As of 2014, 1,561,500 prisoners were held in state and federal correctional facilities in the United States. An additional 744,600 individuals were held in local jails. Although this was down slightly from 2013, the number of incarcerated persons has increased dramatically over the past 40 years. In 1980, there were only 302,377 prisoners in state and federal correctional facilities, and 163,994 in local jails. In addition, 37.4% of all incarcerated individuals were African-American, and 7.23% were female.

It is important to note, however, that many of these prisoners were parents. In a 2007 report prepared for the Annie E. Casey Foundation, 55% of prisoners in state penitentiaries and 63% of prisoners in federal penitentiaries had children under the age of 18 [1]. Given this large population of children who have incarcerated parents, it would be worthwhile to determine if the parents’ criminal backgrounds have any effects on the likelihood that their adult children will also be involved in criminal activity.

In order to determine the impact that parental incarceration may have on offspring arrests, several theories and results from recent studies regarding the effects of parental absence and parental influences must be considered. First, research has shown that spouses of incarcerated persons may not be able to adequately supervise their children. This increase in unsupervised time may result in increased opportunities for adolescent delinquency and a proclivity to criminal activity later in life [2]. Second, children who have absent fathers may have less attachment to them, thus resulting in emotional instability and lowered job prospects [3-6]. Third, absent fathers may increase the likelihood that male adolescents may leave home earlier than desired and may develop unhealthy associations with delinquent peers [7,8]. Fourth, when mothers go to prison, the children are usually placed in the custody of relatives who are sometimes unprepared both emotionally and financially to care for the children [9,10]. Hence, the children may suffer due to the loss of income and opportunities. Fifth, many incarcerated parents are housed in prisons far from their primary residences, and phone calls from prisons are typically very expensive [11]. Hence, the incarceration of a parent may strain the limited financial resources of the children’s caregivers, possibly resulting in less communication between the incarcerated parents and their children. Due to this reduced parental contact and because of the stigma attached to parental incarceration, young adults who have incarcerated parents may seek support and advice from peers, which may further contribute to the young adult’s delinquency [11]. All of the above suggest that parental incarceration may increase the likelihood that the adult child of an incarcerated parent will engage in some type of criminal activity in the future.

Although some research has examined the emotional, psychological, or economic effects of parental incarceration on children, very little research has been devoted to the effects of parental incarceration on the criminal behavior of adult offspring [9,12-24]. Much of this prior research is qualitative in nature or pertains to delinquent behavior on the part of minor children. The following review will only examine quantitative studies or those studies that examined the criminal behavior of adult children.

In Murray et al. [17] the authors used two data sets: one from a longitudinal survey of 15,000 persons living in Stockholm and the other from a survey of 400 males living in London. Both surveys were conducted prior to the 1980s; hence, the data are somewhat out-ofdate. In addition, the only explanatory variables that were used were social class and parental incarceration. The adult children were ages 19-30 years in this study. A logistic regression was used, and the results suggest that parental incarceration was a significant risk factor for offspring offending in both Sweden and England. Hence, the children of incarcerated parents are more likely to be convicted of a crime than others. However, parental incarceration was a much stronger risk factor in England than in Sweden. There are several possible reasons for this result. First, prison sentences were shorter in Sweden than in England during the period in question. Second, in Sweden, prisons were open; prisoners were able to go home every 3 or 4 months, mail was uncensored, and private (conjugal) visits were allowed. Third, in Sweden, more emphasis was placed on the rehabilitation rather than the punishment of juvenile offenders. Hence, given the above, it is possible these lenient prison policies may have mitigated the effects of parental incarceration on children.

A more narrowly focused study was conducted by Huebner and Gustafson [16]. In this study, the effects of maternal incarceration on adult children were examined. This study used data from the National Longitudinal Survey of Youth 1979 (NLSY79). As noted by the authors, the NLSY79 is the only nationally representative data set that examines relationships between mothers and their children, even in the context of incarceration and criminal behavior. Two dependent variables were used in this study: an adult conviction variable that equaled one if the adult child had a criminal conviction between the years 1994 and 2000, and an adult probation variable that equaled one if the adult child had been on probation during the period 1994-2000. Various control variables were used including an index of childhood delinquency, a maternal absence dummy variable, a maternal incarceration dummy variable, and several offspring demographic variables. Using a logistic regression, results of this study suggest that adults who had mothers who were incarcerated were four times more likely to be on probation and three times more likely to be convicted of a crime. Interestingly, childhood delinquency and race were insignificant factors, but gender was very significant; men were more likely to be on probation and to have been convicted of a crime than women.

Another study that used the NLSY79 data is Harper and McLanahan [20]. In this study, the focus was on the effects of paternal absence on offspring incarceration. Although not specifically examining the effects of paternal incarceration on offspring incarceration, this is one of the few studies to look at the effects of a father’s absence on the likelihood that a child will be involved in criminal activity. Using NLSY79 data, the authors constructed a longitudinal data set of offspring incarceration. A time-varying dummy variable was constructed in order to denote whether or not a child between the ages of 15 and 30 was incarcerated; this was used as the dependent variable. Various explanatory variables were used including childhood family structure, father absence, race, income, and other socioeconomic and demographic variables. Using a logistic regression, results indicated that children raised in father-only or mother-father family structures were much less likely to be convicted of a crime later in life. All other family structures increased the likelihood that a child would be convicted of a crime, including a father-stepmother structure.

Finally, Bryant et al. [24] looked at the effects of severe emotional disturbances on youth criminal activity. Although the primary focus of this study was on emotional disturbances, one of the explanatory variables used was parental incarceration. All individuals included in this data set were juvenile clients of the South Carolina Department of Social Services. Using a logistic regression, it was found that having a parent who was incarcerated greatly increased the likelihood that a child would be a minor offender, but parental incarceration had little effect on whether or not the child would be a major offender.

The purpose of the present study is to determine the effects of parental incarceration on adult offspring criminal behavior. The present study differs from prior research in several ways. First, this study uses the 1997 version of the National Longitudinal Survey of Youth, which is much more recent than the data used in any of the prior studies on this topic. Second, both maternal and paternal incarceration is used as explanatory variables. Third, only adult offspring (ages 21-34) are examined. Fourth, this study looks at arrests, not convictions. The next section discusses the empirical technique that is used in the present study.

In order to determine if parental incarceration has a significant impact on the probability that an offspring will have contact with the criminal justice system, guidance was obtained from Huebner and Gustafson [16] and Harper and McLanahan [20]. In Huebner and Gustafson [16], the dependent variables were adult offspring conviction and adult offspring probation. In Harper and McLanahan [20], the dependent variable was the incarceration of the adolescent or adult child. In the present study, a dummy variable denoting whether or not the adult offspring was arrested is used as the dependent variable. The reason for using an arrest dummy variable is because this variable will capture many more contacts between the adult child and the criminal justice system than conviction, probation, or incarceration. This study only examines adult children (ages 21-34).

Regarding explanatory variables, because only adult arrests will be examined, measures of juvenile delinquency (for example, peer influences when the child was an adolescent) will not be used as explanatory variables. Instead, most of the explanatory variables will be demographic and socioeconomic characteristics of the adult offspring, such as race gender, age, marital status, and educational attainment.

The only juvenile family structure variables that will be used as explanatory variables are the following:

Mother serve prison sentence: As far as you know, did your mother ever serve a prison sentence for a conviction before your 16^th birthday?

Father serve prison sentence: As far as you know, did your father ever serve a prison sentence for a conviction before your 16^th birthday?

Given the above, the following model is estimated in the present study:

Y_i=α0+α1 FATHER IN PRISON+α2 MOTHER IN PRISON+α3 CONTROL VARIABLES (1)

Y_i is a dummy variable which equals one if the individual was arrested and zero otherwise, FATHER IN PRISON is a dummy variable that equals one if the father was incarcerated before the child’s 16^th birthday and zero otherwise, MOTHER IN PRISON is a dummy variable that equals one if the mother was incarcerated before the child’s 16^th birthday and zero otherwise, and CONTROL VARIABLES is a vector of socioeconomic and demographic variables. It is expected that, given prevalent theories on the effects of parental absence on children, having incarcerated parents will increase the likelihood of being arrested. In the data set used in the present study, “incarceration” only refers to time served in prison for convictions, not time served while awaiting trial.

Control variables used in the present study include the following: gender, race, region of residence, ratio of income to poverty line, age, household size, employment status, marital status, educational attainment, and whether or not individual smokes. These variables were used in prior studies.

Data used in the present study were obtained from National Longitudinal Survey of Youth, 1997 (NLSY97). The NLSY was constructed to be a nationally representative sample of the civilian non-institutionalized population at the time of the initial survey in 1979. A second survey with a different cohort was started in 1997. The 1979 NLSY consisted of 12,686 young men and women who were between the ages of 14 and 22 when they were first surveyed in 1979. The 1997 NLSY consisted of 8,984 men and women between the ages of 12 and 16 Interviews with NLSY respondents were conducted annually until 2011 but since then have been conducted biannually. Each age-sex cohort is represented by a multi-stage probability sample drawn by the Bureau of the Census from a list of sampling areas that had been constructed for the Monthly Labor Survey. The NLSY employs extensive household interviews in the selected sampling areas in order to obtain as random and as representative a sample as possible. In the present study, data for the years 2006-2011, and 2013 were used. After deleting all missing observations, the sample size was 28,886. In the NLSY97, the arrest variable denotes whether or a not a respondent was arrested since their last interview. Means of variables are presented on Table 1. Regarding some of the more noteworthy statistics, 3.6% of individuals were arrested at some time during the survey period, 1.14% had a mother that was incarcerated, and 6.3% had a father that was incarcerated. The age range for individuals included in this data set is 21 to 34.

Table 1: Means of variables.

Since the dependent variable is binary, a logistic regression was used to estimate equation (1). In addition to using the entire sample, various subsamples were also examined. Those subsamples include the following: persons 30 and younger, persons 25 and younger, and persons 25 and older. These subsamples were examined in order to determine if parental incarceration had differing effects for different age groups. Results are presented on Tables 2-5. These results suggest that having a mother or father in prison greatly increases the likelihood that a person will be arrested. This result holds for the main sample and for all of the subsamples except for the 21-30 aged subsamples. For the 21-30 aged sample, having had a mother in prison has no significant effect on the likelihood that a person will be arrested. For the entire sample, the odds of a person being arrested are 1.35 times greater for a person whose father was incarcerated and 1.51 times greater for a person whose mother was incarcerated.

Table 2: Logistic regression results: Entire sample (ages 21-34).

Table 3: Logistic regression results: Ages 21-30.

Table 4: Logistic regression results: Ages 21-25.

Table 5: Logistic regression results: Ages 25-34.

Regarding the effects of the other explanatory variables on adult criminal behavior, single men who are unemployed and not college educated are more likely to be arrested than others. For the entire sample, men are 2.43 times more likely to be arrested than women. Another interesting result is that individuals who smoke are much more likely to be arrested than others. Smokers are 3.46 times more likely to be arrested than non-smokers.

Over 2 million people are incarcerated in the prisons and jails of the Unites States. A majority of these prisoners have children that are left in the care of family, friends, or the state. The impact of these separations, especially on young children, has been well-documented and well-researched. One potential impact that has not been extensively examined is the effect of parental incarceration on the likelihood that their adult children will also engage in criminal activity. While prior research has looked at the effects of parental incarceration on offspring convictions and incarcerations, the present study is the first study that examines the effect of parental incarceration on adult offspring arrests.

Using a very large data set compiled from the 1997 version of the NLSY, the results of the present study suggest that parental incarceration increases the probability that an adult child will be arrested. This result is significant because it suggests that criminal justice reform that seeks to reduce or eliminate prison sentences for certain nonviolent offenses may also reduce the criminal activity of future generations.

It is important to note, however, that parental incarceration is not the single most important factor in the determination of the criminal activity of young adults. Demographic and socioeconomic factors, such as employment, marital status, educational attainment, and even smoking, have much greater impacts on the likelihood that a young adult will commit a crime than parental incarceration. Therefore, public policies that increase employment and educational opportunities should be an integral part of any crime reduction effort. Given the lack of supporting empirical evidence in this regard, more research is warranted in this area.