Keywords

At the close of 2015, nearly 2.2 million adults were incarcerated in federal, state, and local prisons and jails, and another 4.7 million people were under the surveillance of probation or parole agencies in the USA (Kaeble & Glaze, 2016). While both the number and fraction of adults incarcerated have fallen from peak levels observed in late 2008, incarceration rates in the USA continue to be dramatically higher than those in other countries, and exposure to incarceration is pervasive in some socio-demographic groups. Figure 2.1 shows incarceration rates in 2015 in the USA compared with rates in western Europe. Americans are more than ten times as likely to be in prison or jail as people living in Denmark, Sweden, and the Netherlands and four times more likely than residents of the UK (Pettit & Sykes, 2017).

Fig. 2.1
figure 1

Source US rates are from Kaeble and Glaze (2016); European rates are from Aebi et al. (2016)

Incarceration rates in selected western European nations and the USA, 2015.

Full size image

Simple counts of the number of people incarcerated or the percentage of the population in prison or jail do not show the extent to which contact with the criminal justice system is stratified by race and ethnicity. In the USA, incarceration is disproportionately concentrated among African-American and Latino men, particularly those with low levels of formal schooling. Table 2.1 presents estimates of adult exposure to incarceration by race and ethnicity. In 1985, eight-tenths of one percent of non-Hispanic white men age 20–34 were incarcerated in jails and prisons, compared to 5.9% of non-Hispanic black men and 2.3% of Hispanic men. By the end of 2015, approximately 1.6, 9.1, and 3.9% of young, white, black, and Hispanic men were incarcerated on any given day, respectively.

Table 2.1 Exposure to íncarceration by race and ethnicity, men age 20–34
Full size table

Growth in the criminal justice system over the last half century and its disproportionate concentration among disadvantaged groups has spawned increasing interest in research on punishment and inequality, particularly on the effects of criminal justice contact for individuals, families, and communities. Scholars routinely find that adult exposure to the criminal justice system has labor market, financial, educational, health, romantic, and political consequences for people with criminal records (Pettit & Western, 2004; Johnson & Raphael, 2009; Maroto, 2015; Massoglia, 2008; Pager, 2007; Pettit, 2012; Sykes & Maroto, 2016; Uggen & Manza, 2002; Western, 2006).

The repercussions of incarceration are not limited to current and former inmates. Growing concern about mass incarceration has resulted in a proliferation of research on the consequences of parental incarceration for children and families. Over the last decade, research has shown that maternal, paternal, and parental incarceration are associated with a host of negative outcomes for children, including lower academic achievement, grade retention, and educational discontinuation (Cho 2009a, b, 2011; Hagan & Foster, 2012a, b; Haskins, 2016; Turney & Haskins, 2014) and greater likelihoods of material hardship, economic disadvantage, and severe deprivation (Geller, Garfinkel, & Western, 2011; Hagan & Foster, 2015; Schwartz-Soicher, Geller, & Garfinkel, 2011; Sugie, 2012; Sykes & Pettit, 2015). These consequences and hardships strain bonds between parents and children (Arditti, 2012; Braman, 2004; Comfort, 2008; Waller, 2002; Western, Lopoo, & McLanahan, 2004) and can fuel intergenerational inequalities (Foster & Hagan, 2007, 2015).

Yet, despite the expanding body of research on childhood exposure to incarceration and its effects, few studies have assessed how differences in data and methods across studies may influence estimates of children’s exposure to parental incarceration and effects of parental incarceration on child outcomes as well as their associated implications. In this chapter, we explore how methods of data collection and analysis influence the measurement of children’s exposure to parental incarceration and its consequences. We begin with a discussion of conventional data sources employed to measure criminal justice contact and its correlates, and we examine how different data collection strategies influence estimates of the overall level of childhood exposure to parental incarceration and differences across socio-demographic groups. Next we review some of the recent literature on the effects of parental incarceration on child outcomes to illustrate how and why different data sources and methods influence observed findings. We conclude with recommendations for future directions in research and practice.

Measuring Children’s Exposure to Parental Incarceration

Studies designed to estimate children’s exposure to parental incarceration using a range of different data sources have generated three consistent findings: (1) There is growth in children’s exposure to parental incarceration over time; (2) there is a cumulative increase in exposure to parental incarceration over the life-course; and (3) there is inequality in children’s exposure to parental incarceration, both over time and over the life-course, across social and demographic groups. However, estimates of the percentage of children experiencing parental incarceration, and inequality in exposure to parental incarceration, vary in relation to sampling design, choice of respondent, and question wording in ways that may have important implications for research on the effects of parental incarceration on child outcomes and social inequality more generally.

Point-in-Time Estimates

Children’s exposure to parental incarceration is typically measured either at a point-in-time or as a lifetime risk. Point-in-time measures of children’s exposure to parental incarceration using data from Surveys of Inmates of State and Federal Correctional Facilities estimated that nearly 1.5 million minor children in the USA had a parent in state or federal prison in 1999 (Mumola, 2000). These estimates are derived from survey questions about the number and ages of biological children of inmates housed in state and federal correctional facilities, weighted by the total number of inmates incarcerated in the respective facility types (Pettit, Sykes, and Western, 2009). Similarly constructed estimates that also include the biological children of inmates housed in local jails suggest many more, or closer to 2.1 million children, had a biological parent incarcerated in any type of correctional facility at the turn of the century (Sykes & Pettit, 2014).

Figure 2.2 demonstrates that the number and percentage of children with a parent currently incarcerated in a federal, state, or local correctional facility has grown along with penal expansion [see also Glaze and Maruschak (2010) for a discussion of growth in the number of children with a parent in state or federal prison]. Our estimates suggest that at the end of 2015, 2.5 million children had a parent incarcerated in a federal, state, or local correctional facility. It may be important to keep in mind that estimates including children with a parent in a local jail facility are typically a third or more higher than estimates that limit attention to children of parents incarcerated in only state and federal correctional facilities. Estimates of parental exposure to the criminal justice system, more generally, are even higher. One recent study suggests that nearly half of American children have a parent with an arrest record (Vallas, Boteach, West, & Odum, 2015).

Fig. 2.2
figure 2

Source Authors’ calculations from the Surveys of Inmates, Bureau of Justice Statistics Annual Inmate Counts, and the Current Population Survey

Number and percentage of children with a parent incarcerated, USA 1980–2015.

Full size image

Data from surveys of inmates can be used to estimate racial and ethnic inequalities in children’s exposure to having a parent in prison or jail. Parental incarceration is much more common for Black and Hispanic children than for non-Hispanic White children, although the exact magnitude of racial and ethnic inequality depends on the scope of inquiry. More expansive measures of criminal justice contact—such as those that include short stints in local jails—tend to show higher levels of system involvement but lower levels of racial and ethnic inequality in exposure to incarceration for a number of different reasons related to criminal justice processing and respondent reporting. Drawing on data from Surveys of Inmates in State and Federal Correctional Facilities, Glaze and Maruschak (2010) and Mumola (2000) find that Black children are 7.5–9 times more likely than White children, and Hispanic children are 2.5–3 times as likely as White children, to have a parent in prison. More expansive measures of parental incarceration that include parents housed in local jail facilities confirm racial inequality in exposure to parental incarceration but exhibit smaller differences between racial and ethnic groups in comparable years.

Table 2.2 displays race and ethnic inequalities in children’s exposure to having a parent incarcerated, including parents housed in local jail facilities. In 1985, six-tenths of one percent of White children had a parent incarcerated in prison or jail, compared to 4.1% of Black children and 2.0% of Hispanic children. Estimates from 2015 are much higher than those recorded 30 years earlier and racial inequality in parental incarceration persists. In 2015, parental incarceration rates for Hispanic children were approximately twice as high as for White children, while Black children were over five times more likely than White children to have a parent incarcerated.

Table 2.2 Childhood exposure to parental incarceration by age 18
Full size table

Lifetime Risk Estimates

Children’s exposure to parental incarceration can also be measured as a lifetime risk, or the chance that a child or children within a specified group has been exposed to having a parent incarcerated in a given period of time. Lifetime risks of exposure to parental incarceration have been generated using data from a number of different sources and vary quite significantly. Although studies that estimate lifetime risks of parental incarceration largely agree on the trends over time and across cohorts, discordant estimates of children’s exposure to parental incarceration have been attributed to differences in sampling strategies across data sources, question wording or scope conditions, and who responds to the questionnaire (e.g., mother, father, or child) (see, e.g., Sykes & Pettit, 2014). These differences not only influence estimates of children’s risk of exposure to parental incarceration but may also have important implications for studies investigating the effects of parental incarceration on child outcomes, as we will discuss in greater detail below.

Data from the Surveys of Inmates in State and Federal Correctional Facilities (SISFCF) have been used to provide estimates of children’s risk of ever having a parent imprisoned in a federal or state correctional facility. When the SISFCF data are combined with inmate totals from the Bureau of Justice Statistics (BJS), birth cohort counts from the Detailed Natality Files, and population counts from the Current Population Survey (CPS), children’s exposure to parental incarceration can be estimated using life-table methods (Mueller & Wildeman, 2016; Sykes & Pettit, 2014; Wildeman, 2009). Wildeman (2009) used this method to estimate the fraction of children born in 1978 and 1990 that could expect to have a parent incarcerated by age 14. Other studies have applied these methods to estimate children’s risk of parental incarceration in later birth cohorts and to age 18 (see Mueller & Wildeman, 2016; Sykes & Pettit, 2014).

Lifetime risks of parental incarceration are, by definition, always higher than point-in-time estimates generated by the same data because they represent cumulative exposure to having a parent incarcerated. Wildeman (2009) relied on data gathered from people housed in state and federal correctional facilities and found that 1 in 25 White children and 1 in 4 Black children born in 1990 could expect to have a parent spend at least a year in a state or federal correctional facility before his/her fourteenth birthday. Subsequent estimates confirm that lifetime risks of parental imprisonment, by age 14 or 18, are much higher for all racial groups than point-in-time estimates listed in Table 2.2 (Sykes & Pettit, 2014; Wakefield & Wildeman, 2013; Wildeman, 2009).

Table 2.3 illustrates these and other estimates of exposure to parental incarceration generated by commonly used data sources. Row 1 reports estimates generated by Wildeman (2009) using data from the Survey of Inmates and population counts, and row 2 reports estimates from Sykes and Pettit (2014) using the same data and method applied to later birth cohorts and through age 17 (i.e., up to age 18). Both sets of estimates underscore that exposure to parental incarceration has grown over time and that the cumulative increase in the lifetime risk of parental incarceration is observed by early (age 14) and late (age 17) adolescence. Further, both sets of estimates illustrate large differences in children’s risk of having a parent incarcerated, with Black children 6 to 7 times more likely to be exposed to parental incarceration than White children.

Table 2.3 Exposure to parental incarceration in different studies relying on cohort and cross-sectional data
Full size table

Comparing estimates from other studies included in Table 2.3 illustrates the salience of differences in sampling design, questionnaire wording, and respondents and their effects on estimates of children’s exposure to parental incarceration. Row 3 reports estimates of children’s exposure to parental incarceration generated from data gathered through the National Survey of Children’s Health (NSCH) 2011–2012. The NSCH randomly samples telephone numbers in the USA to locate households with children aged 0–17 years. Within each household, one child was selected at random to be the subject of interview. Unlike previous iterations of the NSCH, the 2011–12 survey included a special supplement on adverse childhood experiences that inquired about parental incarceration, exposure to violence, and other markers of disadvantage and thus provides an opportunity to generate nationally representative estimates of children’s exposure to parental incarceration from a large sample survey.

Row 3 in Table 2.3 shows that the NSCH asked respondents “Did the focal child ever live with a parent or guardian who served time in jail or prison after the child was born?” Compared to estimates of the lifetime risk of parental imprisonment generated by the surveys of inmates, the NSCH data generate higher estimated risks of exposure to parental incarceration for non-Hispanic White children and lower estimated risks of exposure for non-Hispanic Black and Hispanic children. Further exploration indicates that differences in estimates can be reconciled, at least partially, by adjusting for the length of childhood exposure and information about whether the parent incarcerated co-resided with the child prior to his/her incapacitation (Sykes & Pettit, 2014). Adjusting for these factors helps to explain observed differences in estimates generated by the inmate surveys and the NSCH for Black and Hispanic children. However, adjustments aggravated differences between surveys for White children, suggesting that White youth exposed to parental incarceration are either overrepresented in the NSCH or overreport exposure to parental incarceration in comparison to estimates generated from inmate surveys.

Row 4 in Table 2.3 shows estimates of exposure to parental incarceration from the Fragile Families and Child Wellbeing Study (FFCWS), a data source commonly used to study the effects of parental incarceration on child well-being. The FFCWS follows a cohort of nearly 4,900 children born between 1998 and 2000 in 20 large US cities with different welfare policies and labor market conditions (Reichman, Teitler, Garfinkel, & McLanahan, 2001). Roughly three-quarters of births in FFCWS were to unmarried parents. The baseline survey asked both mothers and fathers about father’s incarceration histories [see Geller, Cooper, Garfinkel, Schwartz-Soicher and Ronald (2012) for a summary of differential reporting across survey waves]. These data illustrate how mothers and fathers may differentially experience, and differentially report, contact with the criminal justice system. Moreover, there is evidence that respondents in the FFCWS may significantly underreport contact with the criminal justice system. Geller, Jaeger, and Pace (2016) augment the FFCWS data in one city with administrative records from that state’s criminal justice agency. After matching the survey and administrative records, they found that the number of fathers with criminal justice involvement increased by more than 20% in that city. It is unclear whether, or to what extent, data from the other cities in the FFCWS underestimate exposure to parental incarceration. Yet, data show that nearly 50% of children in the FFCWS had ever experienced paternal incarceration by age 9, or Wave 5 of the survey (Haskins & Jacobsen, 2017).

Finally, Row 5 in Table 2.3 shows estimates of parental incarceration generated by data from the National Longitudinal Study of Adolescent Health (AddHealth). AddHealth is a nationally representative sample of adolescents in grades 7–12 in the USA during the 1994–95 school year. The survey includes students from 132 schools in 80 different communities. More than 90,000 students completed in school questionnaires between September 1994 and April 1995 (Harris, 2013). In 2001–2002 (Wave 3), AddHealth began collecting information on whether the respondent’s biological father ever served time in jail or prison. Foster and Hagan (2007) estimate that 12% of respondents had a biological father ever incarcerated.

Each of the data sources described in Table 2.3 has been used to estimate children’s exposure to parental incarceration or other dimensions of criminal justice contact, differences across social and demographic groups in the risk of exposure to parental incarceration, and the consequences of parental incarceration on child outcomes. There is general agreement across studies that children’s exposure to incarceration has increased in concert with growth in incarceration. However, children’s exposure to parental incarceration measured over the life-course is much higher than exposure measured at any point-in-time. Additionally, exposure to parental incarceration is disproportionately concentrated among Black and Hispanic children, for both lifetime risks and point-in-time estimates. There are reasons to think that estimates from some surveys are better, more valid, indicators of underlying levels and differences in exposure to parental incarceration in the general population. For example, estimates from the NSCH are more likely to closely approximate the prevalence of parental incarceration in the USA than cohort-based studies like FFCWS and AddHealth. Yet each of these data sources have important strengths that have made them valuable resources for the study of the effects of parental incarceration on a wide range of child outcomes.

At the same time, however, differences in aspects of sampling design and measurement produce important differences in estimates of children’s exposure to parental incarceration and racial and ethnic inequalities in exposure across surveys. These differences are important to recognize, and better to reconcile, in order to fully understand how exposure to parental incarceration affects children. For example, with respect to point-in-time estimates of parental incarceration, Mumola (2000) and Glaze and Maruschak (2010) focused attention on the number of children with a parent in a state or federal prison. Much of our work, in contrast, has sought to draw attention to those children as well as children with parents incarcerated in local jails (see, e.g., Sykes & Pettit, 2014). Where and when a parent is incarcerated can have important implications for the effects of incarceration on child outcomes and there are important reasons that researchers may preference different measures. State and federal prison terms are typically longer than one year, and parents in prison may be housed long distances from their children. These factors may present significant obstacles and/or expenses to maintaining contact with children. In contrast, jail stays are usually less than a year and parents in jail are more likely to be housed closer to their home and/or children (Braman, 2004; Comfort, 2008), providing opportunities for more frequent or regular contact. At the same time, jail stays may be associated with significant churning, or movement in and out of jail for short periods of time, leading to family instability and increased uncertainty in children’s lives (see, e.g., Comfort, 2008).

Differences in question wording help explain some of the variability in measures used to estimate children’s lifetime risk of parental incarceration generated by surveys like the NSCH, FFCWS, and AddHealth. For example, the NSCH asks “Did the focal child ever live with a parent or guardian who served time in jail or prison after the child was born?” This measure leaves open the possibility that non-focal children in the household may or may not have also experienced parental incarceration. If all children in residence were subject to parental incarceration, then the prevalence of parental incarceration is not underestimated in the NSCH. Yet, if the focal child did not experience parental incarceration but non-focal siblings were exposed to parental incarceration—especially if non-focal siblings are older than the focal child—then estimates of parental incarceration are underestimated in the NSCH.

In the FFCWS baseline questionnaire, paternal incarceration is initially measured by a series of mother and father reports on his presence in jail and if interviewers conducted the survey in a correctional facility. However, the type of facility and the length of time incarcerated were not asked so it was not possible to disentangle short stints in jail from long prison stays until Year 1. Unfortunately, facility-type distinctions were discontinued by Year 5 of the survey. The FFCWS began collecting data on the timing and length of incarceration in “jail/prison” by Year 1. Yet, the frequency of jail and prison stays was not recorded between the Year 1 and Year 9 surveys. This is unfortunate, as cumulative disadvantage may be most severe for children who have parents repeatedly exposed to carceral churning.

Although AddHealth has been a vital source of data for the study of parental incarceration on adolescent outcomes, it also includes a fairly broad measure of criminal justice contact that may lead to underestimates of racial and ethnic inequality in exposure to parental incarceration and its effects for child outcomes. Foster and Hagan (2007) note that in Wave 3 of AddHealth, “nearly twelve percent of the sampled youth reported their biological fathers ‘had served time in jail or prison.’ The timing, frequency, and duration of these incarcerations is unknown” (p. 408). The omnibus measure of parental incarceration in AddHealth may also help to explain lower levels of racial and ethnic inequality in exposure to parental incarceration in AddHealth. More expansive measures of criminal justice contact typically show lower levels of racial inequality in exposure. In the next section of the paper, we explore how these measurement differences affect substantive findings in research on parental incarceration.

How Differences in Data and Methods May Help Reconcile Divergent Findings

Differences in survey design are important for understanding observed differences in children’s exposure to parental incarceration across surveys as well as estimated inequalities in children’s exposure to parental incarceration. These differences may also help to explain divergent findings, with respect to the relationship between parental incarceration and children’s outcomes. To draw attention to these issues, we focus on how features of data and method may help explain discrepant research findings on parental incarceration and educational attainment.

Recent studies use data from AddHealth and FFCWS to examine whether and how parental incarceration influences children’s educational progress at different points in the life-course. Drawing on AddHealth data, Foster and Hagan (2007) show that having a father incarcerated is negatively associated with years of education, and that paternal incarceration also results in lower grade point averages for children (Foster & Hagan, 2009). Yet, research relying on data from the FFCWS provides mixed evidence for the effects of parental incarceration on measures of childhood development and early educational progress (Geller et al., 2012; Haskins, 2014, 2016; Wildeman & Turney, 2014). How does one resolve these discordant findings between surveys?

In a recent paper published in Demography, Turney (2017) attempts to reconcile these findings using data from the FFCWS to estimate variation in the effects of paternal incarceration on children’s problem behaviors and cognitive skills in middle childhood (i.e., to age 9). By taking into account children’s differential risk of experiencing paternal incarceration—as measured by father’s residential status, family poverty, and neighborhood disadvantage—she is able to explore a variety of child outcomes for children exposed to high, medium, and low probabilities of paternal incarceration. Importantly, Turney (2017) finds that “the null average effects on cognitive skills are consistent with the null test scores among younger children (Geller et al., 2012; Haskins, 2014), but they are inconsistent with the negative average effects on children’s high school grade point averages (Foster & Hagan, 2007) and other educational outcomes (Murray, Loeber, & Pardini 2012). Together, these findings suggest that the average consequences of paternal incarceration for children’s cognitive skills may increase as children progress through school (also see Turney & Haskins, 2014), and future research should directly consider this possibly” (p. 382). Thus, divergent findings of educational inequality in the lives of children exposed to paternal incarceration may be resolved if researchers examine grade progressions across the life-course.

Turney’s (2017) hypothesis about academic achievement worsening with grade advancement is certainly plausible and worth exploring, and her study was rigorously executed and is very convincing. To her hypothesis, we posit an additional possibility: The sampling frames of these data sources are too different to compare across study findings. The FFCWS is a stratified sample, where the first stage of sampling is based on welfare policy regimes and local labor market characteristics (Reichman et al., 2001), while the first stage of the stratified sample in AddHealth—from which Foster and Hagan (2007, 2015) draw their conclusions—is based on secondary educational institutions (Harris, 2013). While it is possible that academic achievement worsens over a child’s life-course if exposed to paternal incarceration, AddHealth sampled schools and then adolescents while the FFCWS sampled welfare and economic contexts of cities and then hospitals and births. Thus, the units were sampled from different conceptions of the population and thus should not be compared without adjusting for sampling differences between surveys. Furthermore, the youth in both surveys aged through the educational system during different periods of carceral growth, which may obscure the relationship between exposure to parental incarceration and academic achievement for a specific grade during a particular point-in-time.

Another possibility for these divergent findings is that conventional surveys underestimate the number of children exposed to parental incarceration and do so in ways that have important implications for the relationship between parental incarceration and child outcomes. The undercounting of people with criminal records in social surveys may introduce discrepancies and bias in survey estimates of parental incarceration, thereby lowering estimated differences in effect sizes and compromising significance tests between children exposed and unexposed to parental incarceration. Geller et al. (2016) have observed the undercounting of fathers with criminal records for one of the cities in the FFCWS. To address problems of undercounting in sample surveys, Sykes and Maroto (2016) developed a method of adjusting national survey sampling weights for differences in adult exposure to incarceration using the Survey of Inmates and other population-based data sources. Their method highlights how both effect sizes and significance tests are impacted by relying solely on survey measures that may underestimate exposure to incarceration (see also Western, Braga, Hureau, & Sirois 2016 on bias in effect sizes and significance tests when survey respondents are missing in subsequent waves of longitudinal data). Sykes and Maroto’s method can be extended to adjust sampling weights in other national surveys that measure exposure to parental incarceration. It is also worth noting that similar problems arise when surveys underestimate racial inequality in exposure if those surveys use an omnibus measure of criminal justice contact or parental incarceration. In either case, researchers should consider sampling designs as potential explanations for divergent findings and exercise care in scoping their conclusions based on these issues.

Conclusions and Future Directions

Millions of American children are exposed to parental incarceration every year, and having a parent incarcerated in jail or prison has become a defining feature in the lives of a disproportionate number of African-American and Latino children. The collateral consequences of parental incarceration on child outcomes, in a variety of domains, are increasingly well-documented. Yet, less attention has been paid to how the data and methods—from sampling design to the conceptualization and measurement of parental incarceration—may influence estimates of children’s exposure to parental incarceration and the effects of parental incarceration on child outcomes. In this chapter, we have provided a careful consideration of whether and how sampling design, choice of respondent, and question wording influence estimates of children’s exposure to parental incarceration, the relationship between parental incarceration and child outcomes, and explanations for the effects of parental incarceration on children’s academic achievement.

Differences in data and methods are most consequential, in our view, not simply for understanding the prevalence and consequences of parental incarceration in the aggregate but are most important when trying to determine between (or within) group differences in exposure to parental incarceration and its effects. Sampling methods and survey designs of commonly used studies can obscure racial and ethnic inequalities in parental incarceration and its effects on children.

We have a few key recommendations for future research and practice. Future research should be more attentive to how features of data and method influence the effects of parental incarceration and their implications for accounts of inequality. Turney’s (2017) hypothesis about differential risks and treatment effects across the life-course is important and illuminating. However, an inquiry into these differential risks may require further exploration into how the sampling designs of different data sources may themselves produce differential effects. For instance, it could be that the sampling designs of various surveys either miss particular groups of people because they are not attached to households (see Pettit, 2012) or because the initial stage of a stratified sample is based on city contextual attributes (i.e., welfare policies and economic conditions), educational institutions, or correctional facilities. Decomposing how much of an outcome is due to how the analytical unit was sampled, as well as how the outcome was measured across different surveys, would clarify a great deal of discordant findings in the literature.

Second, future research should investigate how changes in social policies after the Great Recession impacted the consequences of parental incarceration for children. Many of the current data sources, especially the Surveys of Inmates, are more than a decade old, and many of the cohort-based studies selected samples well before the Great Recession. Since the collapse of the housing and financial markets in 2007, states have devised a number of programs and policies to decarcerate state prisons and jails, and the youth in a number of these studies are now teenagers or young adults. New data collection efforts must be planned to understand how these policy shifts after the recession have impacted long-term exposure to, and differential effects of, parental incarceration in America for children born during the twenty-first century.

Third, practitioners and policymakers should carefully consider the objectives of their interventions and proposed policy solutions in light of existing data limitations and measurement differences. If, for example, policymakers and practitioners seek to establish short-term policies and programs targeting children with parents incarcerated, decision-makers should consider relying on population-based point-in-time estimates of parental incarceration. Similarly, policymakers and practitioners intervening in specific domains of social life or for specific issues (e.g., educational retention or health limitations) should rely on data and findings that have their samples drawn from those ecological contexts (i.e., from schools or health services).

Finally, research should focus on additional types of exposures to the criminal justice system. Much of the literature explores exposure and effects associated with incarceration, but little is known about the fate of children with parents who were arrested but not convicted or convicted but not incarcerated. The net widening of criminal justice contact may also influence child well-being through its effects on parents, either directly or indirectly. These areas of research are important and are beginning to be investigated (see Maroto & Sykes, Forthcoming; Sugie & Turney, 2017). The sequence of criminal justice contact—being surveilled, stopped, arrested, plead/tried, convicted, and incarcerated—should be carefully considered in the design and implementation of new studies focused on how incarceration, and the carceral state more generally, infiltrates the lives of children and structures inequalities across generations.