Why We Need Improved Data to Better Understand the Consequences of Incarceration for American Families

Amanda Geller, New York University, Garrett T. Pace, University of Michigan

A growing body of research shows that having a parent imprisoned affects more than two million American children – more than two of every hundred – in various ways. But much remains to be learned about exactly how parental incarceration matters, and when and for whom the effects are most damaging or, alternatively, potentially protective. This research is hard to do, however, because data are limited. Here we review the types of data currently available for studies of fathers’ incarceration and its effects – and offer recommendations for future data collection.

Data Challenges and Availability

Many of the factors making parental incarceration an important issue for policy are the same factors that create challenges for data collection. Father incarceration is common; however, incarceration disproportionately affects racial and ethnic minorities in disadvantaged communities – populations very hard to reach and document with traditional social science data collection methods. Administrative sources of information can also be limited, because many fathers are noncustodial parents, and may not be their children’s legal guardians.

To date, research on U.S. parental incarceration has relied on three kinds of data: qualitative interviews, administrative data recorded by the criminal justice system and by employment and social service agencies, and data from surveys conducted on households sampled from the entire U.S. population. Each of these sources has real strengths. Qualitative research provides rich descriptions of the challenges faced by incarcerated men and their families. Administrative data track details of fathers’ criminal histories and time behind bars. Population-based surveys allow us to compare households and families that do and do not include fathers serving time in jail or prison. The most useful “longitudinal surveys” track the same families over time, allowing researchers to observe changes that happen both before and after fathers are incarcerated.

On their own, however, each of these data sources is limited. Qualitative studies tend to focus on small numbers of incarcerated fathers and their families, who are unlikely to be representative of all such fathers and families. These studies also cannot readily tell the difference between family changes caused by fathers’ imprisonment and those caused by other kinds of social disadvantage. Administrative datasets present challenges for family research, since many incarcerated fathers are not connected to their families in such records. Finally, longitudinal surveys struggle with high attrition rates when tracking disadvantaged families over time, while one-shot surveys contain only limited information about fathers’ imprisonment and its potential consequences.

Does It Help to Match Survey and Administrative Data?

Can researchers learn more by merging different information sources? In a pilot study, we assessed potential gains that could be achieved by combining different kinds of data sets – specifically by supplementing a leading household survey on children and families with data from an administrative database with state criminal history records. The household survey contained both mothers’ and fathers’ reports about fathers’ involvements with the criminal justice system; and the administrative database contained detailed records submitted by police departments for all adult arrests in the state for finger-printable misdemeanor or felony offenses. Using records for 333 fathers, files were matched using personal identifiers including first and last names, Social Security numbers, and date of birth. (Notably, respondents’ personal details were protected throughout the process: personal identities were not revealed to researchers, and the family details contained in the survey were not revealed to the state agency).

To assess the gains made by data matching, we looked closely at the families with incarceration histories indicated in the administrative data, the survey responses, or both. The value added by matching data was assessed based on the extent to which administrative data identified details of fathers’ arrests and incarceration that were not reported in the surveys (including when we resolved conflicting information gathered from mothers and fathers).

  • Of the 333 fathers included in the record search, 77 were also found in the administrative records, 55 of whom had arrests clearly timed before the relevant surveys happened. Notably, a substantial number of these fathers or their partners were not interviewed. Others had been recorded in the survey as “not incarcerated.” Many of these fathers reported that they had been charged with a crime but did not report convictions, and were therefore not asked about incarceration.
  • Still other fathers (or their partners) told the survey takers that they had been incarcerated, yet had no such indications in administrative records. This could happen due to misreporting, or because such fathers were arrested in another state, were incarcerated in a federal government or immigration facility, or had their arrests sealed for youthful status or other reasons, or were convicted of offenses that did not require finger printing.

In total, matching the state administrative records with household survey data increased the number of fathers known to have arrest histories by more than a fifth, and provided much more detail, including dates, charges, and specific custodial and non-custodial sentences.

Recommendations for Future Data Collection

Although administrative criminal records are not perfect, we strongly recommend combining such information with more comprehensive survey questions. Studies using such matching will require extra preparation and funding, as well as procedures for informed consent and protecting respondents’ personal data. Matching can be strengthened by surveys that ask fathers more detailed questions about when and where they were arrested and incarcerated, and whether their records are sealed. Administrative records, in turn, could include richer information about family ties. Although it must be done carefully with limitations in mind, data matching can clearly enhance the completeness and quality of research on the family effects of incarceration.

Read more in Amanda Geller, Kate Jaeger, and Garrett T. Pace, “Surveys, Records, and the Study of Incarceration in Families.” The ANNALS of the American Academy of Political and Social Science 665, no. 1 (2016): 22-43.
September 2016