Children of the American Prison Generation: Student and School Spillover Effects of Incarcerating Mothers

Abstract

Formal equality and judicial neutrality can lead to substantive inequality for women and children, with social costs that extend beyond individuals and families and spill over into the larger social settings in which they are located. We consider the uniquely damaging effects of an “equality with a vengeance” (Chesney-Lind & Pollack 1995) that resulted from “tough on crime” policies and the 1980s federal and state sentencing guidelines that led to the incarceration of more women and mothers. We argue that legal equality norms of the kind embedded in the enforcement of sentencing guidelines can mask and punish differences in gendered role expectations. Paradoxically, although fathers are incarcerated in much greater numbers than are mothers, the effect threshold is lower and the scale of effect on educational outcomes tends to be greater for maternal incarceration. We demonstrate both student- and school-level effects of maternal incarceration: the damaging effects not only affect the children of imprisoned mothers but also spill over to children of nonincarcerated mothers in schools with elevated levels of maternal incarceration. We find a 15 percent reduction in college graduation rates in schools where as few as 10 percent of other students' mothers are incarcerated. The effects for imprisoned fathers are also notable, especially at the school level. Schools with higher father incarceration rates (25 percent) have college graduation rates as much as 50 percent lower than those of other schools. The effects of imprisoned mothers are particularly notable at the student level (i.e., with few children of imprisoned mothers graduating from college), while maternal imprisonment effects are found at both student and school levels across the three measured outcomes. We demonstrate these effects in a large, nationally representative longitudinal study of American children from the 1990s prison generation who were tracked into early adulthood.

Women presently make up less than 10 percent of the U.S. prison population. Over the last three decades, however, the imprisonment of women increased about sixfold, compared to threefold for men (Pattillo, Weiman, & Western 2004). The majority of imprisoned women are mothers, and the number of imprisoned women doubled in the last decade (Wildeman 2009). The rapid rise in women's and mothers' imprisonment has resulted more from changes in sentencing than from changes in crime (Steffensmeier et al. 2006). The collateral consequences for women, mothers, and children, as well as the spillover effects on their schools and communities, are largely unstudied (Hagan & Dinovitzer 1999). The increased imprisonment of women has potentially long-lasting and intergenerational significance when viewed in terms of the culturally prescribed roles of mothers in child care, child raising, the emotional lives of children, and the surrounding social lives of children's schools and communities.

Prisons are now a major stratifying institution in American society. Akin to the military and universities in earlier generations, they create and maintain inequality within and between generations and communities (Wakefield & Uggen 2010). We focus in this article on influences of incarcerated mothers and fathers on school-based intergenerational educational inequality. Recent work highlights the need for comparative intergenerational research on maternal as well as paternal imprisonment effects (Murray & Farrington 2008; Wakefield & Uggen 2010). We address this need by analyzing a national longitudinal data set organized around youth and their parents in their families and schools, with a focus on the educational transitions of youth from adolescence to and through early adulthood.

The important work of Garland (2001a, 2001b), Western (2006), and others has focused attention on the massive imprisonment of American men. Kruttschnitt (2010: 34) emphasizes the too-often-neglected parallel rise in the imprisonment of women. Although far fewer women compared to men may be incarcerated overall, Kruttschnitt suggests that the threshold at which the removal of these women to prison negatively impacts their families and communities may be notably lower than the threshold when men are removed. Thus the paradox is that incarcerating many fewer women may have an even greater impact than incarcerating far more men. The primary role of women in child care may actually make Kruttschnitt's hypothesis more of a well-grounded prediction than an unexpected paradox. Rothman (2005: 53) reminds us of the foundation for this expectation by emphasizing that motherhood is “a, and maybe the, prime relationship, primary in the lifespan of the person being mothered, primary in establishing our understanding of what it is to be connected with another human being.”

Kruttschnitt broadly poses the important resulting question: “What is the cost to society and individual families when women are sent to prison?”

This question has four distinguishable implications that we assess in terms of the educational impact on children. The first is the possibility at the individual level that although the great majority of inmates are male, the effect of imprisoning mothers may be larger on children than that of imprisoning fathers (Dallaire 2007; Murray & Farrington 2008). The second possibility is that at the aggregate level children experience concentrated school spillover effects of the incarceration of many surrounding mothers, even when the mothers incarcerated are not the children's own mothers (Clear 2007; Hagan & Dinovitzer 1999; Rose & Clear 1998; Sampson & Loeffler 2010). Research on the spatial clustering of incarceration shows that the prison sentencing of women with children is residentially concentrated (Brazzell 2008). Cho (2011) examines prevalence effects of maternal imprisonment in her work on school dropouts in a Chicago sample. Our research focuses, for the first time in a national sample, on the concentration within schools of both maternal and paternal imprisonment. The third possibility we examine is that the threshold of the impact of maternal imprisonment spilling over onto children beyond the immediate family may be lower than that for paternal imprisonment (Kruttschnitt 2010). The fourth possibility is that all these effects may be most consequential for the increasingly important educational outcome of college graduation (Western 2006).

The concept of spillover effects has been used previously, for example, to refer to the beneficial effects of group characteristics, such as the average number of years of schooling in a state or city, on individual outcomes. Thus Angrist and Pischke (2009: 193) suggest that “living in a state with a more educated workforce may make everyone in the state more productive, not just those who are more educated.” This kind of spillover is said to provide a social return to schooling with social benefits for everyone, regardless of which citizens are themselves more educated. As noted below, and in contrast with this focus on beneficial spillover effects on schooling, we are concerned that imprisonment's detrimental spillover effects include social costs involving schooling.

We will consider the potentially damaging effects of the imprisonment of women during an unprecedented social experiment begun in the 1980s with the implementation in the United States of federal and state sentencing guidelines (Hagan 2010). This experiment applied legal arguments about norms of equality to women—with little or no attention to how increasing the imprisonment of mothers would have unanticipated consequences for children, families, and communities.

By analyzing and comparing how the incarceration of mothers and fathers impacts the education of children, this article reflexively turns the traditional social science of penology back on itself. Smith (2008) has argued that a renewed neo-Durkheimian reflexivity can put a “wedge between culture and control” (182) and in this way expose, if not slow, the “runaway train of disciplinary society” (183). Our analysis pursues this possibility by elaborating a gendered critique of the increased imprisonment of mothers resulting from the application of sentencing guidelines and related “tough on crime” policies. This critique reveals the gendered nature of the ostensibly egalitarian move that has increased the imprisonment of women to unprecedented levels in the United States.

Research on the Effects of Imprisonment

Research indicates multiple reasons why maternal imprisonment may be more influential on children than paternal imprisonment is. First, during the early life course, mothers on average spend more time with children than fathers do and therefore have greater opportunity to influence their children (Murnane, Maynard, & Ohls 1981). Second, when mothers are incarcerated, their children are unlikely to live with their fathers; their living arrangements are more likely to involve separation from the imprisoned biological mother and relocation to a new household with an aunt, grandmother, or foster mother as the new primary caregiver (Glaze & Maruschak 2008; Johnson & Waldfogel 2004; Mumola 2000). Third, because there are far fewer prisons for women than there are for men, mothers tend to be incarcerated farther away from their children, thus making visits and sustained contact and communication with their children less likely (Coughenour 1995): half or more of the children of imprisoned mothers do not see their mothers during their entire incarceration (Glaze & Maruschak 2008; Snell 1994), and studies consistently indicate that family living arrangements and attachments are greatly disrupted by periods of maternal incarceration (Poehlmann 2005; Poehlmann et al. 2010). Fourth, children of incarcerated mothers are exposed over time to more risks in the home environment than are children of incarcerated fathers, which indicates that children of incarcerated mothers are exposed to a more adverse accumulation of childhood disadvantages (Johnson & Waldfogel 2004). Fifth, children living with incarcerated mothers have a greater tendency to experience multiple incarceration-related events (e.g., being present at sentencing) compared to children with incarcerated fathers (Dallaire & Wilson 2010), and these incarceration-connected experiences may further add to the stresses associated with maternal incarceration. Finally, research on teachers reveals that these key socializing agents see maternal imprisonment as more damaging than is paternal imprisonment (Dallaire et al. 2010), and cultural norms may have different effects on how children are perceived and treated at school when mothers are imprisoned instead of fathers. Prior research therefore suggests that maternal imprisonment should be more consequential for children than paternal imprisonment would be, even though paternal imprisonment is a more frequent and familiar part of children's lives.

The collateral damage and spillover effects of maternal imprisonment on children, families, schools, and their surrounding communities are the particular focus of this article. The spillover effects of maternal incarceration are a gendered example of the potential ramifying effects of “equality with a vengeance” (Chesney-Lind & Pollack 1995)—in this case, on women, children, schools, and communities. While policy makers may often have been blind to such ramifying effects, much as society once was to the secondhand effects of smoking, the public is increasingly confronted with the more widely dispersed effects of policies like maternal imprisonment. Patricia Hill Collins (2010: 12) observes how such policies broadly institutionalize inequality:

Social structures such as neighborhoods, schools, jobs, religious institutions, recreational facilities, and malls are the institutional expressions of social inequalities of race, class, gender, age, ethnicity, religion, sexuality, and ability. These structures are typically hierarchical and offer unequal opportunities and rewards. When people travel among neighborhoods, they notice these structural inequalities. Increasingly, media enables people to see structural inequalities, both locally and globally.

The purpose of this article is to capture the interplay of children, parents, and schools as they individually and collectively experience the inequalities of maternal and paternal incarceration. We will see that schools in particular are important mediating institutional mechanisms through which the individual and spillover educational inequalities resulting from maternal and paternal incarceration are played out in the transitions from youth to adulthood in neighborhoods and communities across the United States.

Murray and Farrington (2008) highlight the importance of comparative research on maternal and paternal imprisonment as it relates to children's problem behaviors, but they also note there has been very little such research. Furthermore, the research that is available on the collateral consequences on children of parental imprisonment is focused on individual family units without distinguishing the gender of the parent incarcerated or considering the reach of these effects beyond the families that immediately experience them. That is, this work has neglected school-level and spillover effects of either maternal or paternal imprisonment on children who are exposed to the imprisonment of parents other than their own. With worldwide and historic high levels of imprisonment, this is likely an important time to examine the spillover effects of heightened levels of vicariously experienced parental incarceration, and especially maternal incarceration.

The little research that is available suggests mixed effects of maternal and paternal imprisonment on various child outcomes. For example, a study of inmates reveals that adult children of an incarcerated mother are more likely to be incarcerated than are children of incarcerated fathers (Dallaire 2007). However, another recent study comparing effects of maternal and paternal incarceration finds no cognitive differences at three years of age (Geller et al. 2009). Other research finds correlations with children's problems at school. Incarcerated mothers and substitute caregivers report that their children's school and learning difficulties are the children's greatest problems (Bloom & Steinhart 1993). Parental incarceration correlates with children's lower class rankings (Stanton 1980), failure of courses, and dropping out of school (Trice & Brewster 2004), truancy in adolescence, and failure to complete exams in the UK (Murray & Farrington 2008). However, some important recent research suggests that maternal incarceration neither increases grade retention for children in grades K through 8 (Cho 2009a) nor decreases elementary schoolchildren's math and reading standardized test scores (Cho 2009b). The inconsistency of parental effects on children's educational outcomes may involve the age of the child. Negative effects of parental imprisonment may accumulate and therefore be expressed more consistently at older ages and across a range of middle and later educational outcomes (e.g., school dropout [Cho 2010] and educational attainment [Foster & Hagan 2007, 2009] in midadolescence and through early adulthood).

Longitudinal research with designs that systematically vary the sampling of students and schools is required to trace potential impacts of maternal and paternal imprisonment across varied school settings and from childhood through adolescence and into adulthood. Such research is both timely and overdue following four decades of sentencing reform leading to steadily rising levels of imprisonment of mothers and fathers in the United States.

However, this research must proceed in new ways. Inequalities are spatially and institutionally concentrated, as Patricia Hill Collins (2010) suggests, with inequalities resulting from the incarceration of mothers and fathers concentrated beyond prison walls and within the classrooms and schools where children are educated. When inequalities such as heightened levels of maternal and paternal imprisonment are concentrated, they may create collective contextual effects that can spill over into the lives of others who do not directly experience the parental incarceration. Drawing on Sharkey's (2008, 2010) work on neighborhoods and children's cognitive performance, we suggest two perspectives from which to view school-level as well as individual-level maternal and paternal incarceration effects on the educational outcomes of children.

The first perspective focuses on the effects of interrupted parent-child relationships and the associated absence of role models. Rothman's (2005) observation of the primacy of the mother-child relationship in the life span and, more broadly, in establishing connections with others implies the salience of maternal incarceration in undermining successful life trajectories. Wilson (1996) and Sampson and Wilson (1995) include the associated absence of maternal and paternal role models as a source of social isolation that can confine cognitive landscapes and perceived pathways to educational and other attainments. Elder (1994) highlights the importance of maternal and paternal relationships and role models as the foundation for linked lives that establish and sustain educational and other kinds of transitions and trajectories. The effects of interrupted parent-child relationships can flow both at the individual level within families and at the school level among students who are influenced not only by disruptions of their own families but also by the spillover influence from disruptions and absences in the families of others.

The second perspective on parental incarceration's effects on children's educational achievement emphasizes the reduced availability of economic and educational resources (e.g., Jencks et al. 1972). These resources can both stimulate and support transitions and trajectories of educational achievement, while their absence can correspondingly weaken or break these progressions. Again, the effects of the presence or absence of these resources can operate at both individual and school levels. We incorporate a variety of measures of the economic and educational resources of individuals and schools in this article—for example, by including not only household income but also the mean educational level corresponding to family incomes. The latter measure can capture spillover effects resulting from the collective advantages or disadvantages created by clustering in schools of parents with similar incomes.

Both the relationship and resource perspectives just outlined are assumed to unfold in their consequences over stages of the life course. Especially at the school level, the assumption is that there is relatively limited variation in surrounding circumstances confronting students from childhood through adolescence and in the transition to adulthood. Individual circumstances involving parental imprisonment will vary, but collective circumstances, especially in the 1990s, often simply vary from bad to worse, thus exposing youth in many urban schools to increasingly and persistently unfavorable environments. The important additional implication of the focus on spillover effects is therefore to call attention to the policy consequences of the effects of parental incarceration through the persistent impact of surrounding families with imprisoned mothers and fathers at the broader level of schools.

In sum, the tendency of parental incarceration to damage educational trajectories of children may play out over the length of the life course and may be imposed through the environmental influence of families at the level of schools as much as or more than within the families themselves. In assessing effects of disrupted relationships and educational and economic resources, we must also take into account an alternative perspective that involves predisposing conditions such as parental alcoholism and neighborhood crime and drug problems. These and other predisposing conditions at the individual and school levels are included in the models estimated in this article.

Gender and the Sentencing Guidelines Movement

While the literature we have reviewed suggests how the mass imprisonment of parents might influence the intergenerational outcomes of children through families and schools, and while this literature gives grounds for special concern about the effects of the imprisonment of mothers, we also have seen that there is too little research to provide a systematic and consistent base of knowledge about these concerns for policy purposes. Yet in the absence of this knowledge, in the 1980s federal and state legislatures radically restricted in a historically unprecedented way the discretion of judges to consider gender-linked parenting responsibilities in decisions about imprisonment.

During Ronald Reagan's second presidential term, a law-and-order political agenda led to a powerful U.S. Sentencing Commission and restrictive federal sentencing guidelines (Hagan 2010: Chapter 5). Stith and Koh's (1993) legislative history captures the political contradiction at the core of this legal movement, which was “conceived by liberal reformers as an anti-imprisonment and anti-discrimination measure, but finally born as part of a more conservative law-and-order crime control measure” (223). Liberal post–civil rights reformers supported determinate sentencing reforms as a presumed means of ending racial and judicial disparities resulting from the discretion built into indeterminate sentencing provisions. However, conservative law-and-order opponents of the civil rights reforms and ghetto rebellions supported determinate sentencing as a means of reducing the discretion of “activist” judges and standardizing more severe punishment (Beckett 1997; Murakawa 2006; Wacquant 2009). The resulting legislation greatly increased the imprisonment of minority offenders, especially by targeting crack cocaine, and it caught mothers as well as fathers in its widening and deepening net. Thus, in the name of equality, the guidelines movement dictated that women receive sentences comparable in severity to men's, regardless of any special responsibilities for children and families.

While sentencing guidelines “treated all persons as formally equal,” Savelsberg (1992: 1348) explains that the guidelines simultaneously disregarded substantive social inequalities. These inequalities involved mothers who bore unique responsibilities for child care and for whom separation from their children was particularly consequential. This meant that liberal equality in sentencing for women translated into substantive inequalities for mothers. Barbara Katz Rothman (2000: 248) explains, “Liberal feminism works best to defend women's rights to be like men, to enter into men's worlds, to work at men's jobs for men's pay, to have the rights and privileges of men. But what of our rights to be women? The liberal argument, the fairness argument, the equal rights argument, these all begin to break down when we look at women who are, or are becoming, mothers.”

Being locked in to the male world of imprisonment is an injurious form of equality for mothers and has consequences for the children and communities that depend on them. Chesney-Lind and Pollock (1995; also Chesney-Lind 2006) characterize the implications for women, children, and communities of the “tough on crime” approach with their reference to “equality with a vengeance.” These consequences go beyond sentencing guidelines to include practices and policies stretching from policing to punishment and prisons to parole. Chesney-Lind's work with several collaborators established the increasingly punitive sanctioning of women and mothers, particularly when convicted of drug offenses (e.g., Bloom et al. 1994). Chesney-Lind and Pollock (1995) further identify the vengeful nature of this treatment in the shift away from the placement of women in special institutions (e.g., cottage-style housing) and toward sentencing women to new prisons designed to look like those built for men. Yet this did not mean that the actual treatment received by women was equal (McCorkel 2003). For example, the case files of young women continued to include references to physical attractiveness and sexual experience that singled them out for differential treatment (Rosenbaum & Chesney-Lind 1994).

These issues of equal treatment and unequal outcomes are reflected in the empirical research literature on gender and sentencing that further confirms how the imprisonment of women and mothers has increased. Simon and Ahn-Redding (2005) discuss two early schools of thought regarding gender and sentencing. They observe that while most researchers believed that women received preferential treatment, others continued to speculate that judges were more punitive toward women, especially when the offense violated gender role expectations (Steffensmeier et al. 1993: 74; see also Simon & Ahn-Redding 2005). The qualitative and quantitative research conducted by Kathleen Daly (1987a, 1987b, 1989a, 1989b) on the period before the passage of sentencing guidelines further complicates this picture. Daly observed in her interviews that preguideline judges were concerned they would “break up families” and “punish innocent family members” by incarcerating “familied offenders” (see Daly & Bordt 1995: 163). The implication of Daly's analysis was that previously observed sex effects that favored women were more specifically an extension of greater leniency toward women who were also mothers (Daly 1989a). These findings regarding family care conflict with the subsequent policy demanded by the U.S. Sentencing Commission for equal treatment of men and women with similar offense records and charges, regardless of parenting responsibilities.

Daly argues instead for institutionalizing the priority given to familied parents' care of their children. She concludes that “equal treatment of defendants whose responsibilities for others not only varied but differed by gender may be unjust” (see Daly & Bordt 1995: 163). Daly further reasons that a source of the problem was making unfamilied males rather than familied females the normative standard of comparison for sentencing. The cultural priority that made unfamilied men the comparative standard led to increasing imprisonment for women. Alternatively, Daly (1995) argues for prioritizing the special gender-linked demands on mothers, and she reverses the presumptive reliance on unfamilied males as the normative standard of comparison by asking, “[M]ight women-normed guidelines be more humane, more defensible?” (166).

In the legislation-induced urgency of making regulatory decisions, the U.S. Sentencing Commission came down on the side of guidelines that gave judges less discretion and directed them to punish mothers equally on the narrower basis of charges and prior record, while disregarding gender-linked family responsibilities. This policy experiment had profound implications for mothers and children. It meant treating as equal accused men and women whose family-connected vulnerabilities were actually quite different. Since judges previously incarcerated many fewer women than men, the new guidelines meant that annual rates of change in imprisonment for women, and thus mothers, increased much faster than those for men and fathers from the 1990s onward (Chesney-Lind & Pollack 1995; Harrison & Beck 2005; Sabol et al. 2009).

The class of young people whose lives are disrupted by separation from their imprisoned mothers and fathers is larger today than ever before in American history (Pattillo et al. 2004). This major change in American penal policy, which extends well beyond the example of sentencing guidelines, has massively increased punishment, from arrest through imprisonment, without corresponding knowledge of its consequences.

Methods for Studying the Children of Imprisoned Parents

We use four waves of the National Longitudinal Study of Adolescent Health (Add Health), which began in 1995 with a stratified sampling design of grades 7 to 12 nested within 132 representative U.S. schools (Harris 2009; Harris et al. 2009). These adolescents were born during the steep upturn, beginning in the late 1970s, that more than quadrupled inmates in U.S. prisons. Add Health collected information through four waves: in-home parent and child survey interviews in 1995 (at ages 12 to 21) and child interviews in 1996 (ages 13 to 22), 2001 to 2002 (ages 18 to 26), and 2007 to 2008 (ages 24 to 32).

We use retrospective data from Wave IV to measure the occurrence and timing of imprisonment of mothers and fathers, as well as prospective measures predominantly from Wave I to measure parents' likelihood of experiencing incarceration. Interviewers in Wave IV respectively located and interviewed 92.5 and 80.3 percent of the eligible Wave I sample members. More than 10,000 adolescents (and, later, young adults) participated in all four waves of data collection, and 9,421 of these participants received longitudinal sample weights (Chantala 2006).

We linked the Add Health survey to a supplementary collection of educational data from school transcripts—the Adolescent Health and Academic Achievement supplement (AHAA)—which tracks Add Health respondents to their mid-20s (Muller et al. 2007). Approximately 91 percent of Wave III respondents signed a release form for collection of supplementary school transcript data. We consider three academic outcomes: (1) high school grade point average (GPA) measured on a four-point scale from the AHAA supplementary survey, (2) a Wave IV measure at the average age of 27 of educational outcomes on a 13-point scale from completion of eighth grade to postbaccalaureate professional education, and (3) a Wave IV binary measure of college completion. The last measure specifically assesses the college/noncollege divide that became salient during the expansion of postsecondary education after World War II (Western 2006).

Measures and Models of Student and School Effects

We use hierarchical linear modeling (HLM) to estimate variation in educational outcomes measured within and between the schools (Raudenbush & Bryk 2002). HLM allows adjustments for nonindependence resulting from clustered school sampling through calculation of robust standard errors (Gottfredson 2001). We use the school weight at Wave I as well as the longitudinal sample weight at Wave IV. We present descriptive information about the variables used in this analysis in Table 1 and Appendix A.

Table 1. Descriptive Statistics

Reference Categories: ^aFemale = 1; Male = 0; ^bNon-Hispanic White; ^cAll other family types.

Respondents reported if and when their biological fathers and mothers had been in jail or prison. Nearly 15 percent of the Add Health youth reported in Wave III that their biological fathers “had served time in jail or prison.” In Wave IV, the respondents more specifically reported how old they were when their biological fathers first went to jail or prison. By Wave IV in 2008, 14.5 percent of fathers and 4 percent of mothers had spent time in jail or prison. We coded mothers' and fathers' imprisonment to the child's age of 18, so as to temporally precede their educational outcomes in their mid-20s.

We aggregated the measures of biological fathers' and mothers' imprisonment within schools to create school-level measures of parents imprisoned before their interviewed children's 18th birthdays. These measures are direct indicators of the interruption involved in the parent-child relationship perspective introduced above. Recall that the effects of interrupted parent-child relationships can flow both at the individual level within families and at the school level among students who are influenced not only by disruptions of their own families but also by the spillover effects of disruptions and absences in the families of others.

We have arrayed proportions of mothers and fathers imprisoned in the weighted sample of U.S. schools in ascending order in Figure 1. In some sampled American schools, as many as one-quarter of the fathers and one-tenth of the mothers experienced incarceration during the respondents' childhood and adolescence. By using reports of parental imprisonment at the individual level and school level as measures of maternal and paternal imprisonment, we can estimate separately individual and school-level spillover effects of parental imprisonment and the interruption of parent-child relationships on children's educational outcomes.

Figure 1. School Mean Levels of Maternal and Paternal Imprisonment between Adolescent Ages 0–18.

We considered many characteristics in addition to parental imprisonment as plausible further influences on educational outcomes at student and school levels. For example, we included several measures related to the economic and educational resources perspective also introduced above. The most familiar are perhaps the income level of the household and the mean income level of families whose children attend the school. We introduced the related educational-resource measures of the number of full-time teachers and the proportion of teachers with master's degrees. In addition, we controlled for the school's proportion of two-biological-parent families and the school's size, urbanicity, and level of public funding.

We further acknowledged above the importance of controlling for the alternative possibilities that predisposing problems in the schools might account for differences in educational outcomes. For this reason, we incorporated school attendance and school dropout rates as further likely determinants of educational outcomes. We also included area crime rates as another possible exogenous source of both parental imprisonment and educational detainment.

As already noted, we also included student-level variables that correspond to a number of school-level variables. Perhaps most important, we controlled at the individual level for the imprisonment of the mother and father before the child's age of 18 as well as the household income. We also included whether the mother or father graduated from college, was an alcoholic, or was a smoker. We further incorporated a measure of the closeness of the child to his or her biological father and mother. To parallel the school-level control for area crime rate, we incorporated a self-report scale of adolescent delinquency. We further included a student-level measure of single-parent status, as well as age, gender, and race/ethnicity.

Multilevel Models of Parental Imprisonment

Using regression estimations for the GPA and highest educational level completed and logit estimates for college completion, we estimate joined individual and school-level HLM equations for the three educational outcomes. We first estimate an individual-level equation separately for students in each school, and this yields regression coefficients (for each predictor) and an intercept term representing the student-input adjusted school outcome for each outcome measure (with the continuous predictors centered on their means) for each school. Our within-school modeling of the three educational outcomes thus takes the following form:

${\text{Educational Outcome}}_{ij}={\beta }_{0j}+\sum _{q=1}^{\Sigma }{\beta }_q{X}_{qij}+{ϵ}_{ij},$

where β _0j is the intercept; X_qij is the value of covariate q associated with respondent i in school-level j; and β_q represents the partial effects on the child's educational outcome of both the respondent father's and mother's imprisonment, college education, alcoholism, smoking, perceived closeness with the child, single-parent status, race/ethnicity, gender, age, household income, along with the child's self-reported delinquency. The error term, ɛ_ij, is the unique contribution of each student, which is assumed to be independently and normally distributed with constant variance σ ².

Second, we estimate the school-level equation, in which the intercept terms for each school represent the dependent variable adjusted for student intake characteristics, and which we attempt to explain with school-level characteristics. This between-school equation thus takes the following form:

${\beta }_{0j}={\theta }_{00}+{\theta }_{01}\left(\text{paternal imprisonment}\right)+\dots \text{.}+U_{0j},$

where θ ₀₀ is the school's overall average educational outcome, and θ ₀₁ is the regression coefficient of the effect of paternal imprisonment measured as a school-level mean score on the school's overall average educational outcome. The additional school-level covariate measures include maternal imprisonment, single-parent status, household income, area crime rate, school attendance and dropout rates, school size, urbanicity, public funding, number of full-time teachers, and number of teachers with master's degrees. We standardized the preceding variables to place these school-level measures on a common metric. We tested for significant cross-level interactions with race/ethnicity, and these were not statistically significant. U _0j is the school-level error term, assumed to be normally distributed with a variance of τ.

Because the model parameters are initially estimated separately for each school, the input characteristics are not assumed to have a constant effect across all schools, and this allows the HLM modeling to provide a more accurate representation of the complex multilevel error structure.

The Findings

From the resulting multilevel analysis, we learn not only about student-level sources of variation—including parental imprisonment—in educational outcomes, but also about the influence of variation in the aggregation and concentration of imprisoned mothers and fathers between schools on these outcomes, with the range of other variables taken into account at student and school levels. This allows us to move beyond the level of individuals to uniquely determine whether the aggregation and concentration of parental imprisonment at the school level has collective spillover effects on the educational outcomes of children resulting from historically high levels of incarceration of mothers and fathers in the United States (see Sampson & Loeffler 2010).

Table 2 summarizes the HLM results for high school GPA. Maternal imprisonment has significant negative effects on GPA both for individual students and for schools with relatively more mothers imprisoned. The student-level impact of maternal imprisonment is relatively large, negative, and statistically significant. The further spillover school-level effect of maternal imprisonment is also negative, net of the significant positive school-level effects of multiple teachers with master's degrees and high student attendance. The other significant student-level positive predictors of GPA are being female, having a college-educated father and mother, having a two-biological-parent family, and having a high household income. Being African American, as well as having a father or mother who smokes, negatively predicts student-level GPA. However, the significant negative student- and school-level effects of maternal imprisonment on high school GPA are net of all these other significant effects. Note that in this model the effect of paternal imprisonment is insignificant at both the student and the school levels.

Table 2. HLM Model of Predictors of Transcript Recorded Cumulative GPA across High School Years Attended with Robust Standard Errors

^* p < .05

^** p < .01

^*** p < .001

^† significant at p < .10

Reference Categories: ^aFemale = 1; Male = 0; ^bNon-Hispanic White; ^cAll other family types.

Table 3 summarizes the HLM results for the 13-point highest level of education completed by students. In this table, both paternal and maternal imprisonment now have significant negative effects on highest level of education achieved for individual students, as well as significant spillover effects in schools with both more fathers and more mothers imprisoned. While at the student level the effect of maternal imprisonment is much larger than the effect of paternal imprisonment, at the school level the effect of maternal imprisonment is about the same as that of paternal imprisonment. Mean household income also now has significant positive effects at both the student and the school levels. A high number of teachers with master's degrees again has a significant positive school-level effect, yet the effects of schools that have more mothers and fathers imprisoned are again comparatively strong at the school level. At the student level, being female and having a father and mother with a college education are again significantly positive. Maternal smoking is once again negative. Being African American has a significant positive effect with all these other variables in the equation, and this may reflect a residual affirmative action component. Again, the significant negative student- and school-level effects of paternal and maternal imprisonment on highest education level achieved are net of all these other significant effects, with the maternal effects larger than the paternal effects at the student level.

Table 3. HLM Model of Predictors of Respondent's Education Level at Wave IV with Robust Standard Errors

^* p < .05

^** p < .01

^*** p < .001

^† significant at p < .10

Reference Categories: ^aFemale = 1; Male = 0; ^bNon-Hispanic White; ^cAll other family types.

We turn finally in Table 4 to a logit model of the important divide between the college-educated and non-college-educated Americans in this sample of young adults. Both paternal and maternal imprisonment are repeated here as significant negative predictors of college graduation for individual students, and there are further significant negative spillover effects on college graduation in schools that have more imprisoned fathers and mothers. As in each of the previous models, significant amounts of teachers with master's degrees again have a positive school-level effect, as does the size of the school, while receiving public funding is a negative factor. In relative terms, the effects of schools that have relatively more mothers and fathers imprisoned are once again comparatively strong at the school level. At the student level, being female and having a father and mother with a college education are again positive and significant. So is a relatively high mean household income at the student level. Perceived closeness with one's biological mother is now also significant. And again, being African American has a significant positive effect with all the other variables in the equation, and again, this perhaps reflects a residual affirmative action effect in college admissions. Still again, as in previous tables, the significant negative student- and school-level effects of paternal and maternal imprisonment on college completion are net of all other significant effects.

Table 4. HGLM Model of Predictors of College Degree Completion at Wave IV (Population Average Model with Robust Standard Errors)

^* p < .05

^** p < .01

^*** p < .001

^† significant at p < .10

Reference Categories: ^aFemale = 1; Male = 0; ^bNon-Hispanic White; ^cAll other family types.

We graphically display the pronounced effects of student- and school-level parental imprisonment on the prospects of attaining a college degree in Figure 2. The bar graphs presented in this figure provide separate estimates of the student- and school-level effects of maternal and paternal imprisonment on college graduation by using the logistic regression equation in Table 4. All variables other than mother and father imprisonment before age 18 at the student and school levels are set at their mean values to estimate these bar graphs.

Figure 2. Proportion Graduating from College as a Function of Paternal and Maternal Imprisonment.

Figure 2 reveals the importance of mothers' imprisonment in childhood and adolescence as a barrier to the completion of college as a young adult. The far right side of Figure 2 shows a pervasive student-level effect of imprisonment of one's own mother that is relatively impervious to the effects of the imprisonment of other mothers in the school. Regardless of school-level variation, among the children of mothers who are imprisoned, the college graduation rate is between 1 and 2 percent. These children of incarcerated mothers have relatively little chance of successfully completing college. On the other hand, as we discuss below, among children whose mothers are not imprisoned, there is substantial variation in how the imprisonment of other mothers in their school affects their college graduation rates.

Overall, the college graduation rate in the United States is over 40 percent for children of parents who are not imprisoned during the children's childhood or adolescence and children who attend schools in which few other parents are incarcerated. However, this graduation rate drops to about 30 percent for children whose mothers are not imprisoned but who go to school where as few as 6 percent of other mothers are imprisoned. These are schools where the imprisonment rate of mothers is about one standard deviation above the mean. The small size of this standard deviation in combination with the significance of the school-level effect of maternal incarceration is an indication of the low threshold at which maternal imprisonment spillover consequences become notable for children whose mothers are not incarcerated but who attend these schools. When about 10 percent of the mothers in a school are imprisoned, the graduation rate drops to about 25 percent among youth whose mothers are not imprisoned.

The effects on college graduation rates of one's own father being imprisoned are smaller than those for imprisoned mothers and also more sensitive to the imprisonment of other fathers with children in the school, while the effects of the imprisonment of one's own mother are larger and thus less sensitive to the incarceration of other mothers. Furthermore, when one's own father and mother are not imprisoned, higher thresholds of other paternal compared to other maternal imprisonment are involved in producing similar levels of spillover effects on college graduation rates. As noted earlier, paternal imprisonment is also more common than maternal imprisonment.

If one's father is not imprisoned and 15 percent of other fathers in the school are imprisoned, the student college graduation rate is reduced to about 30 percent. If one's father is not imprisoned and 25 percent of other fathers in the school are imprisoned, the college graduation rate is reduced to about 20 percent. These reductions are obviously quite notable, but the thresholds indicated above for maternal spillover effects of similar magnitude are lower, which may be even more notable.

Also notable is that when one's own father is imprisoned, and 25 percent of other fathers in the school are imprisoned, the student college graduation rate drops to 10 percent. However, recall that among children whose own mothers are incarcerated, the likelihood of college graduation is far lower—indeed, only about 2 percent.

The more general point is that in high-incarceration schools in the United States, even among youth whose fathers and mothers are not incarcerated, the college graduation rate is reduced by as much as half. The more specific point is that among children of mothers who are imprisoned, the children's chances of incarceration are very low, and beyond this, when one's own mother is not imprisoned, the threshold for spillover effects of other mothers' imprisonment is lower than that for fathers. The educational outcomes for children tend to be more sensitive to maternal than paternal incarceration.

Conclusion

This article reports on the educational consequences for children of imprisoning their own mothers and fathers as well as the mothers and fathers of other children in the schools they attend. Thus we are concerned not only with individual sources of student variation in educational outcomes, but also with how the spillover effect of the highly concentrated increase in the imprisonment of parents in some school settings may disadvantage students in progressing to higher levels of educational attainment. These student and school effects can impose long-lasting social costs.

Several of the damaging predictions at the outset of this article are confirmed. The first is that the negative effect of imprisoning mothers is notably larger than that of imprisoning fathers in four of six comparisons of children's educational outcomes, although in a fifth comparison the effects have about the same size and significance, while paternal effects are marginally bigger in a sixth comparison. The effects of maternal imprisonment are especially clear and persistent at the individual student level, which is consistent with the assumption that mothers have a primary influence on their own children. However, the second confirmed prediction is that there is a further negative spillover effect on children's educational outcomes of incarcerating mothers in all three cases, as well as of imprisoning fathers in two of three cases—even when the mothers and fathers are not the children's own. The third confirmed prediction is that the threshold of the spillover impact of maternal imprisonment on children's educational outcomes is lower than that of paternal imprisonment. The fourth is that all these effects are highly consequential for the important educational outcome of college graduation.

College graduation is increasingly the educational credential that is most consequential for the occupational and therefore socioeconomic success of Americans who are entering and advancing through the labor market today. Absence of a college degree is an increasingly consequential barrier to upward mobility in American society. We have seen that the children of mothers who are imprisoned relatively rarely graduate from college. We have further seen that even attending a school where relatively few mothers are imprisoned notably reduces the likelihood that the children of other mothers will graduate from college. When as few as 6 percent of the mothers in a school are imprisoned, the overall rate of graduation for other children in the school is reduced from about 40 to 30 percent, and when 10 percent of the mothers in a school are imprisoned, the graduation rate is reduced to about 25 percent.

The legislation of federal and state sentencing guidelines in the 1980s, which radically reduced the discretion that previously allowed judges to take children into account in imposing less severe sentences for mothers, is a relatively new development. This change overturned more than two centuries of judicial autonomy in the United States. The current sentencing guidelines leave judges less room to consider women's family responsibilities or the social capital they provide their families and communities as mitigating factors that would encourage probation as an alternative to prison. These guidelines were a significant part of a “tough on crime,” law-and-order era that increased imprisonment of women and mothers. We did not find significant interactions between race and parental imprisonment effects: when black and white mothers come before the courts and are imprisoned, their children are about equally likely to experience diminished educational outcomes. Nonetheless, black women are disproportionately arrested and brought to these courts, and their children are therefore far more affected overall.

We have demonstrated the harmful consequences in terms of educational detainment associated with maternal incarceration. As the intergenerational consequences unfold for the children of the “prison generation,” it becomes possible and potentially important to map these outcomes back onto their exclusionary origins in a shortsighted regime of increased imprisonment. The analysis presented in this article investigates sources of both individual-level and spillover effects on intergenerational educational outcomes of children of incarcerated parents and children in schools with elevated numbers of incarcerated parents. We found strong evidence that the school-level threshold of maternal imprisonment effects on the educational outcomes of children—especially college completion—was notably lower than parallel effects of paternal imprisonment, which are nonetheless also significant. Our thesis was that this lower threshold of maternal effects is the product of the refusal to accommodate family differences and vulnerabilities of women in the demands of sentencing guidelines for imprisonment.

Two perspectives on the vulnerabilities and subsequent diminished educational outcomes resulting from parental incarceration at both the individual and the school levels stress interruption of parent-child relationships and the loss of educational and economic resources. An alternative perspective is that individuals and schools that experience parental incarceration are simply more vulnerable as a result of predispositions involving preexisting background risk characteristics. Yet we do not find that this latter possibility eliminates evidence of either of the previous two perspectives—especially the disrupted parent-child relationship perspective. Our measure of the interruption of the parent-child role relationship is simple and direct: the incarceration of a parent. We also include a range of measures of educational and economic resources, including individual and school-level measures of family income. We further include a range of measures of predisposing background differences, such as parental alcoholism and other nonnormative behaviors. Despite the range of these measures included in our analyses, the influence of the disruption of parent-child relationships involved in parental incarceration, and especially maternal incarceration, both at the individual and the school levels, is persistent and robust. We cannot conclude that our results are definitive, but they are certainly suggestive that the disruption of role relationships resulting from maternal as well as paternal incarceration is detrimental to the educational outcomes of children in adolescence and early adulthood.

Although justified in terms of legal equality norms, the cultural roots of the increased resort to maternal imprisonment likely lie in its use as a means to signal the repudiation of “bad mothers” who are accused of violating maternal role expectations. Legal equality norms embedded in the enforcement of state and federal sentencing guidelines mask and punish differences in gendered role expectations with damaging consequences that unfold later and intergenerationally in the diminished educational outcomes for children. Capturing the plight of incarcerated mothers and their children in the postguidelines era, Flavin (2009: 162; see also, for example, Carlen & Worrall 2004; Gomez 1997; Hagan & Petty 2001) notes that “without a fundamental shift in our approaches to punishment and parenthood, incarcerated women will continue to be scapegoated and widely assumed to be incompetent mothers, should their parenting be acknowledged at all.” Our research suggests the social costs of this punitive policy.

An implication of our research is that the emphasis placed on legal equality for women in criminal law requires reexamination. Meda Chesney-Lind's (2006) call for a renewed consideration of patriarchy, crime, and justice in an era of political backlash underlines the seriously problematic aspects of a contemporary “equality with a vengeance”—a form of equality that severely impacts women, mothers, and their children throughout the life cycle. Our research indicates that the norm of legal equality is counterproductive: it obscures empirically demonstrated consequences that call for a fuller appraisal of the social statuses of mothers and children relative to men in the circumstances of everyday family life.

Criminal courts neglect gender-specific rights of mothers and children, including the potential common ground on which the rights of parents as defendants might be more transparently aligned with the rights of children as victims in need of parental attention, care, and protection. Women's and progressive criminal justice organizations, as well as researchers, may underestimate these interconnections and their broad relevance to affected communities. We have seen that formal equality can lead to substantive inequality for women and children and that the gendered effects of the logic of judicial neutrality are socially costly not just for individuals and families, but also for the schools and the communities in which they are located across the nation.