Optimism and Obstacles: Racialized Constraints in College Attitudes and Expectations among Teens of the Prison Boom

Abstract

Using data from the Future of Families and Child Wellbeing Study, we estimate associations of paternal incarceration with three measures of teens’ attitudes and expectations: (1) optimism about the future, (2) perceived importance of college graduation, and (3) perceived likelihood of college graduation. Results suggest that whereas optimism toward the future and the importance of college are resilient in the face of paternal incarceration, teens’ expectations of actual college completion are reduced, particularly for Black youth. An examination of micro-, meso-, and macro-level mechanisms suggests that early consequences of paternal incarceration influence perceptions of future educational trajectories. Our findings point to the complexity of contemporary teens’ college-related attitudes in the wake of the prison boom. Signs of optimism surfaced, yet obstacles associated with paternal incarceration represent a type of racialized constraint. An updating of the educational expectations literature is needed to account for the racialized nature of the era of mass incarceration in the United States.

INTRODUCTION

The U.S. carceral state ballooned from 1970 to 2010, affecting individuals, families, and communities alike (Travis, Western, and Redburn 2015). Today, nearly 1 in 2 Americans report having experienced the incarceration of an immediate family member (Enns et al. 2019). This includes more than two million children with a currently incarcerated parent (Sykes and Pettit 2014), and about 10 million children whose parent has ever been incarcerated (Underwood 2017). These aggregate statistics mask stark disparities across racial and ethnic groups: scholars estimate that 1 in 4 Black children, 1 in 10 Latinx children, and 1 in 25 White children experience parental incarceration by their early teens (Wildeman 2009), perpetuating patterns of childhood inequality (Wakefield and Wildeman 2014).

In considering the ways this expansion of the criminal legal system has affected families and children, a growing literature examines the effects of a father’s incarceration on educational outcomes consequential for children’s prospects for social mobility. This work primarily focuses on fathers because 93 percent of incarcerated parents are men (Maruschak, Glaze, and Mumola 2010). These studies document the range of academic problems (e.g., cognitive, behavioral) that children with incarcerated fathers face in early and middle childhood (e.g., Haskins 2014, 2016; Jacobsen 2019; Turney and Haskins 2014; Wildeman 2010). A less-developed body of research examines longer-term outcomes by quantitatively assessing the effects of parental incarceration on young adult educational attainment (Hagan and Foster 2012; Mears and Siennick 2016).

This quantitative body of evidence documents the measurable consequences of paternal incarceration, yet it tells us little about how youth understand the meaning of incarceration in their own lives. In contrast, prior qualitative work focuses on coping strategies and psychological well-being after parental incarceration (Arditti and Johnson 2022; Johnson and Easterling 2015; Kautz 2017). These studies demonstrate that over and above parental incarceration’s immediate effects on more standard life-course outcomes, the stigma, stress, and traumatic loss of incarceration may more fundamentally (re)shape children’s sense of possibility and understanding of self. These often-unmeasured effects may weigh most heavily on youth of color, who are at greatest risk of experiencing paternal incarceration.

In this study, we consider these consequences in the context of a process long central to the sociology of education: the formation of attitudes, beliefs, and expectations around schooling. In particular, we analyze adolescent expectations about educational attainment, a key topic of inquiry ever since the Wisconsin model of status attainment placed expectations at its “strategic center” (Haller and Portes 1973:68). Educational expectations are empirically consequential, predicting effort in high school (Domina, Conley, and Farkas 2011), ultimate attainment (Jacob and Wilder Linkow 2011; Morgan 2004), and later life success (Morgan 2004). More fundamentally, educational expectations are deeply imbued with meaning (Deterding 2015; Frye 2012), reflecting adolescents’ observations about the opportunity structure of society and an internalized sense of what is possible for “someone like me” (MacLeod 1995).

Our objectives are threefold. First, we extend theory on the collateral consequences of punishment for educational outcomes by considering postsecondary expectations as an outcome of paternal incarceration. To situate adolescents’ educational expectations within the broader context of their beliefs and attitudes, we also examine two other outcomes: optimism about the future and beliefs about the importance of college. Second, we assess the extent to which the association of paternal incarceration with college expectations varies across racial groups. For youth of color, educational expectations may represent not only their hopes for social mobility, but also the realities and raced nature of social barriers like mass incarceration. Finally, after finding group-specific disadvantages for Black teens in the sample, we assess the potential mechanisms—micro, meso, and macro—through which paternal incarceration shapes educational expectations among Black youth.

Background

A Life-Course Cumulative Disadvantage Framework

As today’s students progress through school, many must grapple with a factor not considered in earlier accounts of the status attainment process: parental incarceration. Existing research documents the wide-reaching effects of parental incarceration on student performance and educational attainment. We conceptualize diminished attainment as a form of social exclusion, or weakening of the social bond. To understand the processes by which legal-system involvement results in social exclusion, prior research relies on a life-course perspective of cumulative disadvantage (Foster and Hagan 2015; Sampson and Laub 1997). This perspective emphasizes the disproportionate concentration of legal-system involvement among disadvantaged populations, as well as the role of system involvement in the accumulation of even greater disadvantage, over the life course and from parent to child (Jacobsen 2019).

In the context of parental incarceration, this cumulative disadvantage is driven by social-psychological factors such as the stigma and stress of having a parent behind bars (Braman 2004; Turney 2014). Such factors interfere with children’s normative socio-emotional development, resulting in increased behavior problems, lower school performance, and exclusionary reactions by others. Indeed, paternal incarceration is associated with lower school readiness and cognitive ability (Haskins 2014, 2016), special education placement (Haskins 2014), increased behavior problems (Haskins 2015; Wildeman 2010), grade retention (Turney and Haskins 2014), and a greater risk of being suspended or expelled from school (Jacobsen 2019). These educational disadvantages are compounded with new challenges as children enter adolescence and young adulthood, when parental incarceration is associated with school absence (Nichols and Loper 2012), reduced performance (Foster and Hagan 2007), and a lower likelihood of high school or college completion (Hagan and Foster 2012; Mears and Siennick 2016).

This framework is a useful starting point for understanding the consequences of paternal incarceration for educational attainment; however, few studies have explored the social-psychological processes within this framework that likely drive long-term outcomes, nor whether such processes operate in similar ways across racial groups. For example, recent research suggests the stigma of incarceration lowers perceptions of one’s own social status, especially for African Americans (Schnittker and Bacak 2013), and that having a formerly incarcerated relative negatively affects others’ perceptions of a person’s parenting quality, deservingness, and personality traits (Brew et al. 2022). Given the “stickiness” of stigma (Braman 2004; Goffman 1963), it is important to consider how the incarceration of a parent may affect an adolescent’s own perceptions. Of particular interest is whether the incarceration of a father affects the formation of attitudes, beliefs, and expectations around schooling. This research is particularly needed for adolescents, who are better equipped than younger children to understand the complexities of a parent’s incarceration (Shlafer and Poehlmann 2010), including how it is perceived by others and the social barriers it may pose (Arditti and Johnson 2022; Johnson and Easterling 2015). Adolescents may also have a heightened knowledge of and sensitivity to paternal incarceration, including the stigma, traumatic loss, and economic instability it creates. More broadly, the formation of attitudes and beliefs is an important part of adolescent development that prior quantitative research on parental incarceration has rarely considered.

Educational Expectations from the Perspective of the Wisconsin Model

Adolescents’ educational expectations have been a regular focus of sociological inquiry ever since the introduction of the Wisconsin model of status attainment (Sewell, Haller, and Ohlendorf 1970; Sewell, Haller, and Portes 1969). The Wisconsin model placed expectations squarely in the center of the causal chain linking social origins to destinations, arguing that “practically all the effect of family’s position on educational and occupational attainments is due to its impact on the formation of status aspirations” (Haller and Portes 1973:62). In this conceptualization, expectations matter because they serve as motivational resources, capable of propelling students forward through their effect on academic effort and investment.

In the decades since its articulation, the Wisconsin model’s assertion that educational expectations independently affect students’ efforts and outcomes has faced sustained scrutiny. The ongoing importance of expectations seems particularly uncertain given the attitudes of what scholars call the “ambitious generation” (Schneider and Stevenson 1999). Since the 1970s, high school students’ expectations of college completion have risen rapidly (Goyette 2008). In 1976, only 52 percent of high school seniors expected to complete a bachelor’s degree, but by 2006, this figure had risen to over 80 percent. Nevertheless, actual college completion has failed to keep pace. Degree attainment for men, and African American men in particular, has remained roughly constant since the 1970s (Jacob and Wilder Linkow 2011). This growing gap between expectations and attainment challenges the assertion that expectations still guide current action and predict future success (Reynolds et al. 2006; Schneider and Stevenson 1999). Recent evidence confirms that the predictive power of expectations has, in fact, weakened across cohorts (Domina et al. 2011) and suggests expectations alone are not enough to guide youth through the challenges of adolescence (Harris, Duncan, and Boisjoly 2002).

And yet, even in a context of weakening predictive power and widespread postsecondary expectations, one’s own expectations still independently predict a wide range of consequential outcomes: high school effort (Domina et al. 2011), GPA and course-taking patterns (Jacob and Wilder Linkow 2011), ultimate degree attainment (Feliciano and Rumbaut 2005; Lundberg 2020; Morgan 2004), and health behaviors (McDade et al. 2011). Indeed, even studies arguing that researchers have overstated the causal effects of expectations still show that expectations maintain an independent, statistically significant effect on attainment (Fishman 2019). Once formed, expectations appear to be fairly stable, declining only in response to large shifts in academic performance (Andrew and Hauser 2011). Given all of these facts, Domina and colleagues (2011:109) conclude that “half a century after its initial formulation, the Wisconsin model continues to accurately describe the relationship between student expectations and effort.”

Even beyond their role in shaping other academic outcomes, educational expectations are theoretically rich constructs in and of themselves. As Kerckhoff (1976) and Bourdieu and Passeron (1977) argue in their influential critiques of the Wisconsin model, expectations reflect more than goals. They are observations about the opportunity structure of society. In this account, expectations predict outcomes because they reveal a sense of possibility or constraint left unmeasured by traditional indices of (dis)advantage. Thus, given the centrality of college in facilitating access to secure, middle-class employment in the United States—both in fact (Autor, Katz, and Kearney 2008) and in popular imagination (Deterding 2015; Rosenbaum 2001)— college expectations are fraught with meaning, not simply estimations of probabilities.

Paternal Incarceration and Educational Expectations: Our Conceptual Model

Taken together, these facts suggest paternal incarceration could consequentially shape students’ postsecondary expectations, both directly and indirectly. Directly, qualitative research from developmental psychology and family studies highlights that adolescents use various cognitive coping mechanisms and adaptation strategies to help them navigate the stigma and other stressors brought on by parental incarceration. These include efforts to regulate thoughts, feelings, and behaviors, and attempts to control environmental circumstances (Compas and Reeslund 2009). This work finds that children of incarcerated parents often adapt by maintaining “strength through control”—a type of internal control strategy that highlights restructuring cognitive actions, behaviors, and beliefs about self to allow for the best fit for one’s current conditions (Johnson and Easterling 2015).

Thus, if personal exposure to familial incarceration restructures students’ “cognitive landscapes and perceived pathways to educational and other attainments,” as Hagan and Foster (2012:261) suggest, then paternal incarceration should constrain the educational expectations of youth, and potentially uniquely youth of color. As such, educational inequality research needs to examine effects of parental incarceration on educational beliefs and attitudes, not only on more distal outcomes that attitudes predict (e.g., GPA, school completion, college enrollment). Thus, extending our cumulative disadvantage framework to account for social-psychological processes, teens may internalize their stigmatized family position in the formation of educational expectations, and this may be one avenue through which parental incarceration results in lower attainment and other markers of social exclusion.

Additionally, paternal incarceration may shape students’ postsecondary expectations indirectly through the accumulation of disadvantage in childhood—manifested as challenges children face in school as a result of their father’s incarceration (e.g., grade retention, special education placement, lower parental involvement, increased behavior problems, and exclusionary discipline). These consequences may occur early in children’s schooling and result in greater consequences in adolescence (e.g., high school achievement and engagement) that may further limit their perceptions of the likelihood that someone like them would finish college.

Beyond these individual (micro-level) mechanisms, there are also small group processes (meso-level) and school- or neighborhood-level factors (macro-level) to consider. In terms of meso-level processes, the Wisconsin model highlights the role of “significant others’ influence” on educational expectations. These include “specific persons . . . [who] communicate [to the youth] their expectations” and who serve as models to the youth (e.g., peers, parents, and other adults; Sewell et al. 1969:84), thus shaping adolescents’ own expectations for attainment. Youth who are at risk of paternal incarceration are already embedded in social networks that are more disadvantaged in terms of social, cultural, and other forms of capital. However, the incarceration of a parent appears to further constrain these networks, resulting in exclusion from higher-performing peers (Bryan 2017) and weakened relationships with adults (Turney and Wildeman 2013), which may hinder the development of expectations. In terms of macro-level factors, paternal incarceration is often accompanied by broader “packages of risk” (Giordano and Copp 2015), that is, contexts of disadvantage that may confound the association between paternal incarceration and college expectations. These include school and neighborhood disadvantage (Haskins 2017; Haskins and McCauley 2018) and teens’ own police contact. All of these co-occur with and potentially shape the development of educational expectations in adolescence.

Racial Heterogeneity and Adolescence

These questions demand attention for all youth, but they are particularly urgent for youth of color. Black and Latinx teens face many social barriers, including discrimination and structural racism, that their White peers do not experience. For youth of color, vulnerabilities “pile up faster” (Sampson and Laub 1997:21), adding to the accumulation of disadvantage hypothesized by Hagan and Foster (2012). In such a context, college graduation can appear less attainable. A body of sociological research points to this phenomenon as a problem of leveled aspirations (e.g., Bell 2019; MacLeod 1995; Willis 1977). Although educational expectations are generally stable, the expectations expressed by Black and Latinx youth and youth who grow up in disadvantaged contexts exhibit more instability across the life course. These youth often start with high aspirations, then shift their expectations to more closely align with perceived realistic opportunities based on their social position (Berzin 2010; Bozick et al. 2010; Cheng and Starks 2002; Kao and Tienda 1998; MacLeod 1995; Willis 1977; Young 2006).

Moreover, research in communities of color has hypothesized that experiencing traumatic loss, like that of losing a parent to incarceration, lowers expectations in the transition to adulthood (Smith 2014; Young 2006). Beyond these acute traumas, lower expectations for youth of color are likely informed by the lived experiences of adults in their communities who have faced barriers to employment and higher education based on their race and criminal legal history (Pager 2003; Stewart and Uggen 2020). Thus, for these youth who already face social and economic disadvantage, paternal incarceration fosters a greater accumulation of “continuing negative consequences for later development,” restricting “future options in conventional domains” such as college (Sampson and Laub 1997:21).

Despite all these reasons to question the value of academic effort, research demonstrates that Black students remain deeply invested in the American “achievement ideology” that positions schools as the primary means for upward mobility (Mickelson 1990; Nichols et. al 2010). Contrary to the prediction that minoritized students’ structural exclusion and oppression should cause them to devalue education (e.g., Ogbu 1978), sociologists consistently find that—conditional on family socioeconomic status—Black (and Latinx, to some extent) students actually report more positive attitudes about school and higher education expectations (e.g., Ainsworth-Darnell and Downey 1998; Diamond and Huguley 2014; Feliciano and Lanuza 2016; Morgan 1996; Reynolds and Pemberton 2001). Amid realities like fewer resources, differential discipline, racism and discrimination, and mass incarceration (Bell 2019; Ipsa-Landa and Conwell 2014), students of color express “optimism in the face of despair” (Harris 2008).

This background yields competing predictions about the effects of paternal incarceration on Black and Latinx students’ educational expectations. Given the accumulation of disadvantage already faced by students of color, paternal incarceration could be particularly consequential for their expectations for the future. At the same time, the broader optimism consistently reported by Black and Latinx teens suggests they may be more buffered against the effects of paternal incarceration than their White counterparts.

Current Study

In this article, we advance an understanding of the consequences of paternal incarceration for youth attitudes and expectations, especially youth of color, as they navigate the transition to adulthood. Specifically, we use data from the Future of Families and Child Wellbeing Study to examine the association between paternal incarceration and adolescents’ college expectations. To situate teens’ college expectations within the broader context of their beliefs, we also examine two supplementary outcomes: adolescents’ optimism about the future and their beliefs about the importance of college. Including these two additional outcome measures allows us to test the sensitivity of our results. By examining the association between paternal incarceration and optimism about the future, we are accounting for the possibility that our results for college expectations are driven by broader beliefs about the future rather than narrow assessments of college itself. By testing the association between paternal incarceration and teens’ beliefs about the importance of college, we contrast adolescents’ beliefs about their likelihood of college graduation with their reports of how important it is that they graduate.

Next, given the stark racial disparities in both paternal incarceration and educational outcomes in adolescence, and the fact that stigmatization experiences related to the legal system may be greater for youth of color than for White youth (Pager 2003; Schnittker and Bacak; Stewart and Uggen 2020), we assess whether associations vary by race. Finally, we test the extent to which three groups of mechanisms—micro (individual academic factors), meso (significant others’ influence), and macro (contextual disadvantage)—play a role in explaining associations between paternal incarceration and teens’ perceived higher education trajectories.

Data and Methods

The Future of Families and Child Wellbeing Study

The Future of Families and Child Wellbeing Study (FFCWS)¹ is a cohort of nearly 5,000 children born in large, U.S. cities between 1998 and 2000. Unmarried parents were oversampled, making the data over-representative of disadvantaged urban families. Since the baseline data-collection effort, data have been collected at five additional waves. Conducted at approximately one (Y1), three (Y3), five (Y5), nine (Y9), and fifteen years (Y15) after the child’s birth, these waves each utilized a variety of data collection measures, including detailed surveys of parents and children, home visits and observations, and cognitive and medical tests. With its detailed incarceration histories and rich developmental data on children and families, FFCWS is a primary data source for studies of parental incarceration. The current study uses all six waves, including the most recent Y15 follow-up. Y15 includes interviews with both the teen and the primary caregiver (PCG), and home visits for a random subsample of families. Data collection occurred between winter 2014 and fall 2016.

Analytic Sample

Of the 4,898 children in FFCWS at baseline, 3,444 completed a youth survey at Y15. Much of this sample attrition occurred in the early years of the survey (see Table A1 in the online supplement) and was most extensive among immigrant families. The retention rate between Y9 and Y15 was over 90 percent. Of these participating youth, we remove cases in which the father was unknown or deceased leading up to Y1 (n = 28) or the youth had missing data on any of our outcome measures (n = 18). We also dropped cases for whom the father’s incarceration history was unknown (n = 193). These changes resulted in an analytic sample of N = 3,205. Of the more than 70 variables used in our analyses, only five are missing more than 10 percent of observations. The variable with the most missing data (18 percent) assesses the father’s cognitive ability. We retain records with missing values using multiple imputation with chained equations (White, Royston, and Wood 2011) and combine results across 20 multiply imputed datasets using Rubin’s (1987) rules. The analytic sample is summarized in Table 1, and these descriptive statistics are broken out by race and incarceration status in Table A1 in the online supplement.

Table 1.

Descriptive Statistics for Variables Used in Analyses

Variable	Full Sample
	Mean	SD
Outcome Variables
Perceived likelihood of college graduation	0.650	0.477
Importance of college graduation	0.854	0.353
Optimistic about future	3.407	0.497
Explanatory Variable
Father incarcerated between Y1 and Y9	0.357	0.479
Control Variables
Early paternal incarceration	0.342	0.474
Father lives with child Y1	0.696	0.460
Father deceased	0.101	0.301
Mother criminal charges	0.094	0.292
Father impulsivity (z-scores)	0.000	0.977
Mother impulsivity (z-scores)	0.003	0.977
Father substance use	0.153	0.360
Mother violent victimization by father	0.115	0.320
Mother and father married	0.242	0.429
Mother and father cohabiting	0.359	0.480
Mother and father very good relationship quality	0.473	0.499
Mother reads to child	0.319	0.466
Mother or father not a U.S. citizen	0.136	0.343
Father age (16 to 67 years)	28.343	7.226
Mother lived with both parents at age 15	0.422	0.494
Father lived with both parents at age 15	0.447	0.497
Mother postsecondary education	0.365	0.482
Father postsecondary education	0.321	0.467
Father cognitive ability (z-scores)	−0.022	0.997
Mother cognitive ability (z-scores)	0.022	1.001
Mother religious attendance	0.386	0.487
Mother depression	0.153	0.360
Father unemployed	0.263	0.440
Mother unemployed	0.217	0.413
Mother income-to-poverty ratio (0 to 36.1)	1.847	2.231
Mother children in household (0 to 9)	2.304	1.304
Mother multi-partner fertility	0.360	0.480
Child Black	0.544	0.498
Child Hispanic	0.232	0.422
Child male	0.513	0.500
Child age in months	186.959	8.060
Child low birth weight	0.093	0.291
Child poor health Y1	0.335	0.472
Y9 Mediators
Academic
Home-based parental involvement in schooling	3.167	0.701
School-based parental involvement in schooling	0.905	0.451
Special education	0.101	0.301
Grade retention	0.161	0.368
Behavioral
Externalizing behavior problems	−0.013	0.981
Suspension or expulsion	0.183	0.387
Y15 Mechanisms
Micro-level
High School Involvement
Grade point average	2.853	0.671
School attachment	0.390	0.331
Meso-level
Significant Others’ Influence
Talked with teacher/coach about college	0.317	0.465
Talked with counselor/other about college	0.489	0.500
College graduation important to parent	0.917	0.276
Talked with parent about college	0.364	0.481
Most classmates likely to graduate college	0.327	0.469
Most friends likely to graduate college	0.461	0.499
Macro-level
Contextual Factors
Neighborhood disadvantage	−0.005	0.817
School majority students of color	0.551	0.498
School majority free or reduced-price lunch	0.484	0.500
Police contact	0.268	0.443
N	3,205

Data Source: Future of Families and Child Wellbeing Study.

Note: Sample limited to cases in which the father was living and known to the mother by Y1 and for which valid data were provided for youth attitudes and paternal incarceration. Twenty sample-city dummy variables are not shown for parsimony. Results are based on the first of 20 multiply imputed datasets. Y1 = year-one survey, Y9 = year-nine survey.

Measures

Outcome variables.

We constructed all of our outcome variables using questions from the Y15 survey. Our primary measure, college expectations, comes from the question, “How likely are you to graduate from college?” Options include “very likely,” “somewhat likely,” “not very likely,” and “very unlikely.” Despite the disadvantaged nature of our sample, the distribution of responses is consistent with prior research. For example, 70 percent of 15- and 16-year-olds in the 1997 cohort of the National Longitudinal Survey of Youth believed their chances of graduating from college by age 30 were greater than 50/50, and 57 percent believed their chances were greater than 75 percent (Reynolds and Pemberton 2001). Similarly, teens in our sample expressed high expectations; about 65 percent said they were “very likely” to graduate from college, compared to less than 3 percent who answered they were “very unlikely” or “somewhat unlikely.” Given these response patterns, we constructed a dichotomous variable taking a value of one if the teen responded that college graduation is “very likely” and zero otherwise.

To situate teens’ expectations within the context of their broader attitudes toward college and their futures more generally, we analyze two supplemental outcomes. First, we contrast teens’ beliefs about how likely they are to graduate from college with their report of how important it is to graduate. Specifically, teens were asked, “How important is it to you that you graduate from college?” Notably, this question focuses on the importance of college to the respondent, not in society generally, capturing their personal investment in the “achievement ideology” assessed in prior research (e.g., Harris 2008). As with the likelihood measure, teens overwhelmingly responded that graduating was “very important” (85 percent), so we constructed a dichotomous measure of whether the teen answered “very important” as opposed to “somewhat important” (13 percent) or “not very” (2 percent). Second, we account for the possibility that expectations about college reflect beliefs about the future broadly, rather than college itself. For this, we measure adolescents’ optimism about the future using the mean of four items ranging from 1 = “strongly disagree” to 4 = “strongly agree”: “I think good things are going to happen to me,” “I believe that things will work out, no matter how difficult they seem,” “In uncertain times, I expect the best,” and “I am optimistic about my future.”

Explanatory variable.

Our primary explanatory variable is the incarceration status of the teen’s biological father. We utilize a summary measure of whether the father has ever been incarcerated, constructed using direct reports (e.g., “Has father spent any time in jail since the last interview?”) and indirect reports (e.g., “What was father doing most of last week?”). Consistent with prior FFCWS research, we consider the father to have been incarcerated if either the mother or father reports incarceration. For our purposes, we focus on incarcerations that occurred during the child’s life. Given that incarceration was not asked about directly at baseline, we measure incarcerations that occurred between Y1 and Y9. We do not include incarceration by Y15 in order to establish the appropriate temporal order with our mediating variables (Y9), and because few new incarcerations occurred between Y9 and Y15 (see Figure A1 in the online supplement).

Control variables.

To model the relationship between paternal incarceration and our outcomes, we use the extensive data that FFCWS offers on children’s contexts and potential correlates of incarceration. We focus on indicators that prior research has highlighted as risk factors of paternal incarceration (e.g., Turney and Wildeman 2013; Wildeman 2009), such as fathers’ substance abuse, domestic violence, and impulsivity. We account for early correlates of college expectations, including mother’s and father’s postsecondary attainment, cognitive ability, and reading to the child at Y1. We also control for many other characteristics related to child health and development, family context, and the mother’s criminal history. Finally, we control for sample city and the parents’ marital status at baseline to account for the oversampling of unmarried parents. The full list of control variables is presented in Table 1.

Mediators and other mechanisms.

To better understand the association between paternal incarceration and college expectations, we include 18 variables measured at Y9 (end of observation period for paternal incarceration) or Y15 (year at which outcome variables are observed). We conceptualize Y9 mediators as variables that, based on our theoretical framework, are early outcomes of paternal incarceration that may partially explain an association with lower college expectations at Y15. We refer to Y15 mechanisms as later outcomes of paternal incarceration and other correlates of incarceration at micro-, meso-, and macro-levels that co-occur with lower college expectations. These are the contemporaneous pathways that prior literature suggests are associated with postsecondary expectations and outcomes.

In stepwise fashion, the Y9 mediators include academic characteristics such as home-based and school-based parental involvement in schooling (Haskins and Jacobsen 2017), special education placement (Haskins 2014), and grade retention (Turney and Haskins 2014), alongside behavioral characteristics like externalizing behavior problems (Wildeman 2009) and school discipline (Jacobsen 2019). Our Y15 mechanisms include a measure of high school GPA (Hagan and Foster 2012) and school attachment as a proxy for integration and connectedness (Bryan 2017). We also include meso-level indicators that consider various forms of “significant others” influences that previous literature suggests are associated with postsecondary expectations and outcomes (Cheng and Starks 2002): whether the teen talked to teachers, coaches, or counselors about college (Stanton-Salazar 2011); whether the teen discussed the importance of college with parents (Sandefur, Meier, and Campbell 2006); and the teen’s perception of the likelihood their friends and classmates will graduate college (Alvarado 2021). Finally, we include macro-level measures of disadvantage that could co-occur in contexts where incarceration is more common, including neighborhood disadvantage, police contact, and the racial and poverty context of the school the teen attends (full list presented in Table 1).

Analytic Approach

To examine the association between paternal incarceration and adolescents’ college expectations, we estimate a series of linear probability and standard linear regression models. We use linear probability models for our two binary outcomes and linear regression for our measure of optimism about the future. We use linear probability as opposed to logistic regression models for our two binary outcome variables in order to use a consistent modeling approach for ease of interpretation across our outcome variables (coefficients of interest are similar in direction and statistical significance when using logistic regression). We use extensive covariate adjustment, examining regression coefficients from bivariate models (no control variables) and full models with all controls. Even with a long list of observable covariates, it is possible our explanatory variable—paternal incarceration—is partially capturing unobserved aspects of family disadvantage. This is an important challenge because for the majority (68 percent) of teens in our sample whose fathers were incarcerated, the incarceration history began very early in the child’s life (by Y1) or prior to the child’s birth. This means many baseline characteristics of the father and his relationship with the mother are already plausibly affected by incarceration. Therefore, as an additional control variable, we include whether the father spent time in jail or prison by Y1. As a sensitivity analysis, we reran our models excluding children whose fathers were incarcerated prior to Y1; results are substantively unchanged.

Results

Sample Description

Table 1 presents descriptive statistics for all the variables used in our analysis. More than three-quarters of our analytic sample is Black (54 percent) or Latinx (23 percent), and about 14 percent have a biological parent who is not a U.S. citizen. About one-third of fathers (32 percent) and mothers (37 percent) have any postsecondary education, and about one-quarter (26 percent of fathers, 22 percent of mothers) reported a period of unemployment at some point during the study. Only 24 percent of biological parents were married at their child’s birth (36 percent cohabiting), and less than half of mothers reported having a very good relationship with the father one year later, even though 70 percent of fathers were still living with their child at least some of the time. Between Y1 and Y9, 36 percent of fathers spent time in prison or jail.

Despite these disadvantages, youth in our sample were generally optimistic about their futures and not only valued college but expected to graduate. In fact, for our measures of optimism about the future (3.42 versus 3.40) and the importance of college graduation (86 versus 85 percent), we find negligible and non-significant mean differences between teens with and without childhood paternal incarceration. However, we do see descriptive differences for our main outcome of interest: teens’ perceived likelihood of college graduation. Teens whose fathers were incarcerated during their childhood have a 16 percent reduction in college expectations (59 versus 68 percent; t = 5.50; p < 0.001).

Association of Paternal Incarceration with Attitudes and Expectations

Table 2 presents our results for the association between paternal incarceration and each of our three outcome variables. For each outcome, we present a bivariate model first, followed by one that includes all our controls. Overall, results parallel the prior descriptive statistics for the associations of paternal incarceration between Y1 and Y9 with each of our outcome variables. Associations between paternal incarceration and teens’ optimism about the future in both the bivariate and full models (Models 1 and 2) are positive, but the coefficients do not reach statistical significance (b = 0.023, se = 0.018 for bivariate; b = 0.016, se = 0.021 for full model). Similarly, for teens’ beliefs about the importance of college graduation (Models 3 and 4), neither the bivariate nor the full models show significant associations with paternal incarceration (b = 0.004, se = 0.013 for bivariate; b = 0.028, se = 0.015 for full model).

Table 2.

Linear Regression and Linear Probability Models of Optimism about the Future, Importance of College Graduation, and Perceived Likelihood of College Graduation

Explanatory Variables	Optimistic about Future				Importance of College Graduation				Perceived Likelihood of College Graduation
	Model 1: Bivariate		Model 2: Add Controls		Model 3: Bivariate		Model 4: Add Controls		Model 5: Bivariate		Model 6: Add Controls
	b	se	b	se	b	se	b	se	b	se	b	se
Paternal Incarceration
Father incarcerated between Y1 and Y9	0.023	(0.018)	0.016	(0.021)	0.004	(0.013)	0.028	(0.015)	−0.096 ***	(0.018)	−0.047 *	(0.020)
Control Variables
Early paternal incarceration			−0.017	(0.022)			−0.014	(0.015)			−0.013	(0.020)
Father lives with child Y1			−0.030	(0.021)			0.020	(0.015)			−0.003	(0.020)
Father deceased			0.008	(0.027)			−0.007	(0.022)			0.019	(0.028)
Mother criminal charges			−0.056	(0.033)			−0.037	(0.026)			−0.044	(0.032)
Father impulsivity			−0.011	(0.010)			−0.011	(0.007)			−0.017	(0.010)
Mother impulsivity			−0.014	(0.009)			−0.007	(0.007)			−0.009	(0.009)
Father substance use			−0.036	(0.029)			0.007	(0.020)			−0.008	(0.029)
Mother violent victimization by father			0.024	(0.029)			−0.001	(0.022)			−0.008	(0.030)
Mother and father married			0.052	(0.029)			−0.002	(0.021)			0.039	(0.028)
Mother and father cohabiting			0.010	(0.022)			−0.016	(0.016)			−0.025	(0.021)
Mother and father very good relationship quality			0.006	(0.022)			0.013	(0.015)			0.036	(0.021)
Mother reads to child			0.051 **	(0.019)			0.008	(0.014)			0.024	(0.018)
Mother or father not a U.S. citizen			−0.072 *	(0.034)			0.063 *	(0.024)			−0.012	(0.032)
Father age			0.002	(0.001)			−0.002	(0.001)			−0.001	(0.001)
Mother lived with both parents at age 15			−0.032	(0.019)			−0.005	(0.013)			−0.025	(0.018)
Father lived with both parents at age 15			−0.026	(0.020)			−0.013	(0.015)			0.003	(0.020)
Mother postsecondary education			−0.003	(0.022)			0.045 **	(0.015)			0.046 *	(0.021)
Father postsecondary education			−0.039	(0.024)			0.018	(0.016)			0.049 *	(0.022)
Father cognitive ability			−0.023 *	(0.012)			−0.010	(0.008)			−0.021 *	(0.010)
Mother cognitive ability			−0.005	(0.010)			−0.004	(0.007)			−0.012	(0.010)
Mother religious attendance			0.021	(0.018)			0.000	(0.013)			0.007	(0.017)
Mother depression			−0.032	(0.026)			−0.019	(0.019)			0.003	(0.025)
Father unemployed			0.010	(0.023)			0.030	(0.016)			0.009	(0.022)
Mother unemployed			−0.002	(0.022)			0.017	(0.016)			0.000	(0.022)
Mother income-to-poverty ratio			0.006	(0.004)			0.007 *	(0.003)			0.023 ***	(0.004)
Mother children in household			−0.003	(0.007)			−0.010	(0.006)			−0.010	(0.008)
Mother multi-partner fertility			−0.009	(0.021)			−0.021	(0.016)			0.019	(0.020)
Child Black			0.167 ***	(0.029)			0.054 **	(0.021)			0.103 ***	(0.026)
Child Hispanic			0.096 **	(0.035)			0.026	(0.025)			0.015	(0.032)
Child male			0.029	(0.017)			−0.059 ***	(0.012)			−0.124 ***	(0.017)
Child age in months			0.005 **	(0.002)			−0.002 *	(0.001)			−0.002	(0.001)
Child low birth weight			0.000	(0.033)			0.012	(0.021)			−0.004	(0.029)
Child poor health Y1			−0.030	(0.019)			−0.007	(0.014)			−0.046 *	(0.018)
Constant	3.399 ***	(0.011)	2.195 ***	(0.317)	0.853 ***	(0.008)	1.323 ***	(0.244)	0.685 ***	(0.010)	0.917 **	(0.306)
N	3,205		3,205		3,205		3,205		3,205		3,205

Data Source: Future of Families and Child Wellbeing Study.

Note: Sample limited to cases in which the father was living and known to the mother by Y1 and for which valid data were provided for youth attitudes and paternal incarceration. Models of college importance and likelihood of college graduation are linear probability models because the outcomes are binary. The 20 sample-city dummy variables are not shown for parsimony. Results are combined across 20 multiply imputed datasets. SE = robust standard errors. Y1 = year-one survey, Y9 = year-nine survey.

^*p < 0.05;

^**p < 0.01;

^***p < 0.001 (two-tailed).

Whereas teens with incarcerated fathers have similar optimism about the future and are similarly invested in the importance of college graduation, compared to their peers without childhood paternal incarceration, their perceived likelihood of graduating from college is significantly lower. Results in Models 5 and 6 indicate that the association between paternal incarceration is negative and statistically significant with and without the addition of controls. Specifically, paternal incarceration is associated with a b = 0.096 decrease in the probability of perceiving that college graduation is very likely (p < 0.001). This association declines by half when controls are included, suggesting much of the association is driven by observed heterogeneity between children whose fathers were incarcerated during their childhood and those whose fathers were not. When controls are included, paternal incarceration is associated with a b = 0.047 decrease in the probability of perceiving that college graduation is very likely (p < 0.05).

Although a five-percentage-point decrease may appear modest, the general resilience of college expectations indicates this is a substantively large effect. Note that the magnitude of the decrease in college expectations associated with paternal incarceration substantially surpasses most other predictors included in Table 2. Indeed, this approximately five-percentage-point penalty is almost exactly the same magnitude as the positive effect of having a mother (b = 0.046) or father (b = 0.049) who completed postsecondary education. Thus, it seems the “negative credential” (Pager 2003) of paternal incarceration carries almost the same psychic weight as the positive credential of a parent’s postsecondary degree.

Heterogeneity by Race and Ethnicity

Thus far, our analyses suggest that teens who experienced paternal incarceration during childhood (ages 1 to 9) report significantly lower expectations of college completion than do their urban peers, but associations do not seem to reflect general negative orientations toward schooling or one’s future. However, these overall models may mask important variation across racial/ethnic groups, given the highly racialized nature of the legal system. Therefore, given racial disparities in incarceration rates, exposure to parental incarceration, and educational outcomes, Table 3 presents linear regression and linear probability models of the association of paternal incarceration among our three outcomes, this time within race/ethnic subsamples. Results for Black teens are found in Panels A and B, White/other² teens in Panels C and D, and Latinx teens in Panels E and F.

Table 3.

Linear Probability Models of Association of Paternal Incarceration with Optimism about the Future, Importance of College Graduation, and Perceived Likelihood of College Graduation, by Race and Ethnicity

Models	Black				White/Other				Latinx
	Panel A: Bivariate Models		Panel B: Add Controls		Panel C: Bivariate Models		Panel D: Add Controls		Panel E: Bivariate Models		Panel F: Add Controls
	b	se	b	se	b	se	b	se	b	se	b	se
Optimism about the Future
Father incarcerated between Y1 and Y9	−0.050 *	(0.023)	−0.045	(0.026)	0.089	(0.047)	0.196 **	(0.057)	0.039	(0.038)	0.062	(0.045)
Constant	3.491 ***	(0.015)	2.346 ***	(0.420)	3.277 ***	(0.022)	3.232 ***	(0.763)	3.358 ***	(0.022)	1.189	(0.663)
Importance of College
Father incarcerated between Y1 and Y9	0.007	(0.017)	0.010	(0.018)	−0.025	(0.035)	0.040	(0.044)	−0.002	(0.029)	0.070 *	(0.033)
Constant	0.859 ***	(0.011)	1.595 ***	(0.325)	0.842 ***	(0.015)	1.189 *	(0.611)	0.852 ***	(0.016)	0.819	(0.500)
Perceived Likelihood of College Graduation
Father incarcerated between Y1 and Y9	−0.101 ***	(0.023)	−0.073 **	(0.025)	−0.178 ***	(0.045)	0.049	(0.051)	−0.071	(0.040)	−0.049	(0.046)
Constant	0.719 ***	(0.014)	1.256 **	(0.419)	0.707 ***	(0.019)	0.486	(0.650)	0.596 ***	(0.022)	0.505	(0.646)
N	1,745		1,745		717		717		743		743

Data Source: Future of Families and Child Wellbeing Study.

Note: Sample limited to cases in which the father was living and known to the mother by Y1 and for which valid data were provided for youth attitudes and paternal incarceration. Control variables are not shown for parsimony. Results are combined across 20 multiply imputed datasets. SE = robust standard error. Y1 = year-one survey, Y9 = year-nine survey.

^*p < 0.05;

^**p < 0.01;

^***p < 0.001 (two-tailed).

Beginning with Panels A and B, we present results across our three outcomes among Black teens in the FFS sample who differ by paternal incarceration. The top row of Panel A presents bivariate associations between paternal incarceration between Y1 and Y9 and Black teens’ optimism about the future. Among Black teens with incarcerated fathers, the association between paternal incarceration and optimism about the future becomes negative (b = −0.050, se = 0.023, p < 0.05) but is no longer statistically significant (b = −0.045, se = 0.026) after controls are included (top row Panel B). For Black teens’ perceived importance of college graduation, the associations are positive but not significant with and without controls (b = 0.007, se = 0.017 for bivariate; b = 0.010, se = 0.018 for full). Finally, for Black teens whose fathers were incarcerated, the perceived likelihood of graduating college is negative and statistically significant, both in our bivariate (b = −0.101, se = 0.023; third row Panel A) and full (b = −0.073, se = 0.025; third row Panel B) models. Notably, the negative associations documented here for Black teens—around a seven-percentage-point decrease—are even larger than those estimated for the entire sample above (−0.047 versus −0.073). This suggests paternal incarceration may not only be most common among Black teens, it may also be most consequential.

Moving to analyses within White/other teens in our sample, the top row of Panels C and D present associations between paternal incarceration and optimism about the future. In contrast to results for Black teens, the association between paternal incarceration and optimism is positive (b = 0.089, se = 0.047) and becomes statistically significant when controls are included (b = 0.196, se = 0.057, p < 0.01). White/other teens with incarcerated fathers appear to be more optimistic about the future than their White/other peers in the FFCWS with no paternal incarceration experience, when adjusting for observable controls. Supplementary analyses using seemingly unrelated regression reveal that the differences between the negative coefficients for Black teens and positive coefficients for White/other teens are statistically significant (see the online supplement).

For our measure of the importance of college graduation (second row Panels C and D), the association with paternal incarceration is negative but not statistically significant (b = −0.025, se = 0.035) and becomes positive with the addition of controls but does not achieve significance (b = 0.040, se = 0.044). Whereas the association of paternal incarceration with White/other teens’ perceived likelihood of college graduation is negative and statistically significant prior to adding controls (b = −0.178, se = 0.045, p < 0.001), the direction of association again becomes positive when controls are added. However, the coefficient is not significant, suggesting it is driven by observed heterogeneity between White/other teens whose fathers were incarcerated and those whose fathers were not. Supplementary analysis suggests this association with perceived likelihood of college graduation for White/other teens is significantly different from the negative association we saw for Blacks teens (see Table A3 in the online supplement).

Finally, Panels E and F present our analyses of Latinx teens. This subsample has some similarities and differences with the Black and White/other teen subsamples. The top row presents our results for the association between paternal incarceration and optimism about the future; coefficients across both models are positive but not statistically significant (b = 0.039, se = 0.038 bivariate; b = 0.062, se = 0.045 full model). Turning to the importance that Latinx teens place on college graduation, the association with paternal incarceration is negative but not statistically significant (b = −0.002, se = 0.029), yet becomes positive and significant after controls are included (b = 0.070, se = 0.033, p < 0.05). Finally, the association between paternal incarceration and perceived likelihood of college graduation for Latinx teens is negative with and without the addition of controls (b = −0.071, se = 0.040 bivariate; b = −0.049, se = 0.046 full model) but does not reach statistical significance. Although not statistically significant, this coefficient for Latinx teens with childhood parental incarceration (−0.049) is of similar magnitude as the results (−0.047) found in the full model (Table 2).

Apart from the above primary findings, it is also worth noting that—across all outcomes—Black teens report significantly more positive attitudes and beliefs than do observably similar White/other teens (see Table A2 in the online supplement). For example, Black teens are 5.4 percentage points more likely to report it is “very important” to graduate college. Again, this is a substantively large effect: roughly equivalent in magnitude to being female (5.9 percentage points) or having a college-educated mother (4.5 percentage points). Even within this relatively disadvantaged sample, we replicate and extend prior research demonstrating Black students’ optimism and academic orientation (e.g., Diamond and Huguley 2014; Harris 2008).

Exploring the Effects of Y9 Mediators and Y15 Mechanisms

As Table 3 demonstrates, the negative association between paternal incarceration and college expectations appears to be concentrated among Black teens. Therefore, in our final set of analyses, we explore Y9 mediators and Y15 mechanisms to help us understand what might be behind the negative association between paternal incarceration and college expectations among the Black teen subsample. Table 4 presents the results. Model 1 starts where the third row of Panel B left off: the association of childhood paternal incarceration on Black teens’ perceived expectation for college graduation after our full set of controls. This relationship (b = −0.073, se = 0.025, p < 0.01) is negative and statistically significant.

Table 4.

Linear Probability Models of Association of Paternal Incarceration with Perceived Likelihood of College Graduation among Black Youth

	Model 1: Include Control Variables		Model 2: Control for Academic Mediators		Model 3: Control for Behavioral Mediators		Model 4: Control for High School Involvement		Model 5: Control for Influence of Significant Others		Model 6: Control for Contextual Disadvantage		Model 7: Full Model
	b	se	b	se	b	se	b	se	b	se	b	se	b	se
Explanatory Variable
Father incarcerated between Y1 and Y9	−0.073**	(0.025)	−0.072**	(0.025)	−0.065*	(0.025)	−0.060*	(0.025)	−0.065**	(0.025)	−0.068 **	(0.025)	−0.050*	(0.025)
Y9 Mediators
Academic
Home-based parental involvement in schooling			0.005	(0.017)									0.006	(0.016)
School-based parental involvement in schooling			0.057 *	(0.026)									0.046	(0.024)
Special education			−0.048	(0.039)									−0.025	(0.038)
Grade retention			−0.009	(0.033)									0.003	(0.032)
Behavioral
Externalizing behavior problems					−0.033 *	(0.013)							−0.021	(0.013)
Suspension or expulsion					−0.041	(0.029)							−0.008	(0.028)
Y15 Mechanisms
Micro-level: High School Involvement
Grade point average							0.163 ***	(0.018)					0.139***	(0.018)
School attachment							−0.170 ***	(0.034)					−0.102**	(0.035)
Meso-level: Significant Others
Talked with teacher/coach about college									0.046*	(0.023)			0.027	(0.023)
Talked with counselor/other about college									0.080**	(0.023)			0.056*	(0.023)
College graduation important to parent									0.358***	(0.043)			0.325***	(0.044)
Talked with parent about college									0.036	(0.024)			0.024	(0.023)
Most classmates likely to graduate college									0.037	(0.025)			0.022	(0.025)
Most friends likely to graduate college									0.131***	(0.023)			0.101***	(0.023)
Macro-level: Contextual Disadvantage
Neighborhood disadvantage											0.027	(0.016)	0.029	(0.015)
School majority students of color											0.063 *	(0.031)	0.049	(0.029)
School majority free or reduced-price lunch											−0.066 *	(0.031)	−0.039	(0.030)
Police contact											−0.088 **	(0.026)	−0.042	(0.025)
Constant	1.256 **	(0.419)	1.198 **	(0.427)	1.235 **	(0.418)	0.774	(0.404)	0.605	(0.411)	1.211 **	(0.416)	0.167	(0.406)
N	3,205		3,205		3,205		3,205		3,205		3,205		3,205

Note: Future of Families and Child Wellbeing Study. Sample limited to cases in which the father was living and known to the mother by Y1 and for which valid data were provided for youth attitudes and paternal incarceration. Control variables are not shown for parsimony. Results are combined across 20 multiply imputed datasets. SE = robust standard errors.

^*p < 0.05;

^**p < 0.01;

^***p < 0.001 (two-tailed).

For Models 2 and 3, we include several academic and behavioral consequences of paternal incarceration measured at Y9. Model 2 focuses on the academic: home-based and school-based parental involvement in schooling, special education placement, and grade retention. Model 3 considers the behavioral: behavior problems and school discipline. Accounting for each of these sets of Y9 mediators—previously documented consequences of parental incarceration for children’s educational outcomes—appears to explain part of the association³ with perceived likelihood of college graduation for Black teens. Our behavioral mediators (Model 3) account for more of the relationship, reducing it by 12 percent. Even so, the coefficient (b = −0.065, se = 0.025, p < 0.05) remains negative and statistically significant.

Model 4 considers the first set of Y15 measures contemporaneous to our outcome. These include micro-level mechanisms such as GPA and school attachment. These contemporaneous high school involvement measures appear to explain a larger portion of the association (b = – 0.073 versus b = −0.060; 18 percent) but still do not reduce the association between paternal incarceration and the perceived likelihood of college graduation for Black teens to insignificance. Model 5 provides a test of “significant others’” influences (parent, coach, teacher, counselor, peers) that previous literature suggests is associated with postsecondary expectations and outcomes. Including these meso-level potential mechanisms⁴ explains a portion of the association (b = −0.073 versus b = −0.065; 11 percent). However, even here, a negative and significant relationship remains.

Model 6 estimates the effect of including macro-level mechanisms measuring contemporaneous contextual factors of neighborhood and schools that could co-occur in contexts where incarceration is more common. This includes contextual disadvantage and police contact for Black teens. On their own, these macro-level mechanisms only explain about 7 percent (b = – 0.073 versus b = −0.068) of the association. Looking across Models 1 and 6 of Table 4, we see that behavioral mediators (Model 3) measured as early as Y9 carry lasting effects on Black teens’ perceived likelihood of college graduation, with only teen’s contemporaneous achievement in high school accounting for more of the association (12 versus 18 percent).

Finally, Model 7 includes both Y9 mediators and all of the Y15 mechanisms. Together, this combination explains about 32 percent of the association between paternal incarceration and college expectations for Black adolescents in our sample. Even so, we still see a negative association between childhood paternal incarceration and Black teens’ college expectations. Indeed, paternal incarceration between Y1 and Y9 is associated with about a five-percentage-point lower probability (b = −0.05, se = 0.025, p < 0.05) of believing that one is very likely to graduate from college.⁵

Discussion

For the past 50 years, educational expectations have held a central place in sociological accounts of the transition to adulthood. Expectations, as Haller and Portes (1973:68) put it, are where youths’ hopes for the future come together with their “realistic appraisal of possibilities conveyed . . . by significant others.” As today’s teens form their expectations, millions need to grapple with a new message: one conveyed by parental incarceration. Over and above its immediate effects on family life and school success, what message does parental incarceration send to students about their prospects for the future?

We addressed this question using data from FFCWS, documenting that teens who have experienced paternal incarceration report significantly lower college graduation expectations than do their urban peers. Indeed, we estimate that the “negative credential” (Pager 2003) of a father’s incarceration shapes college expectations to the same degree as the positive credential of a father’s postsecondary education. To contextualize this large negative effect on college expectations, we also estimated whether paternal incarceration predicts similar disadvantages in teens’ responses to two related measures: their beliefs about the importance of college and their optimism about their future more broadly. Across these additional outcomes, we find no evidence of disadvantages among teens affected by paternal incarceration. If anything, our results partially point toward greater positivity and signs of optimism. Thus, even as they assert the importance of college and remain broadly hopeful about their futures, teens experiencing paternal incarceration appear to anticipate a substantially steeper climb to college graduation.

Considering the raced nature of mass incarceration, we found evidence that these negative effects are most salient for Black teens. Given that Black children face paternal incarceration at substantially higher rates than their peers, one could imagine that the signal sent by a father’s incarceration would be less salient for these youth. Unfortunately, our results appear to show the opposite. When it comes to shaping their academic expectations, paternal incarceration is not only most common among Black youth, it is also most consequential. Even compared to other Black teens growing up in similarly disadvantaged contexts, those who faced paternal incarceration during childhood report significant reductions in their expectations for college completion. Notably, these reduced expectations emerge despite the fact that Black teens in this sample report a significantly greater investment in the importance of college graduation.

Other interesting findings surfaced in our race/ethnic subsample analysis that are worth noting. Literature often groups Black and Latinx youth together, but we found differing results when separating them. At least within the FFCWS, Latinx youth looked more similar to their White/other peers; we think this could be due to sample attrition being most extensive among immigrant families. Likely, the Latinx teens still participating at Y15 are, on average, more advantaged, and we see this across various demographic measures (see Table A2 in the online supplement).

Do these reduced academic expectations simply reflect paternal incarceration’s negative effects on academic performance? As best we can tell, the answer to this question is no. Controlling for a wide range of academic outcomes, educational decisions, and family experiences that occur as early as 3rd grade only reduces—but does not remove—the significant negative effect of paternal incarceration for Black youth. Instead, incarceration itself appears to promote a precise sense of racialized constraint about postsecondary possibilities—potentially a more realistic understanding of the obstacles the marker of stigma from incarceration carries for Black families. This aligns with previous work on incarceration and perceptions (e.g., Brew et al. 2022; Schnittker and Bacak 2013) and demonstrates how stigma, stress, and traumatic loss due to incarceration could fundamentally (re)shape children’s sense of possibility and understanding of self.

By updating educational expectations research for the era of mass incarceration, our results demonstrate another way parental incarceration structures racial inequality for the next generation. Over and above its previously documented negative effects on students’ academic performance (e.g., Haskins 2014; Turney and Haskins 2014), we find that paternal incarceration also negatively affects students’ sense of academic possibility. It does so not by undercutting teens’ perspectives on the importance of higher education, nor by constricting their hopes for the future more broadly. Instead, incarceration seems to send the specific message that college completion is less likely for “teens of the prison boom.” Future work, especially qualitative work, should seek to identify processes through which parental incarceration influences adolescents’ educational success or failure, particularly with respect to the formation of attitudes, beliefs, or expectations around schooling.

Several limitations of our findings should be reiterated. First, although our multivariate results control for whether the mother and father were married at baseline, our sample is over-representative of unmarried families. Our findings may therefore not be generalizable beyond disadvantaged urban families. Second, although we include a long list of potential confounders, there may be unobserved differences between teens who have experienced the incarceration of a father and those who have not, for which we have not accounted. Ideally, we would address this by examining within-individual change in college-related attitudes associated with paternal incarceration, but this requires more waves of data than are available in FFCWS.

A final limitation to note is that, although we find strong evidence that paternal incarceration is associated with comparatively low college expectations, these expectations—in absolute terms—are still very high. About 60 percent of the teens in our sample who experienced paternal incarceration report that college graduation is “very likely” and an additional 37 percent report it is “somewhat likely.” At a time when only 42 percent of the adult population has an associate’s degree or higher (Ryan and Bauman 2016), many of our respondents will likely find their expectations were unreasonably high. Nevertheless, insofar as unreasonably high college expectations are now normative among teens (Schneider and Stevenson 1999), the high level reported by affected teens reinforces our broader finding that these teens are generally optimistic within the same broad cognitive landscape as their peers.

As the “children of the prison boom” (Wakefield and Wildeman 2014) age into adolescence, it is becoming possible—and necessary—to more fully account for the intergenerational harms inflicted by the era of racialized mass incarceration in the United States. In this study, we extended prior research on paternal incarceration’s educational effects by analyzing how it shapes students’ expectations and ideas about the future. In so doing, we brought incarceration research into conversation with sociology of education’s long-standing interest in the sources and consequences of academic attitudes, particularly among students of color. Within this relatively disadvantaged urban sample, we found that “achievement ideology” remained high: even the experience of paternal incarceration was not enough to shake teens’ belief that college graduation was “very important” for their future success. And yet, even if paternal incarceration did not cause teens to question the importance of college, it diminished their expectation that college graduation was actually possible for them. Thus, signs of optimism surfaced, yet obstacles associated with paternal incarceration represent a type of racialized constraint felt most strongly among impacted Black teens. Our findings point to the complexity of contemporary teens’ college-related attitudes in the wake of the prison boom and suggest an updating of the educational expectations literature is needed to account for the racialized nature of mass incarceration in the United States.

Biographies

Anna R. Haskins is the Andrew V. Tackes Associate Professor of Sociology and the Associate Director of the Initiative on Race and Resilience at the University of Notre Dame. Her research examines how three of America’s most powerful social institutions—the education system, the family, and the criminal legal system—connect and interact in ways that both preserve and mitigate social inequality, with emphases on early educational outcomes, intergenerational impacts, and disparities by race/ethnicity.

Wade C. Jacobsen is an assistant professor in the Department of Criminology and Criminal Justice at the University of Maryland, College Park. He is also a faculty associate at the Maryland Population Research Center and a member of the research team for the Maryland Longitudinal Data System Center. His research examines how interactions with institutions (schools, criminal legal system) and social networks shape the development of risky behaviors in childhood, adolescence, and young adulthood.

Joel Mittleman is an assistant professor of sociology at the University of Notre Dame, where he is also a faculty affiliate of the Gender Studies Program and the Center for Research on Educational Opportunity. Mittleman studies inequality with a focus on gender, education and LGBTQ+ populations.