heterogeneous across populations (Cortés, Grau, and Rivera 2019).

Taken together, the Chilean context helps motivate the theoretical expectations. The mix of punitive conditions and limited support may reinforce disadvantage for some women while offering opportunities for others, suggesting heterogeneous responses rather than uniform change.

Methods and Data

We use data from the RDFC, an intensive longitudinal study that followed a cohort of women released from prison in Santiago, Chile, between September 2016 and March 2017. The target population was all Chilean women serving custodial sentences of at least 30 days in the only prison facility for sentenced women in Santiago, and who were released either under parole or after completing their entire sentence. Because Chile does not distinguish institutionally between jail and prison, the 30-day criterion serves as an eligibility cutoff rather than a marker of short-term confinement.⁵ Over 80% of the population of interest participated in the study (n = 225).⁶ Participants were followed over 12 months, with five interviews conducted during that time. The current paper uses only the information collected at baseline (one to two weeks before release), when the questions about prison experience were asked.

For this study, the surveys were complemented with official data provided by the National Prosecutorial Office, guaranteeing an observation window of at least ten years beyond the time spent in prison for their index offense (i.e., the offense and conviction that led to their participation in the RDFC). These data include all cases referred to the prosecutor's office since the participants were 18 years old up until 2022 (i.e., about five years after release from the index conviction). We were able to match the information for 208 of the 225 women. Three women were excluded from the analysis as they were less than 20 years old at the time of release, meaning we only observed them for less than two years before the index conviction.

The final analytic sample corresponds to 205 women. At the time of release, they were, on average, 37 years old and had served an average sentence of 27 months, with a range from 30 days to 15 years. In total, 45% of the participants had an index sentence of one year or less, while 25% had served more than three years. In terms of the index offense, 46% were released after a sentence for a nonviolent property offense (mostly theft), 37% for drug trafficking, and 17% for violent crimes, in most cases, robbery. The index sentence was the first incarceration term for only 23% of the participants. Among those with a prior incarceration history, the number of prior terms ranged from one to 54, indicating that some women had substantial criminal-justice contact.

Measures

Offending: The dependent variable is the number of criminal cases in which an individual was prosecuted during a ten-year period. This period includes the five years before incarceration and the five years after release for the index offense; the time spent in prison is not included. We divided these ten years into 20 six-month intervals—ten before prison and ten after release. For each individual, we counted the number of prosecuted cases in each interval.

Prison Experience: We measure the experience of incarceration at baseline using two sets of indicators. The first set captures punitive exposures, tapping into deprivations central to the “pains of imprisonment” framework that may map onto both perceived costs (deterrence) and negative stimuli (strain).⁷ Specifically, we include Sentence length—the number of months sentenced, standardized to address skew, and Absence of release privileges, a dichotomous measure coded 1 when no release privilege was granted (e.g., Sunday or weekend release, or controlled daily permits to study or work outside) and 0 otherwise; all such benefits imply time outside prison walls. Victimization is a dichotomous variable coded 1 when the respondent reports any physical or sexual victimization by staff or other incarcerated people and 0 otherwise. In terms of social ties, we measure Lack of visits, coded 1 when the respondent reports no in-person visits in the past six months from immediate or extended family, partners, or friends (and 0 otherwise), and Social isolation, coded 1 when the respondent reports having no friends inside the facility (and 0 otherwise). Higher values denote more punitive experiences.

The second set of indicators captures access to rehabilitative opportunities that incarceration may provide, many of which are largely absent on the outside. We include three indicators, each one dichotomous with 1 indicating access: Access to employment/training (worked in a prison job or participated in a vocational/training course), Access to education (completed at least one grade/level while incarcerated), and Access to intervention (participated in any rehabilitative/treatment program). Because all women were released from the same facility, prison-level environment measures do not vary in this sample and are therefore not included. All descriptive statistics are presented in Table 1.

Table 1.

Descriptive Statistics.

	Mean	SD	Min	Max	N
Offending
Total offense counts five years prior to incarceration	15.67	18.41	0	146	205
Total offense counts five years post incarceration	10.59	16.15	0	99	205
Prison experience
Punitive exposures
Sentence length (months)	26.57	33.01	1	180	204
Absence of release privileges	0.80	-	0	1	201
Victimization	0.18	-	0	1	202
Lack of visits	0.20	-	0	1	196
Social isolation	0.33	-	0	1	205
Rehabilitative opportunities
Access to employment/training	0.50	-	0	1	203
Access to education	0.27	-	0	1	202
Access to intervention	0.35	-	0	1	202

Analytic Approach

To answer the research questions, we estimate a dual group-based trajectory model (DGTM), an extension of the traditional group-based trajectories that directly models the association between the developmental course of two related outcomes (Nagin and Odgers 2010).

We first estimate pre- and postprison offending trajectories independently, using a Poisson model and a cubic time function for each group. Time is measured relative to the index prison sentence (five years prior to and five years postrelease), with the outcome being the number of prosecuted cases in each six-month interval. We select the number of groups using Nagin's (2005) criteria: the Bayesian Information Criterion (BIC), the average posterior probability of group assignment, and the odds of correct classification. Given the sample size constraints, we prioritize parsimony, selecting a solution that captures the key features of the data while ensuring a minimum number of cases per group (Appendix Table A1).⁸ We then trim the model, removing nonsignificant higher-order terms.

We next estimate the DGTM using the original trajectories as starting values. The DGTM provides a set of conditional probabilities that reflect the likelihood that an individual will follow a specific trajectory after release, conditional on the preincarceration trajectory, P(posttrajectory k| pretrajectory j). These probabilities provide a richer description of continuity versus change than a single correlational measure, as they recover both the direction of change and the conditionality on prior patterns of behavior. As such, they are our primary evidence for RQ1 (see also Frisch 2018).

Finally, the DGTM allows transition probabilities to vary with covariates (Jones and Nagin 2007; Larroulet et al. 2021b; Nagin and Odgers 2010), using a constrained multinomial logit:

π_{k | j} (w_{i}) = \frac{e^{γ_{k | j}^{0} + γ_{k}^{^{'}} w_{i}}}{\sum_{k} e^{γ_{k | j k}^{0} + γ_{k}^{^{'}} w_{i}}}, j = 1, \dots, J .

where k represents a participant's offending trajectory after release and j her pre incarceration trajectory,

π_{k | j}

represents the transition probabilities that link pre- and postprison trajectories, and

w_{i}

represents the set of predictors for these probabilities; in our case, the different indicators of the experience of incarceration. The constraint holds the effect of

w_{i}

constant across preprison groups (i.e., no interactions in the main model). We use these estimates to address RQ2 and RQ3. Given the sample size and the collinearity among the experience indicators, we report bivariate specifications (each indicator entered singly, controlling for preprison group membership); sensitivity analyses are summarized at the end of the “Results.”

We select DGTM as our analytic method because it closely reflects the conceptualization of turning points as disruptions or redirections of long-term patterns of behavior (Abbot 1997; Wheaton and Gotlib 1997) and provides an intuitive visual and statistical representation of the process of continuity and change when heterogeneity in the response is expected. It also suits our goals for three reasons. First, as argued by Nguyen and Loughran (2018), trajectories are the appropriate outcome when analyzing turning points, as they capture the process of desistance rather than the binary event of recidivism (see also Bushway et al. 2003) and account for heterogeneity in offending patterns (Eggleston, Laub and Sampson 2009). Second, aligning with Hickert et al. (2021), DGTM directly links pre- and postrelease trajectories via conditional transition probabilities. This is particularly important as a single aggregate measure can mask the fact that the same event—incarceration—may lead to distinct patterns of change (e.g., Laub and Sampson 2003), consistent with the opposing expectations under deterrence versus strain theories. Directionality may also differ systematically by prior trajectory (e.g., Nieuwbeerta, Nagin, and Blokland 2009). Third, DGTM permits the transition process to vary with prison experiences, which is central to testing whether more punitive exposures or rehabilitative opportunities are associated with the continuity and change.

As with any analytic approach, the DGTM entails trade-offs. Trajectory groups are statistical approximations that summarize latent heterogeneity, so some within-group variability and classification uncertainty remain. In addition, because groups are estimated from the cohort's observed distribution, classification is sample-relative: an individual may be labeled “stable” even when absolute change has occurred if that change is small relative to peers, and “changing” when the absolute change is modest but atypical in that cohort.

We complement this group-based analysis with two additional models. First, following Hickert et al. (2021), we implement the C-test, a conditional test that compares two Poisson means to assess whether an individual's postincarceration offending differs from the preincarceration rate beyond chance (Przyborowski and Wilenski 1940). For each woman, we test the postperiod count conditional on the total across both periods and run one-sided tests for decreases and for increases, classifying direction by the side that rejects. We treat p < .20 as a classification threshold for flagging individual change, reflecting the lower power of tests at the individual level. Unlike growth-curve models discussed below, which estimate average change, the C-test directly identifies the direction of individual-level change, which is essential given our opposing expectations in RQ1. We then use the resulting three-category indicator (Down/Stable/Up) as the outcome in multinomial logistic models to test whether our indicators of prison experience were associated with the direction of change.

As a second strategy, we estimate multilevel piecewise growth-curve models (GCM) to quantify average trajectory change and individual heterogeneity (Singer and Willett 2003). This strategy has been used in life-course criminology to examine behavioral change around critical life events (e.g., Bersani, Laub and Nieuwbeerta 2009; Horney, Osgood, and Marshall 1995). The model fits separate pre- and postincarceration slopes, allowing the data to estimate whether trajectories change direction at incarceration. Time is centered such that t = 0 represents time incarcerated (and is excluded from the analysis). We estimated random intercepts (capturing between-person differences in offending levels), and random postincarceration slopes (capturing variation in postrelease change rates). This second strategy provides an estimate of the magnitude of within-person change in concrete units (offenses per semester) and an example of the aggregate effect we aim to unpack. We also introduce preincarceration trajectory memberships as predictors of postrelease slopes via interaction terms, testing whether change varies among different groups. Finally, we rely on this strategy as an additional assessment of whether prison experiences moderate the rate of postincarceration change, and to conduct complementary analyses of age.

Results

Pre- and Postprison Trajectories

Figure 1, panel A presents the three-group solution for offending trajectories prior to incarceration. Approximately 55.6% of the sample is classified into a low-offending group, with an average of fewer than one offense per six-month period, except for the semester immediately preceding incarceration, when the average increases to 1.5 offenses. This group has a mean age of 38 at the time of incarceration, and the majority (57.4%) were released after serving a sentence for drug-related offenses. A second group (34.6% of the sample) corresponds to a medium-risk group, with an average number of offenses per semester ranging from 1.9 to 2.8. The mean age in this group is 32, and the vast majority (71%) were serving sentences for nonviolent property offenses. Finally, the high-offending group comprises approximately 10% of the sample. The mean number of offenses in this group varies between 1.8 and 11.1 per semester. This is the youngest group, with an average age of 29, and most individuals (74%) were incarcerated for nonviolent property offenses, largely theft (93%).

Figure 1.

Offending trajectories preprison and postprison.

Regarding postrelease offending trajectories (panel B), a three-group solution was also selected.⁹ The trajectories and characteristics of the groups largely mirror the preincarceration patterns, although, when considering the average over five years after release, the total number of offenses was lower than prior to incarceration, indicating an overall decline in the number of offenses prosecuted (Appendix Table A2, panel A).¹⁰ This pattern may reflect a combination of factors, including the process of aging and variations in crime rates. We return to this point later.

Approximately 57% belong to a low-risk group, in which the average number of offenses remains below one throughout the follow-up period. This group has a mean age of 39 at the time of release, and the majority (57%) were released after being incarcerated for drug-related offenses. Thirty-five percent of the sample belongs to a medium-risk trajectory, with an average number of offenses per semester ranging from 0.6 to 2.3. The mean age in this group is 34, and most women (70%) were released after serving a sentence for nonviolent property offenses. Finally, the high-risk group comprises only 8% of the sample, with a mean number of offenses that ranges from 1.4 to 8.1 per semester. This group has an average age of 31, and the vast majority (88%) were incarcerated for nonviolent property offenses, particularly theft.

Changes in Offending Trajectories

Panel A of Table 2 displays the conditional probabilities of belonging to a postprison trajectory, given membership in a preprison group. These conditional probabilities provide a statistical representation of prison as a turning point: whether incarceration reshapes (or not) prior behavior, and in which direction. Panel B displays the joint probability of belonging to both a preprison and postprison group, meaning all possible combinations are shown, and their total sums to one.

Table 2.

Probabilities of Group Membership (n = 205).

Panel A: Conditional Probabilities of Group Membership
		Postincarceration
		Low	Medium	High
Preincarceration	Low	82.8	16.6	0.5
	Medium	31.1	54.0	14.9
	High	5.0	64.4	30.6

Panel B: Joint Probabilities of Group Membership
		Postincarceration
		Low	Medium	High
Preincarceration	Low	46.4	9.3	0.3
	Medium	10.7	18.5	5.1
	High	0.5	6.3	3.0

In terms of our first question, while continuity is prevalent—67.9% of the sample follows a pattern of continuity, remaining on the diagonal when we condition postprison to preprison trajectories (Table 2, panel B), change also occurs. Importantly, the level of continuity differs by the preincarceration trajectory group. Consistent with the relative state-dependent nature of offending, individuals in the low-risk preprison group show the highest likelihood of remaining in the same trajectory (83%), with the likelihood of continuity decreasing as trajectories reflect higher levels of involvement in offending. In fact, two-thirds of the continuity is driven by individuals in the low-low trajectories, not by those with higher or more persistent offending patterns. It is important to note that, given the overall decline in total offenses and the cohort-relative classification used by DGTM, “stable” groups still exhibit reductions in absolute terms, though smaller than those moving toward lower-level trajectories. Indeed, when we examine absolute change, we observe “directional coherence” with DGTM transitions (e.g., the medium-to-low transition declines more than medium-to-medium; see Appendix Table A2, panel B).

In terms of changes, a transition toward reducing offending—that is, desistance—is slightly more likely than one toward an increase in offending (17.5% and 14.7%, respectively). In total, 31% of those classified in a medium-risk trajectory prior to the focal prison term transitioned toward a low-risk trajectory afterward, with individuals in the high-risk group exhibiting a high probability (64%) of shifting to a medium-risk trajectory but a lower chance of transitioning to a low-risk trajectory (5%). Interestingly, this lower chance is also observed when analyzing the transition from the low- to the high-risk group, suggesting that, while change occurs, it does so in small, proximal steps. Models that include a time-varying indicator for being sentenced to incarceration during each six-month interval yield substantively similar results.¹¹

Overall, these results suggest that prison is a heterogeneous turning point in terms of offending trajectories, and that this heterogeneity depends, at least to some extent, on the pattern of behavior preceding incarceration. We now turn to our second and third questions, which explore variation in the experience of incarceration and its association with postrelease offending.

Diverse Prison Experiences

While the literature often treats incarceration as a uniform treatment, the experience of confinement likely differs across individuals. We analyze whether specific dimensions of that experience are associated with the probability of transitioning to each postprison trajectory. We distinguish dimensions that reflect a more punitive experience (i.e., sentence length, victimization, isolation), in line with deterrence and strain framings, from dimensions that indicate access to rehabilitative opportunities that may support the transition to the community. Table 3 reports bivariate associations between each indicator and postprison group membership, with preprison trajectory membership included in all models. Coefficients represent the change in log-odds of membership in the medium- or high-risk postprison group relative to the low-risk reference group.

Table 3.

Bivariate Effect of Prison Experience on Conditional Log-Odds of Postprison Trajectory Group (Ref: Low Group).

Variable	Coefficient	95 Percent CI	p Value
Variable	(SE)
Panel A. Medium group
Punitive exposures
Sentence length	−0.45	[−0.88, −0.03]	.038
Sentence length	(0.22)		.038
Absence of release privileges	1.62	[0.42, 2.81]	.008
Absence of release privileges	(0.61)		.008
Victimization	−0.69	[−1.68, 0.30]	.171
Victimization	(0.50)		.171
Social isolation	0.62	[−0.16, 1.40]	.117
Social isolation	(0.40)		.117
Lack of visits	−0.47	[−1.46, 0.51]	.348
Lack of visits	(0.50)
Rehabilitative opportunities
Access to employment/training	−1.19	[−1.98, −0.40]	.003
Access to employment/training	(0.40)		.003
Access to education	−0.09	[−0.95, 0.77]	.841
Access to education	(0.44)		.841
Access to intervention	−0.56	[−1.38, 0.26]	.179
Access to intervention	(0.42)		.179
Panel B. High group
Punitive exposures
Sentence length	−2.69	[−5.70, 0.33]	.081
Sentence length	(1.54)		.081
Absence of release privileges	16.08	[−2,245.33, 2,277.48]	.989
Absence of release privileges	(1,153.78)		.989
Victimization	−1.67	[−3.49, 0.16]	.074
Victimization	(0.93)		.074
Social isolation	0.39	[−0.93, 1.71]	.564
Social isolation	(0.68)
Lack of visits	−1.68	[−3.36, −0.00]	.050
Lack of visits	(0.86)		.050
Rehabilitative opportunities
Access to employment/training	−2.92	[−5.11, −0.74]	.009
Access to employment/training	(1.12)		.009
Access to education	−15.65	[−2,090.45, 2,059.15]	.988
Access to education	(1,058.57)		.988
Access to intervention	−1.21	[−2.92, 0.49]	.164
Access to intervention		(0.87)	.164

Note. Indicators are included in separate multinomial logit models that adjust for preprison trajectory membership. CI = confidence interval.

Among the punitive dimensions (RQ2), longer sentences are negatively associated with membership in the medium-risk group (vs. low-risk), while having no release privileges is positively associated with transitions toward more criminally involved trajectories. No other punitive indicator is statistically distinguishable from zero for medium- versus low-risk membership. In terms of direction, however, victimization and lack of visits are negatively associated with membership in the medium group, consistent with a deterrence expectation. For high-risk membership (vs. low-risk), directions mirror those for the medium-risk comparison, although only victimization and lack of visits reach conventional significance.

In terms of opportunities (RQ3), only access to work or training while incarcerated shows a significant and consistent association with membership in medium- or high-offending trajectories relative to the low-risk trajectory. Access to education and access to interventions are also negatively related to group membership, but with large standard errors and not statistically distinguishable from zero. When transformed into conditional probabilities (Appendix Figure A1), results suggest that access to work/training corresponds to lower probabilities of remaining in a medium and a higher probability of transitioning to a low postprison trajectory for those in a medium-risk preprison one, and a shift in probability from high toward medium among those in a high preprison trajectory, consistent with the idea of desistance being a gradual process.

When we estimated a model including all predictors that were statistically significant in the bivariate models, the direction of the coefficients remained unchanged, but the magnitudes were smaller, and the standard errors increased (Appendix Table A4). Only access to training and lack of visits remained statistically distinguishable from zero (p < .10). The loss of precision is consistent with the modest sample size and the correlations among indicators (Appendix Table A5). The results should be interpreted cautiously and as descriptive associations.

Additional Results

We conduct two additional sets of analyses to complement the DGTM. For our first question, the C-test addresses the direction and extent of change at the person level, while piecewise growth-curve models quantify average continuous (within-person) change (slopes) in concrete units (offenses per semester) and test whether postrelease change rates differ by preincarceration trajectories.

Based on the C-test analysis (α = .20), 51.2% of the women are classified Down, 36.6% Stable, and 12.2% Up. The large excess of Down versus the 20% expected under the null is robust across different α values (see Table 4, panel A, in bold). Compared with the DGTM, the C-test identifies more individual-level declines, whereas the DGTM emphasizes stability.¹² Indeed, nearly half of the women the DGTM labels as Stable in their behavior are classified as Down by the C-test, while agreement is relatively high for Down and Up (97.4% and 66.7%, respectively; Appendix Table A6). Overall, both methods indicate heterogeneity in direction but differ in scale: DGTM (cohort-relative) can mask individual shifts when many move similarly, while the C-test is person level and thus more sensitive to both true shifts and noise in official records. Some divergence is therefore expected, and we view them as complementary, providing a reasonable range for pre- to postrelease change.

Table 4.

Additional Analyses on Changes in Offending Pre- and Postincarceration.

Panel A: Change in Offending Five Years Prior to and Postincarceration (C-test results)
Umbral α	Down	Stable	Up	Excess of “Down”vs. Chance
.10	44.4% (91)	46.3% (95)	9.3% (19)	+34.4
.15	49.3% (101)	40.5% (83)	10.2% (21)	+29.3
.20	51.2% (105)	36.6% (75)	12.2% (25)	+31.2
.25	53.2% (109)	33.2% (68)	13.7% (28)	+28.2

Panel B: Piecewise Growth Curve Models Predicting Offending Count
	(1)	(2)	(3)
	M0: Null	M1: Base Slopes	M2: Prior Trajectories
Preincarceration slope		0.070**	0.068**
		(0.022)	(0.021)
Postincarceration slope		−0.143***	−0.066***
		(0.025)	(0.016)
Trajectory group prior incarceration (Ref: Low)
Medium			1.662***
			(0.131)
High			5.405***
			(0.587)
Medium × Postslope			−0.087**
Medium × Postslope			(0.032)
High × Postslope			−0.471***
High × Postslope			(0.093)
Constant	1.347***	1.937***	0.829***
	(0.110)	(0.200)	(0.113)
var(Intercept)	1.999	2.943	0.346
var(Postslope)		0.0216	0.0032
var(Residual)	9.414	8.999	9.005
ICC	0.175
Observations	4,027	4,027	4,027
Log-likelihood	−10,396.94	−10,343.65	−10,219.65
AIC	20,799.88	20,701.30	20,461.31
BIC	20,818.78	20,745.40	20,530.62

Note. N = 205 women observed over 20 semesters relative to incarceration (time 0 excluded). Robust standard errors in parentheses. Variance components estimated via maximum likelihood.

*p < .05, **p < .01, ***p < .001.

The unconditional GCM (Table 4, panel B; model 1) indicates a significant change in offending patterns around incarceration. Prior to incarceration, offending increases modestly (β = 0.070, SE = 0.022, p = .002), while the slope following release is negative (β = −0.143, SE = 0.025, p < .001). The net shift corresponds to about 0.213 fewer offenses per six months (p < .001), consistent with the aggregate drop observed in prior analyses. This overall pattern, however, is attenuated by 54% when we include preincarceration trajectory group membership (model 2; β = −0.066, SE = 0.016), indicating that the average change in offending following incarceration is negative but small. Importantly, this reduction in magnitude also suggests that change is not uniform across the population. Women in the medium and high pretrajectories show higher levels of offending over time, but also larger postrelease declines than the low-risk group (0.087 and 0.471 fewer offenses per semester, respectively). This heterogeneity pattern holds across specifications.

For our second and third research questions, we estimate multinomial logit models to explore whether prison experiences are associated with C-test change groups (Down/Stable/Up; Appendix Table A7), and piecewise GCMs to examine whether the same experiences are associated with variation in postincarceration slopes (Appendix Table A8). Although indicators are entered in separate bivariate models, few emerge as statistically distinguishable across both approaches. In the C-test models, longer sentences are associated with the lower odds of moving up in offending, the absence of release privileges with the lower odds of remaining stable, and access to work with a reduced likelihood of moving up (relative to Stable and Down).¹³ In the GCMs, only two factors moderate postprison slopes: reporting no visits during incarceration is associated with steeper postrelease declines, whereas participation in job training is associated with a less steep decline, an inconsistent association compared to earlier analyses.¹⁴ More importantly, the GCMs also indicate selection: program participants have lower predicted offending at incarceration (negative intercept effects), and the association between prison experiences and postrelease offending is significantly attenuated once preprison trajectories are included (Appendix Table A8). Although these models are not directly comparable, the estimates are sensitive to specification; accordingly, caution is warranted when interpreting moderation by prison experiences. We return to these limitations in the “Discussion.”

Aging and Turning Points

One of the central challenges in studying changes in offending relates to the role of age. As widely discussed in the field, desistance is influenced by aging (Hirschi and Gottfredson 1983; Laub and Sampson 2003). Yet, as age increases deterministically over time, it is intrinsically difficult to disentangle maturation from event-related change: the “after” of any event is also an “after” in terms of age (Angrist and Pischke 2009). In the case of incarceration, researchers also need to account for individuals’ “spurts” of activity prior to confinement (e.g., Tahamont et al. 2015; Wermink et al. 2013). Thus, the aggregate postoffending decline documented in this and prior studies using pre–post comparisons (e.g., Bottoms and Shapland 2011; Hickert et al. 2021) may reflect a combination of aging, regression to the mean (due to the preprison variation in lambda), and any deterrent or rehabilitative impact of incarceration.

While we recognize the substantive challenge of separating aging from behavioral change due to incarceration (Angrist and Pischke 2009), we consider several approaches to better understand this association. Within the DGTM, we re-estimated trajectories including age as a predictor of group membership both prior to incarceration and postrelease. Age was negatively correlated with membership in higher-offending groups, but assignment remained the same; thus, we report models without age for parsimony. In the DGTM, age at release was not associated with transition into specific offending trajectories after incarceration, suggesting that the association is at least partially captured through preprison trajectories. Finally, we estimated a Mundlak within–between decomposition of age in a piecewise growth-curve model that includes pre- and postincarceration time slopes (Appendix Table A9). The within-person (maturation) component of age is negative and sizable (≈−0.14 per semester), while the between-person (cohort) component is small and becomes nonsignificant once pretrajectory heterogeneity is included (model 5). Importantly, decomposing age (models 3 and 5) attenuates the postincarceration slope, which is consistent with a maturation-driven explanation of the aggregate postrelease decline. Because age and time-since-release are inherently confounded, this analysis does not allow us to identify an incarceration-specific effect distinct from aging, but it suggests that the postrelease decline in the models may reflect a maturation effect and that aging matters above and beyond between-person age differences.

Discussion

This study contributes to the literature on turning points by examining incarceration as an event that can reshape individuals’ criminal trajectories. Our research aligns with the original conceptualization of turning points as changes in long-term patterns of behavior (Abbott 1997; Elder 1985) and follows recent studies in emphasizing the importance of analyzing postrelease offending in relation to preincarceration trajectories (e.g., Frisch 2018; Hickert et al. 2021). Our analysis expands on prior research by focusing on justice-impacted women and explicitly considering the contingent nature of change on individuals’ preincarceration offending patterns. Additionally, we employ this framework to explore how specific elements of the prison experience may help explain variation in postrelease outcomes.

We observe heterogeneous responses to incarceration among the women in our sample. Although continuity is the most common pattern, when change occurs, it is more often toward desistance than escalation. The magnitude of continuity versus change varies by method, with cohort-relative classifications (DGTM) emphasizing continuity even in the context of absolute decline; but, taken together, the different approaches provide a plausible range for continuity and change. The results also confirm that incarceration does not work as a unifying event in life trajectories. Instead, it is associated with increased offending for some and desistance for others. This pattern is not surprising. In their interviews with the Glueck men, Laub and Sampson (2003) documented these same divergent patterns. Given these results, determining the relative weight of strain and deterrence seems an unfruitful endeavor, as both are likely at play. The analytical question is for whom incarceration has particular effects, and why. Importantly, these heterogeneous processes are largely masked when examining aggregate effects, which often emerge as (close to) null, as is the case in this study.

Despite differences in context, gender, and analytic strategy, the directional pattern found in this study—more movement toward desistance than escalation—aligns with the prior studies analyzing pre- to postconfinement changes in offending (Bottoms and Shapland 2011; Frisch 2018; Hickert et al. 2021). Compared with Hickert and colleagues, however, our study indicates a higher likelihood of desistance (51% vs. 32%; C-test results). This result seems surprising given the precarity of the Chilean prison system and the limited preparation for release relative to welfare-state countries with more rehabilitative systems, such as the Netherlands. If incarceration operates as a salient source of strain, we would expect escalation to be more likely in Chile.

Yet the perception of punitiveness and deprivation inevitably “depends on the challenges, opportunities, and life circumstances [incarcerated-individuals] experience in the community” (Bucerius and Sandberg 2022). Women incarcerated in Chile come from highly disadvantaged backgrounds, often facing unemployment, childcare responsibilities, and substance-abuse issues before incarceration (Larroulet et al. 2021a). As reported in other Latin American countries (Agoff, Sandberg, and Fondevila 2020), their daily struggles to support themselves and their children persist during imprisonment, and although access to interventions is limited, prisons do offer opportunities largely absent in the community (i.e., drug treatment, work training). Moreover, as Crewe, Hulley, and Wright (2017) note, incarcerated women experience similar pains to men but with greater severity and additional gendered strains (e.g., loss of family contact and support). The gendered experience of imprisonment could therefore condition the observed differences in change, generating stability in offending despite the heavily deprived experience, or making transition toward desistance more likely in response to the heightened perceived costs of offending. Further research is needed to understand when and why similar patterns are observed across contexts and genders.¹⁵

Consistent with turning points being conditional on prior behavioral patterns, our results show that preincarceration offending trajectories shape individuals’ responses to incarceration. Rather than operating as a universal catalyst for behavioral change, confinement corresponds to behavioral discontinuity for women with medium to high preincarceration offending, while representing continuity or modest change for women with lower involvement, a pattern observed across methods. This shift toward lower offending was particularly likely among those with high-offending trajectories: two-thirds transitioned to less-frequent offending trajectories after release, mainly to a medium-offending trajectory. Quantitatively, these women went from an average of 54 offenses in the five years before incarceration to an average of 17 after release. The results are at odds with research suggesting that low-risk individuals are more susceptible to postincarceration shifts (i.e., negative selection; Brand and Xie 2010; Turney and Wildeman 2015) and more consistent with ceiling effects. This concentration of declines also mirrors Wright et al.'s “social-protection” hypothesis, that life-course events tend to have stronger effects among those most prone to offend (Wright et al. 2001).

The results also suggest that the change is not radical. The criminological literature defines desistance as a process that takes time to unfold (Bersani and Doherty 2018). While turning points are events, they seem to lead to gradual rather than abrupt changes in the life course (Carlsson 2012; Nguyen and Loughran 2018). Our results illustrate this gradual process of change, as we observe partial movements between the patterns of offending, where the chances of moving from a high- to a low-offending trajectory are smaller than the chances of moving toward a medium-level trajectory. Theoretically, this gradual process may reflect the inertial character of trajectories (Abbott 1997), making abrupt changes less likely. While less prevalent in this sample, a similar pattern was observed in terms of escalation: it also tends to occur gradually.

As with any other observational study, these findings should not be interpreted as causal estimates of incarceration's effects. Our analytic strategy describes heterogeneous offending patterns in this population and links them to postrelease transitions. This approach helps reveal continuity and change that would be obscured by a single summary measure of prior history, and it shows that postrelease patterns vary across preincarceration trajectories. We view this as a necessary first step in documenting heterogeneity in responses to incarceration. Future causal work can build on these patterns to estimate incarceration's effects—and heterogeneity in those effects—more directly.

Finally, recognizing that prison is a fundamentally nonuniform experience, we examine whether specific elements of confinement were associated with postrelease transitions. We distinguish primarily between punitive exposures (with unclear expectations given the opposing predictions of deterrence and strain) and rehabilitative opportunities, which—in principle—could partially compensate for pre-existing disadvantages. Punitive experiences are mixed in terms of theoretical alignment: longer sentences and limited family contact were associated with membership in lower-offending trajectories, whereas the absence of release privileges was associated with membership in higher-offending trajectories. This association between reduced family involvement—measured through visits—and lower postrelease offending highlights the nuanced role of families in the case of women, given their greater likelihood of reporting violence within the family and experiencing pressure to co-offend by domestic partners (e.g., Lee et al. 2020; Leverentz 2014).

Among the rehabilitative indicators, only access to job opportunities while incarcerated emerges as a relatively consistent factor across models, associated with a higher likelihood of transitioning to a low-level trajectory after release. This association could partly reflect selection: women who access work opportunities may already be more motivated to desist, which could independently explain postrelease change. Indeed, additional analyses suggest that preincarceration trajectories are associated with access to specific prison experiences. Yet, among justice-impacted women in Latin America, motivation alone may not be enough to mobilize change without structural reinforcement (Rodermond et al. 2023), as their efforts to desist are often constrained by limited access to formal employment and caregiving burdens (Larroulet et al. 2023; Salisbury et al. 2018).¹⁶ While this finding also suggests that prisons can compensate for pre-existing disadvantages, it should not be interpreted as evidence that prisons are appropriate settings for delivering services and opportunities. Rather, it underscores the structural marginalization experienced by women who come into contact with the criminal justice system and the limited opportunities available in the community (Bucerius, Haggerty, and Dunford 2021). Overall, results on prison conditions are inconsistent across models, and most indicators attenuate when additional controls are added, warranting replication in future work.

Limitations

This study has additional limitations. First, the mostly null findings of prison experiences may be a consequence of the small sample, the particularities of the sample and the context, or a true lack of association between specific experiences and changes in offending. We cannot distinguish between them. Second, our analyses of confinement experiences are descriptive and exploratory. As selection into specific opportunities and experiences is likely associated with postrelease outcomes independent of prison experiences, estimated associations with postprison offending could be biased. In addition, given the sample size, we report bivariate associations and avoid adding further controls. This practical decision, however, should not be read as implying independent effects: people experience prison as a bundle of interrelated factors that call for more integrated modeling. Despite these limitations, we believe the question of why and when prison functions as a turning point toward desistance rather than escalation is theoretically relevant, and we hope future research will address it further.

Third, we recognize limitations in our analytic approach. Group based trajectory models, and DGTM by extension, are data-dependent methods. As such, the assignment to a trajectory group depends on an individual's relative position in the sample, and changes in offending are based on those group trajectories. Thus, we might underestimate individual change, particularly when confronted with aggregate declines in offending. For prison experiences, DGTM uses a constrained model that assumes the association between predictors and group membership after prison does not interact with preincarceration trajectories. This assumption can be relaxed to more directly test the idea of turning points as interactions (Wheaton and Gotlib 1997), but our sample size precludes us from doing so. Relatedly, our moderation (interaction) estimates should be interpreted cautiously, given evidence that interaction effects can be unstable when estimated with modest sample sizes (Castillo et al. 2026). The cautions commonly raised for group-based trajectory modeling—for example, the risk of reification—also apply to DGTM. We use these models to describe heterogeneity and, in the case of DGTM, to provide a richer description of continuity and change.

Finally, we acknowledge that it is not possible to disentangle maturation from incarceration-related change when examining shifts in offending, as these processes are fundamentally unidentifiable because time is inherently tied to both (Angrist and Pischke 2009). Thus, any attempt to isolate the specific effect of incarceration from age-graded behavioral decline must be interpreted with caution. As noted earlier, analyses of within-individual change conducted during periods characterized by aggregate declines in offending—whether attributable to aging or shifts in the reach of the criminal justice system—carry the risk of attributing naturally occurring reductions to a particular life event. In this context, the relative and data-driven structure of the DGTM may be particularly useful, as estimating trajectories relative to peers aging over the same period and exposed to comparable institutional conditions can help guard against misclassifying expected age-graded decline as substantive change.

Taken together, our findings underscore that incarceration operates as a conditional turning point. Rather than a uniform catalyst, its association with postrelease change varies across people and contexts, depending on what individuals bring to the experience and on conditions during confinement. The patterns we observe also indicate that behavioral change is gradual, characterized by steps rather than abrupt changes. Justice-impacted women navigate incarceration and reentry under distinct social, economic, and relational conditions. Extending the work on prison as a turning point to this understudied population helps clarify how far these conditional patterns generalize and where they diverge. Rather than asking whether prison “works,” we call for research that clarifies for whom, under what circumstances, and through which pathways change is most likely. More broadly, these patterns invite us to consider the extent to which expanding opportunities outside the carceral system could support desistance without exposing individuals to the well-documented harms of incarceration.

Supplemental Material

sj-docx-1-jrc-10.1177_00224278261456248 - Supplemental material for Prison as a Turning Point? The Conditional Nature of Incarceration in Women's Offending Trajectories

Supplemental material, sj-docx-1-jrc-10.1177_00224278261456248 for Prison as a Turning Point? The Conditional Nature of Incarceration in Women's Offending Trajectories by Pilar Larroulet, Josefina Rivas and Andrea Canales in Journal of Research in Crime and Delinquency

Footnotes

Acknowledgments

The data for this paper come from the study Reintegration, Desistance, and Recidivism Among Female Inmates in Chile. The original project was supported by San Carlos de Maipo and the Colunga Foundation, and received additional funding from the Inter-American Development Bank (IDB) and the Division on Women and Crime from the American Society of Criminology (ASC). We thank Thomas Loughran, the editorial team, and three anonymous reviewers for their comments.

ORCID iDs

Pilar Larroulet

Josefina Rivas

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Fondo Nacional de Desarrollo Científico y Tecnológico (Fondecyt Inicio, grant number 11220966).

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Pilar Larroulet is an Assistant Professor at the School of Criminal Justice. Her research examines life-course processes and the consequences of criminal justice contact, with particular attention to gendered experiences.

Josefina Rivas is a Sociologist with an MA in Sociology from the Pontificia Universidad Católica de Chile. Her research focuses on criminology, social reintegration and quantitative methods for social program evaluation.

Andrea Canales is an Associate Professor of Sociology at the Pontificia Universidad Católica de Chile. Her research examines educational and labor market inequality, with a focus on socioeconomic and gender inequalities, using quantitative methods and causal inference strategies.

References

Abbott

1997. “On the Concept of Turning Point.” Comparative Social Research 16:85‐105.

Agnew

1992. “Foundation for a General Strain Theory of Crime and Delinquency.” Criminology; An Interdisciplinary Journal 30:47‐87.

Agoff

Sandberg

Fondevila

. 2020. “Women Providing and Men Free Riding: Work, Visits and Gender Roles in Mexican Prisons.” Victims and Offenders 15(7–8):1086‐104.

Angrist

J. D.

Pischke

J. S.

. 2009. Mostly Harmless Econometrics: An Empiricist's Companion. Princeton, NJ: Princeton University Press.

Apel

2016. “The Effects of Jail and Prison Confinement on Cohabitation and Marriage.” The Annals of the American Academy of Political and Social Science 665(1):103‐26.

Apel

Horney

. 2018. “How and Why Does Work Matter? Employment Conditions, Routine Activities, and Crime Among Adult Males.” Criminology; An Interdisciplinary Journal 56(3):545‐74.

Audry

Liebling

. 2020. “The Role of Prison Staff in Shaping Carceral Experience.” Punishment & Society 22(3):351‐73.

Auty

Liebling

. 2020. “Exploring the Relationship Between Prison Social Climate and Reoffending.” Justice Quarterly 37(2):358‐81.

Belknap

. 2021. The Invisible Woman. Gender, Crime, and Justice. Thousand Oaks, CA: Sage Publications, Inc.

10.

Bergman

Fondevila

. 2021. Prisons and Crime in Latin America. Cambridge, UK: Cambridge University Press.

11.

Bersani

B. E.

Doherty

. 2018. “Desistance from Offending in the Twenty-First Century.” Annual Review of Criminology 1:311‐34.

12.

Bersani

B. E.

Laub

J. H.

Nieuwbeerta

. 2009. “Marriage and Desistance from Crime in the Netherlands: Do Gender and Socio-Historical Context Matter?”. Journal of Quantitative Criminology 25(1):3-24.

13.

Bottoms

Shapland

. 2011 “Steps Towards Desistance Among Male Young Adult Recidivists.” Pp. 43-80 in Escape Routes: Contemporary Perspectives on Life After Punishment edited by Farrall

En S.

Sparks

Maruna

Hough

. Abingdon, UK: Routledge.

14.

Brand

J. E.

Xie

. 2010. “Who Benefits Most from College? Evidence for Negative Selection in Heterogeneous Economic Returns to Higher Education.” American Sociological Review 75(2):273‐302.

15.

Broidy

Agnew

. 1997. “Gender and Crime: A General Strain Theory Perspective.” Journal of Research in Crime and Delinquency 34(3):275‐306.

16.

Broidy

Cauffman

. 2017. “The Glueck Women: Using the Past to Assess and Extend Contemporary Understandings of Women’s Desistance from Crime.” Journal of Developmental and Life-Course Criminology 3(2):102‐25. doi: 10.1007/s40865-016-0048-z

17.

Broidy

Payne

Piquero

A. R.

. 2018. “Making Sense of Heterogeneity in the Influence of Childhood Abuse, Mental Health, and Drug use on Women’s Offending Pathways.” Criminal Justice and Behavior 45(10):1565‐87.

18.

Broidy

Thompson

. 2018 “Developmental and Life-Course Findings on Women and Girls.” Pp. 624-49 in The Oxford Handbook of Developmental and Life-Course Criminology edited by Farrington

D. P.

Kazemian

Piquero

A. R.

. New York: Oxford University Press.

19.

Bucerius

Haggerty

Dunford

. 2021. “Prison as Temporary Refuge: Amplifying the Voices of Women Detained in Prison.” The British Journal of Criminology 61:519‐37.

20.

Bucerius

Sandberg

. 2022. “Women in Prisons.” Crime and Justice 51(1):137‐86.

21.

Bushway

Thornberry

Krohn

. 2003. “Desisting Offenders’ Self-Reported Reasons for Desistance.” Journal of Contemporary Criminal Justice 19(1):63‐83.

22.

Carlsson

2012. “Using ‘Turning Points’ to Understand Processes of Change in Offending: Notes from a Swedish Study on Life Courses and Crime.” The British Journal of Criminology 52(1):1‐16.

23.

Castillo

Miller

Vize

Baranger

Lynam

. 2026. “When Do Interaction/Moderation Effects Stabilize in Linear Regression?” Advances in Methods and Practices in Psychological Science 9(1):1‐14.

24.

Centro de Estudios Justicia y Sociedad. 2016. Antecedentes Para la Implementación de Programas de Intervención Especializados en Reinserción Social Para la Población Penal Adulta. Santiago, Chile: Instituto de Sociología, Pontificia Universidad Católica de Chile.

25.

Cook

P. J.

Kang

Braga

A. A.

Ludwig

O’Brien

M. E.

. 2015. “An Experimental Evaluation of a Comprehensive Employment-Oriented Prisoner re-Entry Program.” Journal of Quantitative Criminology 31:355‐82.

26.

Cortés

Grau

Rivera

. 2019 Juvenile incarceration and adult recidivism. Serie de Documentos de Trab ajo, FEN, SDT 482. https://repositorio.uchile.cl/handle/2250/168222.

27.

Crewe

Hulley

Wright

. 2017. “The Gendered Pains of Life Imprisonment.” The British Journal of Criminology 57:1359‐78.

28.

Daly

1992. “Women’s Pathways to Felony Court : Feminist Theories of Lawbreaking and Problems of Representation.” Review of Law and Women’s Studies 2:11‐52.

29.

Daza

Larroulet

. 2019. Evaluación de Sesgo en Línea de Base. Centro de Estudios Justicia y Sociedad del Instituto de Sociología, Pontificia Universidad Católica de Chile. https://justiciaysociedad.uc.cl/wp-content/uploads/2021/12/20190305_Sesgos_Linea_Base.pdf .

30.

DiPietro

S. M.

Doherty

E. E.

Bersani

B. E.

. 2018. “Understanding the Role of Marriage in Black Women’s Offending Over the Life Course.” Journal of Developmental and Life-Course Criminology 4(2):162‐87.

31.

D'Unger

A. V.

Land

K. C.

McCall

P. L.

Nagin

D. S.

. 1998. “How Many Latent Classes of Delinquent/Criminal Careers? Results from Mixed Poisson Regression Analyses.” American Journal of Sociology 103(6):1593‐630.

32.

Eggleston

Laub

Sampson

. 2009 “Group-based Trajectories in Life Course Criminology.” Pp. 187-210 in The Craft of Life Course Research. Nueva York: The Guilford Press.

33.

Elder

. 1985. Life Course Dynamics: Trajectories and Transitions, 1968–1980. Ithaca, NY: Cornell University Press.

34.

Elder

Johnson

Crosnoe

. 2003 “The Emergence and Development of Life Course Theory.” Pp. 3-19 in Handbook of the Life Course edited by Mortimer

En J.

Shanahan

. New York, NY: Springer.

35.

Fair

Walmsley

. 2024. World Prison Population List. 14th ed. London: Institute for Crime & Justice Policy Research, University of London.

36.

Frisch

2018. Framing Imprisonment as a Turning Point in the Lives of Criminally-Involved Adults. (Doctoral Dissertation). College Park: University of Maryland.

37.

Gaes

Camp

. 2009. “Unintended Consequences: Experimental Evidence for the Criminogenic Effect of Prison Security Level Placement on Post-Release Recidivism.” Journal of Experimental Criminology 5(2):139‐62.

38.

Gobeil

R. B.

2016. “A Meta-Analytic Review of Correctional Interventions for Women Offenders: Gender-Neutral Versus Gender-Informed Approaches.” Criminal Justice and Behavior 43(3):301‐22.

39.

Haggerty

Bucerius

. 2020. “The Proliferating Pains of Imprisonment.” Incarceration 1(1):1‐16.

40.

Heimer

Malone

S. E.

De Coster

. 2023. “Trends in Women's Incarceration Rates in US Prisons and Jails: A Tale of Inequalities.” Annual Review of Criminology 6(1):85‐106.

41.

Hickert

Bushway

Nieuwbeerta

Dirkzwager

. 2021. “Confinement as a Two Stage Turning Point: Do Changes in Identity or Social Structure Predict Subsequent Changes in Criminal Activity?” Criminology; An Interdisciplinary Journal 59:73‐108.

42.

Hirschi

Gottfredson

. 1983. “Age and the Explanation of Crime.” American Journal of Sociology 89(3):552‐84.

43.

Horney

Osgood

D. W.

Marshall

I. H.

. 1995. “Criminal Careers in the Short-Term: Intra-Individual Variability in Crime and its Relation to Local Life Circumstances.” American Sociological Review 60:655-73.

44.

Hyatt

J. M.

2024. “Understanding Prison Climate: New Insights from Multilevel Modeling.” Crime & Delinquency 70(1):59‐89.

45.

INDH. 2019. Estudio de las Condiciones Carcelarias en Chile. Diagnóstico del Cumplimiento de los Estándares Internacionales de Derechos Humanos en la Privación de Libertad. Santiago, Chile.

46.

Jaynes

C. M.

2020. “The Relationship Between Job Quality and Crime: Examining Heterogeneity in Treatment and Treatment Effect.” Journal of Research in Crime and Delinquency 57(6):693‐740.

47.

Jonas-van Dijk

Eichelsheim

Dirkzwager

Nieuwbeerta

. 2020. “The Effectiveness of Prison-Based Education for Incarcerated Offenders: A Meta-Analysis.” Journal of Experimental Criminology 16:309‐31.

48.

Jones

Nagin

. 2007. “Advances in Group-Based Trajectory Modeling and an SAS Procedure for Estimating Them.” Sociological Methods & Research 35(4):542‐71.

49.

King

R. D.

Massoglia

MacMillan

. 2007. The Context of Marriage and Crime: Gender, the Propensity to Marry, and Offending in Early Adulthood. Criminology 45(1): 33-65.

50.

Kirk

D. S.

Wakefield

2018. “Collateral Consequences of Punishment: A Critical Review and Path Forward”. Annual Review of Criminology 1:171-194.

51.

Kruttschnitt

Gartner

. 2003 “Women's Imprisonment.” Pp. 301-81 in Crime and Justice: A Review of Research edited by Tonry

En M.

. Chicago, IL: University of Chicago Press.

52.

Larroulet

Daza

Bórquez

. 2023. “From Prison to Work? Job-Crime Patterns for Women in a Precarious Labor Market.” Social Science Research 110:102844.

53.

Larroulet

Droppelmann

Daza

Del Villar

Figueroa

. 2021a. Reinserción, Desistimiento y reincidencia en mujeres privadas de libertad en Chile. Santiago, Chile: Centro de Estudios Justicia y Sociedad, Pontificia Universidad Católica de Chile.

54.

Larroulet

Droppelmann

Del Villar

Daza

Figueroa

Osorio

. 2020. “Who is Transitioning Out of Prison?: Characterising Female Offenders and Their Needs in Chile”. International Journal for Crime, Justice and Social Democracy 9(1):112-125.

55.

Larroulet

Loughran

Augustyn

Thornberry

Henry

. 2021b. “Intergenerational Continuity and Discontinuity in Substance Use: The Role of Concurrent Parental Marijuana Use.” Journal of Developmental and Life-Course Criminology 7(2):127‐50.

56.

Laub

Sampson

. 2003. Shared Beginnings, Divergent Lives: Delinquent Boys to Age 70. Cambridge, MA: Harvard University Press.

57.

Lee

J. G.

Gillespie

L. K.

Slocum

L. A.

Simpson

S. S.

. 2020. “Better Together? Intimate Partner Effects on Offending Women.” Journal of Developmental and Life-Course Criminology 6(4):499-523.

58.

Leverentz

A. M

. 2014. The Ex-prisoner's Dilemma: How Women Negotiate Competing Narratives of Reentry and Desistance. New Brunswick: Rutgers University Press.

59.

Listwan

Sullivan

Agnew

Cullen

. 2013. “The Pains of Imprisonment Revisited: The Impact of Strain on Inmate Recidivism.” Justice Quarterly 30(1):144‐68.

60.

Loeffler

Nagin

. 2022. “The Impact of Incarceration on Recidivism.” Annual Review of Criminology 5:133‐52.

61.

Loughran

Nagin

. 2006. “Finite Sample Effects in Group-Based Trajectory Models.” Sociological Methods & Research 35(2):250‐78.

62.

McCuish

Bushway

Lussier

Gushue

. 2025. “The Impact of Incarceration on Reoffending: A Period-to-Period Analysis of Canadian Youth Followed into Adulthood.” Journal of Criminal Justice 96:102335.

63.

McCuish

Lussier

Corrado

. 2018. “Incarceration as a Turning Point? The Impact of Custody Experiences and Identity Change on Community Reentry.” Journal of Developmental and Life-Course Criminology 4:427‐48.

64.

McCuish

E. C.

2020. “Seductions of Exposure Time: The Mismeasurement of Desistance Among Persons Involved in Frequent and Serious Offending.” Canadian Journal of Criminology and Criminal Justice 62(3):29‐50.

65.

McGloin

J. M.

Augustyn

M. B.

Ward

J. T.

. 2025. “Moving Beyond the Average Gang Membership Effect: Investigating Treatment Effect Heterogeneity in Concurrent and Long-Term Outcomes.” Journal of Research in Crime and Delinquency 62(4):597‐655.

66.

Morales

Hurtado

Figueroa

Ortiz

Polanco

Muñoz

. 2016. Estudio Sobre los Niveles de Exclusión Social en Personas Privadas de Libertad. Santiago: Fundación Paz Ciudadana y Fundación San Carlos de Maipo.

67.

Nagin

2005. Group-Based Modeling of Development. Cambridge: Harvard University Press.

68.

Nagin

2013. “Deterrence in the Twenty-First Century.” Crime and Justice 42(1):199‐263.

69.

Nagin

Cullen

Lero Jonson

. 2009. “Imprisonment and Reoffending.” Crime and Justice 38:115‐200.

70.

Nagin

Odgers

. 2010. “Group-Based Trajectory Modeling in Clinical Research.” Annual Review of Clinical Psychology 6:109‐38.

71.

National Academies of Sciences, Engineering, and Medicine. 2022. The Limits of Recidivism: Measuring Success After Prison. Washington, DC: The National Academies Press.

72.

Nguyen

Loughran

. 2018. “On the Measurement and Identification of Turning Points in Criminology.” Annual Review of Criminology 1:335‐58.

73.

Nguyen

Loughran

T. A.

Paternoster

Fagan

Piquero

A. R.

. 2017. “Institutional Placement and Illegal Earnings: Examining the Crime School Hypothesis.” Journal of Quantitative Criminology 33(2):207‐35.

74.

Nieuwbeerta

Nagin

Blokland

. 2009. “Assessing the Impact of Imprisonment on Recidivism.” Journal of Quantitative Criminology 35(2):227‐43.

75.

Nur

A. V.

Nguyen

. 2023. “Prison Work and Vocational Programs: A Systematic Review and Analysis of Moderators of Program Success.” Justice Quarterly 40(1):129‐58.

76.

Owen

B. A

. 1998. In the Mix: Struggle and Survival in a Women's Prison. Albany, NY: Suny Press.

77.

Pager

2003. “The Mark of a Criminal Record.” American Journal of Sociology 108(5):937‐75.

78.

Paternoster

Bushway

. 2009. “Desistance and the ‘Feared Self’: Toward an Identity Theory of Criminal Desistance.” Journal of Criminal Law and Criminology 99(4):1103‐56.

79.

Przyborowski

Wilenski

. 1940. “Homogeneity of Results in Testing Samples from Poisson Series: With an Application to Testing Clover Seed for Dodder.” Biometrika 31(3/4):313‐23.

80.

Rodermond

Van De Weijer

Rosenkrantz Lindegaard

Bijleveld

C. C.

Slotboom

A. M.

Kruttschnitt

. 2023. “Out of Prison, Out of Crime? The complex Interplay Between the Process of Desistance and Severe Resource Disadvantages in Women’s Post-Release Lives.” European Journal of Criminology 20(6):1852‐77.

81.

Rufs

Larroulet

Valenzuela

. 2023. “Encarcelamiento Materno y Problemas de Comportamiento en Niños, Niñas y Jóvenes en Chile.” Revista Española de Investigación Criminológica 21(1):1‐23.

82.

Salisbury

E. J.

Kalantry

Boppre

Brundige

Martínez

. 2018. “Expanding the Feminist Pathways Perspective Beyond the United States: A Profile of Federal Women Prisoners in Argentina.” Women & Criminal Justice 28(2):125‐51.

83.

Sampson

Laub

. 1990. “Crime and Deviance Over the Life Course: The Salience of Adult Social Bonds.” American Sociological Review 55(5):609‐27.

84.

Sampson

R. J.

Laub

J. H.

Eggleston

E. P.

. 2004. On the Robustness and Validity of Groups. Journal of Quantitative Criminology 20(1):37-42.

85.

Sampson

R. J.

Laub

J. H.

Wimer

. 2006. “Does Marriage Reduce Crime? A Counterfactual Approach to Within-Individual Causal Effects.” Criminology; An Interdisciplinary Journal 44(3):465‐508.

86.

Sanhueza

Brander

Reiser

. 2019. “Encarcelamiento Femenino en Chile. Calidad de Vida Penitenciaria y Necesidades de Intervención.” Revista de Ciencias Sociales 32(45):119‐45.

87.

Singer

J. D.

Willett

J. B.

. 2003. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York, NY: Oxford University Press.

88.

Slocum

Simpson

Smith

. 2005. “Strained Lives and Crime: Examining Intra-Individual Variation in Strain and Offending in a Sample of Incarcerated Women.” Criminology; An Interdisciplinary Journal 43(4):1067‐110.

89.

Slocum

L. A.

Medel

Doherty

E. E.

Simpson

S. S.

. 2022. “Unpacking the Criminogenic Aspects of Stress Over the Life Course: The Joint Effects of Proximal Strain and Childhood Abuse on Violence and Substance Use in a High-Risk Sample of Women.” Journal of Research in Crime and Delinquency 59(6):699‐755.

90.

Steiner

Makarios

Travis

. 2017. “Examining the Link Between Institutional Conditions and Recidivism.” Justice Quarterly 34(6):1094‐119.

91.

Sykes

. 1958. The Society of Captives: A Study of a Maximum Security Prison. Princeton, NY: Princeton University Press.

92.

Tahamont

Yan

Bushway

S. D.

Liu

. 2015. “Pathways to Prison in New York State.” Criminology & Public Policy 14(3):431‐53.

93.

Travis

Western

Redburn

. 2014. The Growth of Incarceration in the United Stated: Exploring Causes and Consequences. Washington, D.C.: The National Academies Press.

94.

Turney

Wildeman

. 2015. “Detrimental for Some? Heterogeneous Effects of Maternal Incarceration on Child Wellbeing.” Criminology & Public Policy 14(1):125‐56.

95.

van Ginneken

Palmen

. 2023. “Is There a Relationship Between Prison Conditions and Recidivism?” Justice Quarterly 40(1):106‐28.

96.

Wakefield

Uggen

. 2010. “Incarceration and Stratification.” Annual Review of Sociology 36:387‐406.

97.

Wermink

Apel

Nieuwbeerta

Blokland

A. A.

. 2013. “The Incapacitation Effect of First-Time Imprisonment: A Matched Samples Comparison.” Journal of Quantitative Criminology 29(4):579‐600.

98.

Western

2006. Punishment and Inequality in America. New York, NY: Russell Sage Foundation.

99.

Wheaton

Gotlib

. 1997 “Trajectories and Turning Points Over the Life Course: Concepts and Themes.” Pp. 1‐26 in Stress and Adversity Over the Life Course, edited by Gotlib

En I.

Wheaton

. New York: Cambridge University Press.

100.

Wright

Caspi

Moffitt

T. E.

Silva

P. A.

. 2001. “The Effects of Social Ties on Crime Vary by Criminal Propensity: A Life-Course Model of Interdependence.” Criminology; An Interdisciplinary Journal 39(2):321‐48.

101.

Zweig

Yahner

Visher

Lattimore

. 2015. “Using General Strain Theory to Explore the Effects of Prison Victimization Experiences on Later Offending and Substance Use.” The Prison Journal 95:113‐84.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB