Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Fam Relat. Author manuscript; available in PMC 2010 July 22.
Published in final edited form as:
Fam Relat. 2009 December 1; 58(5): 507–519.
doi:  10.1111/j.1741-3729.2009.00571.x
PMCID: PMC2908317

Mentoring Children With Incarcerated Parents: Implications for Research, Practice, and Policy


We investigated children and families who were participating in a mentoring program targeting children with incarcerated parents. Using multiple methods and informants, we explored the development of the mentoring relationship, challenges and benefits of mentoring children with incarcerated parents, and match termination in 57 mentor-child dyads. More than one-third of matches terminated during the first 6 months of participation. For those matches that continued to meet, however, children who saw their mentors more frequently exhibited fewer internalizing and externalizing symptoms. In monthly interviews with participants, themes emerged about challenges associated with mentoring and reasons for match termination. Implications for researchers, practitioners, and policymakers are discussed.

Keywords: high-risk children, incarceration, mentoring, termination, volunteer programs

As of 2007, more than 1.7 million children have a parent who is incarcerated in a state or federal prison (Glaze & Maruschak, 2008). Children of incarcerated parents are at risk for developing behavior (Murray & Farrington, 2005) and school problems (Arditti, Lambert-Shute, & Joest, 2003) and insecure attachment relationships (Poehlmann, 2005b). Many of these children experience multiple risks, making them an ideal target for interventions. Recently, policymakers have viewed formalized mentoring programs (e.g., Big Brothers/Big Sisters [BBBS]) that match at-risk youth with unrelated adult volunteers as one intervention effort that may address some of the needs of this population (Jucovy, 2003). The political popularity of mentoring may be due, in part, to the low cost of volunteer-run programs. Further, mentoring programs make intuitive sense – all children need guidance and support from role-models (Walker, 2007). Despite their popularity, such targeted programs have not been explored in the empirical literature. The purpose of the current study is to examine the development of mentoring relationships and children’s behavioral outcomes in the context of a mentoring program for children with incarcerated parents.

Mentoring Programs: An Application of Attachment Theory

Attachment theory provides one model for understanding how supportive adults can positively influence children who have a history of disruption or loss in the context of a parent’s incarceration. Early theorists emphasized the detrimental impact of separation from parents on children’s relationships and subsequent outcomes (Bowlby, 1979). Children with a history of unreliable care, significant disruption, or loss may develop insecure attachments. Children with insecure attachments hold views of themselves as unworthy of care and attention, and of others as untrustworthy and unreliable. Further, insecurity is considered a non-specific risk factor for later behavior problems (Deklyen & Greenberg, 2008). Children’s expectations of future relationships, or internal working models, are thought to result from past relationship experiences (Bowlby). For children who have experienced loss as a result of a parent’s incarceration, establishing close relationships may be particularly challenging and additionally complicated when children’s behavior problems interfere with this process. A new relationship with a caring and supportive adult, however, may result in gradual changes in a child’s internal working models (Ainsworth, 1989), thereby influencing the child’s expectations for future relationships and its behavioral outcomes. Non-parental adults such as mentors “can offer a model of care and support [that] may challenge the views that adolescents hold of adults as untrustworthy and of themselves as undeserving of attention and care” (Rhodes, Haight, & Briggs, 1999, p. 187).

Mentoring has been used to describe relationships within clinical, educational, recreational, and workplace settings (Bozeman & Feeney, 2007). Although definitions vary by setting and program, mentoring in the present study refers to frequent, one-on-one contact between an unrelated adult volunteer and a child, with the relationship characterized by mutual commitment, respect, and loyalty (Rhodes & DuBois, 2006). Whereas specific goals vary by program, most mentoring programs focus on reducing antisocial behaviors, improving school outcomes, and creating positive relationships with mentors (Tierney, Grossman, & Resch, 2000).

BBBS is a nationwide mentoring program that pairs unrelated adult volunteers with children aged 5 to 18 years from single-parent households. Participants commit to meeting two to four times per month for 1 year. BBBS has an extensive infrastructure, including intense volunteer screening, match criteria, and ongoing supervision (Furano, Roaf, Styles, & Branch, 1993). Tierney et al. (2000) conducted the largest randomized controlled study of BBBS to date with a sample of 959 youth between the ages of 10 and 16 years. After the first 18 months of program participation, mentored youth reported fewer externalizing behavior problems (e.g., physical aggression) compared to control youth, although internalizing symptoms (e.g., depression) were not measured. Positive outcomes are promoted when mentoring relationships are characterized as close, consistent, and long-lasting (Rhodes & DuBois, 2006). Despite their popularity, however, the efficacy and effectiveness of mentoring programs are not well understood, and research has led to mixed conclusions about their impact (e.g., Tierney et al.; Rhodes & DuBois). The relationship processes involved in mentoring children of incarcerated parents are particularly unclear, as little is known about if and how previous relationship disruption that occurred as a result of parental incarceration affects children’s experiences in a new relationship with a mentor.

Mentoring High-Risk Youth

In addition to challenges in their close relationships, many children affected by parental incarceration experience significant sociodemographic risks. They are likely to live in single-parent, impoverished households characterized by residential mobility, and their caregivers are likely to experience poor mental and physical health (Poehlmann, 2005a). Further, children with incarcerated parents are at risk of exhibiting emotional and behavior problems at home and school (e.g., Arditti et al., 2003; Murray & Farrington, 2005). These risks could influence program participation by limiting children’s ability to maintain consistent contact with mentors or contributing to match termination, ultimately challenging the implementation and effectiveness of programs.

Despite the growing literature on youth mentoring (Rhodes & DuBois, 2006), little is known about the highest-risk participants. Using a subset of data from the national BBBS study, Rhodes et al. (1999) examined mentoring for youth in foster care compared to youth raised by a biological parent. At baseline, foster parents were more likely to report wanting a mentor for their children because their children were insecure, did not trust other adults, or had poor social relationships. At follow-up, foster parents were more likely to report that their children’s social skills had improved and they were more trusting. Using reports from case managers, however, Rhodes et al.’s analyses revealed that there were no differences in the quality or intensity of the mentoring provided to children in foster care compared to youth raised by a biological parent.

Although the experiences of foster youth may parallel some of the challenges associated with parental incarceration (e.g., history of relationship disruption), there is almost no research examining mentoring programs targeting children with incarcerated parents. Research on the first major project for mentoring children of incarcerated parents (Amachi) is currently underway by Public/Private Ventures (Jucovy, 2003). Between 2001 and 2003, Amachi matched 726 children with mentors. In November 2003, only 47% of all matches were still active. The most common reasons for dissolution of matches were changes or disruptions in the child’s life. Although additional information about the effectiveness of Amachi is forthcoming, the preliminary report is the only published research on mentoring children of incarcerated parents to date. The current study adds to the paucity of literature on this topic by examining a different sample of children participating in a mentoring program for children of incarcerated parents.

Research Questions

The current study examined children’s participation in Mentoring Connections (MC), a Department of Health and Human Services (DHHS) funded mentoring program administered through BBBS, designed to serve children between 4 and 16 years of age with an incarcerated parent. A mixed method, concurrent nested longitudinal design (Creswell, Plano Clark, Gutmann, & Hanson, 2003) was used to examine the following research questions:

  1. What is the rate and context of match termination among children of incarcerated parents participating in a mentoring program? Is termination related to children’s relationships with their current caregivers and incarcerated parents or their behavior problems?
  2. During the first 6 months of participation in a mentoring program, what activities do matches engage in, how frequent is their contact, and what do participants report about the strengths and challenges of the mentoring relationship?
  3. Is the frequency or length of contact between mentors and children related to children’s behavior problems during the first 6 months of program participation? Is program participation related to changes in children’s behavior during this time?



Approval for the study was obtained from the Institutional Review Board at the University of Wisconsin. A BBBS case manager accepted referrals and matched children and mentors. Matches were made on the basis of the organization’s national standards, including participant gender, geographic proximity, and availability. Whenever possible, children and mentors were also matched on their preferences regarding age, race, and activities. At this agency, children are placed on a waiting list until they can be interviewed and again until they can be matched with a suitable mentor. Children referred to MC, however, were interviewed and matched with a mentor as soon as possible, without placement on a waiting list. After children were matched, caregivers and mentors signed informed consent forms to participate in the program evaluation, and children aged 10 years and older gave their written assent. Because participants were not enrolled in the evaluation until after the match was made, data regarding the length of time between referral and match for the current sample are not known. National BBSS statistics, however, indicate that it takes approximately 5 months to match a child with an adult volunteer through their organization (Tierney et al., 2000).

Caregivers, children, and mentors were followed on a monthly basis for 6 months. At the start of the program (intake), the case manager administered all measures. Caregivers completed a demographic form and behavior checklist. Children (aged 9 years and older) completed self-report measures assessing their feelings about their current caregivers and incarcerated parents. During each of the first 6 months of program participation, a research assistant contacted caregivers, children, and mentors by telephone and interviewed them separately about their experiences during the previous month. If the match terminated during the first 6 months, termination interviews were conducted with program participants. If participants could not be reached for more than 1 month, the case manager contacted the family and the mentor. If a participant missed an interview but could be reached the following month, the research assistant also asked about the participant’s experiences during the previous month. After 6 months of program participation, research assistants called active matches to schedule home visits. If caregivers or mentors could not be reached by phone, paperwork was sent to the home. If they did not respond, the researcher or the BBBS case manager followed up with a visit.


Any child between the age of 4 and 16 years who was referred to MC and matched with a mentor was invited to participate in the study, with a 100% research participation rate. Between March 2005 and November 2006, 57 matches were made. As a result of match termination, attrition, measurement reasons, and missing data, the sample size for some of the quantitative analyses is smaller than 57 (see Measures section). Attrition analyses revealed no significant differences between groups who did and did not complete 6-month visits for any reason (including match termination) on child, caregiver, family, and mentor variables.

Most of the mentors were single (53%, n = 30), White (61%, n = 35), female (72%, n = 41), and between 18 and 77 years old (M = 38, SD = 16). Most (72%, n = 41) mentors were employed, 16% (n = 9) were students, and 12% (n = 7) were retired or not working outside the home. Children ranged in age from 4 to 15 years (M = 9, SD = 3) and most were girls (60%, n = 34). The majority of children were identified as either Black (49%, n = 28) or multiracial (40%, n = 23). Most had an incarcerated father (86%, n = 49), though four children had incarcerated mothers, and four children experienced the incarceration of both parents. In this paper, we use the term “caregiver” to refer to the child’s primary caregiver. Thus, for the majority of the children who lived with their biological mothers (79%, n = 45), the term “caregiver” refers to the child’s mother. Six children were living with their grandmothers, two were living with aunts, three were in foster care, and one child resided with her father. This profile is similar to nationally representative samples of families affected by parental incarceration; most often incarceration occurs among minority families in which fathers are incarcerated and children are cared for by their biological mothers. On average, however, children in our sample were older than nationally representative samples of children with incarcerated parents (Glaze & Maruschak, 2008).

Families in our sample experienced many sociodemographic risks. The majority of families were living in poverty (n = 39, 68%) and receiving public assistance (n = 48, 84%). Nearly one-quarter (n = 13, 23%) of children were living in households with five or more people. Most caregivers were not married (n = 49, 86%), and 35% had less than a high school education (n = 20). On the basis of previous research with youth using a cumulative risk model (e.g., Sameroff, Gutman, & Peck, 2003), a risk index was created by giving one point for each of the following risks: five or more people living in the home, family is living in poverty, family is using public assistance, caregiver has less than a high school education, and caregiver is not married. If data were missing for one of the risk items, a score of zero was imputed. Nine children (16%) were missing data on one or more risk items. There were no differences between terminated and nonterminated matches on the number of missing risk items.


Monthly interviews

Semistructured interviews with caregivers, children, and mentors were conducted each month during the first 6 months of program participation. Interviews inquired about several areas in the child’s life. For example, caregivers and mentors were asked “What seems to be going particularly well with the child?” and “Have there been any problems with the child in the last month?” Participants were also asked to discuss the frequency and type of contact between the child and the mentor. Caregivers and children were asked to discuss their feelings about the child’s participation in the program, and mentors were asked a similar question regarding their own participation. Although we attempted monthly interviews with all of the children, we faced some challenges with our youngest participants, who found it difficult to communicate by telephone. In these instances, interview questions were reworded to be more developmentally appropriate and additional probes were used. For example, younger children were asked “What do you do with your mentor when you see her? Do you like to do that?”

A total of 252 interviews were conducted with mentors, 184 interviews with children, and 184 interviews with children’s caregivers across the first 6 months of the program. Because of the large number of interviews conducted, detailed notes and direct quotes were written by research assistants during interviews, rather than audio-taping the interviews. Following the interviews, research assistants expanded upon their notes. All interview notes were later entered using a spreadsheet software program and coded using the initial steps of a grounded-theory approach (LaRossa, 2005; Strauss & Corbin, 1997). Our analysis began with open coding of monthly interviews. We grouped initial indicators by common labels or concepts (e.g., instrumental support, family mobility). Next, axial and selective coding was used to identify emergent categories and themes (e.g., activities chosen by the mentor or child) and links between categories (e.g., family mobility and poverty issues permeated multiple categories of responses).

If a match terminated during the first 6 months of the program, participants completed a final interview. During these interviews, participants were asked to discuss match contact, their experiences in the program, and the reasons for ending the relationship. Quantitative analyses using the following measures supplemented the thematic content analysis of interview data.

Mentor-child relationship experiences

During monthly interviews, participants reported the number of times the match had met, the duration of each meeting, and the activities they engaged in during the previous month. Data regarding match contact from mentors and caregivers were highly correlated (Pearson correlations of .49, p < .05, to .88, p < .001). In an effort to have the most complete reports of contact, mentors’ reports were supplemented with caregivers’ reports for months when mentor data were missing. The average number of contacts and the average length of meetings were calculated. Monthly data were available for 52 (91%) matches.

Mentor-child relationship quality

After 6 months of participation, mentors completed the Mentor Survey (adapted from the Youth-Mentor Relationship Questionnaire; Rhodes, Reddy, Roffman, & Grossman, 2005). Mentors matched with children 7 years or older were asked to rate 14 statements focusing on the mentor’s perspective regarding the youth’s feelings of trust, support, and communication in the relationship on a 4-point scale ranging from 1 (not true at all ) to 4 (very true). For example, one item states “When my mentee is upset, I am able to comfort her.” Total scores ranged from 14 to 56, with higher scores indicating higher quality relationships. Cronbach’s alpha was .70. At intake, 37 mentors were matched with children 7 years or older, but by 6 months, 7 of these matches had terminated. We were unable to conduct a 6-month assessment with four of the mentors, resulting in 26 Mentor Surveys.

Children’s behavior problems

Children’s behavior problems were measured at intake and after 6 months using caregivers’ responses to the Child Behavior Checklist (CBCL; Achenbach, 1991). The CBCL contains 120 problem items rated on a 3-point scale: 0 (not true), 1 (somewhat or sometimes true), and 2 (very true or often true). The items fall into one of two broad-band scales: internalizing and externalizing problems. At intake, 18 (32%) CBCLs were missing because they were not administered by the BBBS case manager or because they were not returned by the caregiver, resulting in 39 CBCLs available for analysis. As a result of match termination and attrition, 18 CBCLs were available at 6 months.

Feelings about family relationships

The Inventory of Parent and Peer Attachment (IPPA; Armsden & Greenberg, 1987) was administered at intake to children (9 years and older) to assess their feelings about the caregiver-child and incarcerated parent-child relationships. The IPPA is a self-report measure containing 25 items that assess adolescents’ relationships with their parents. Children in the current study were asked to complete two versions of the form, one focusing on their current caregiver (IPPA-CG) and the other for their incarcerated parent (IPPA-IP). The wording of the IPPA was adjusted to reflect statements pertaining to the caregiver-child relationship (e.g., “caregiver” replaced “mother” when appropriate). At intake, 27 of the 31 children who were 9 years or older completed the IPPA-CG. Three of these children chose not to complete the IPPA-IP because they reported not having a relationship with the incarcerated parent. Composite scores were computed for both measures, with higher scores indicating more positive relationships. Cronbach’s alpha was .86 for the IPPA-CG and .92 for the IPPA-IP.


A mixed-methods approach was used to address each of the research questions. For each question, we present quantitative results first, followed by qualitative analyses to supplement the quantitative findings. Descriptive statistics and bivariate correlations are presented in Table 1. All quotes were used with permission and names have been changed to protect confidentiality.

Table 1
Correlations and Descriptive Statistics for Variables of Interest

Match Termination

Quantitative analysis

Of the 57 matches enrolled in our sample, more than one-third (n = 18; 32%) of the matches were not meeting 6 months after the start of the program. Of these 18 prematurely terminated matches, 5 (9%) terminated within the first month of the program. Four additional matches (7%) terminated after 2 or 3 months of program participation and nine more matches (16%) terminated between 4 and 6 months after the start of the program.

To test for differences between terminated matches and matches that were still meeting after 6 months, chi-square analyses were performed with categorical variables and a series of one-way ANOVAs was performed with continuous variables. These analyses revealed no statistically significant group differences on any mentor characteristics, including gender, age, ethnicity, or employment status. Further, there were no differences on child gender, age, ethnicity, kin relation to caregiver, or behavior problems assessed at intake. When compared on IPPA-IP and IPPA-CG scores at intake, there were no differences between terminated and active matches. Further analyses revealed no group differences on caregiver marital status, education, ethnicity, and household income. Analyses revealed that children who did not experience match termination, however, experienced more cumulative family risks, F(1, 55) = 4.87, p = .03. When individual risk variables, however, were compared to children who experienced premature match termination, there were no group differences.

Context and reasons for termination

Through interviews, we documented five primary reasons for premature match termination in the MC program: scheduling conflicts, personal or family issues, residential mobility, mentors underestimating the commitment, and match incompatibility. These categories were not mutually exclusive; participants often listed multiple challenges in the relationship that ultimately resulted in the match’s termination. Although none of the participants indicated that parental incarceration was the primary reason for match termination, some participants discussed how parental incarceration was indirectly associated with challenges in the mentoring relationship. For example, some mentors discussed how children’s weekend visits with their incarcerated parents limited the time that mentors and children could spend together. In another case, the BBBS match specialist cited the parent’s release from prison and changes in the child’s living arrangements as particularly stressful for the child and the match, eventually resulting in the match’s termination.

Inconsistency in meetings and scheduling difficulties were commonly discussed as challenges and reasons for termination. One mentor said, “It’s getting hard to see my little sister. She is booked Monday through Thursday with school stuff and Saturdays are family days.” Scheduling difficulties were often related to other issues within the child’s family. Several mentors expressed concerns that telephone numbers changed frequently and that caregivers were inconsistent or forgetful of meeting times. One mentor said, “I’ve tried calling the past two to three weeks and my calls have not been returned. Last time I tried to meet, I prepared a lunch and went to pick Lauren up, but her mom had forgotten.” Another mentor said “I suspect other mentors are having this problem as well – when a family goes through challenges [such as the incarceration of a parent], it really makes our meeting together even harder to accomplish.”

Monthly interview data revealed that children in the program experienced a number of family transitions, including moves to different homes and schools, changes in caregivers, births and deaths in the family, and the remarriage of a primary caregiver. These challenges were often cited as reasons for match termination. When children changed caregivers, it was most often a result of the parent’s incarceration. Most other transitions, however, were not directly attributed to the parent’s incarceration; rather, they were more often described as either indirect consequences of incarceration or related to sociodemographic risk factors. Families moved frequently and in two cases, children ran away from home. One mentor said, “He is one of many kids, so he may not get much attention at home.” Another mentor said, “I haven’t been in touch with Lee in the past month because he moved.”

Several participants discussed problems with mentors underestimating the level of commitment that was required by the program or the expectations that children and caregivers might have for the mentors. One mentor said, “I have not met with my mentee in the last month … I knew that consistency was important, but I kidded myself that I was going to stick to making those appointments.” Many mentors felt that children’s homes were crowded, chaotic, and typically in flux – dynamics that presented unexpected challenges for some mentors working with this population. One mentor expressed concerns related to the child’s living situation stating, “It’s unclear to me who works. There is not food in the house and it’s in disarray. There are always people in and out and a lot of yelling. The family is stressed and life is chaotic.” Another mentor said, “Damen is kind of ‘homeless.’ He is in a very unstable living situation. I’m trying to help him find a job, but his friends are bad influences. I give him $20 a week so he doesn’t get money by stealing or selling drugs.”

Mentors also expressed financial concerns related to mentoring. BBBS emphasizes that mentoring does not require money and seeks to provide matches with free or reduced-cost activities. Yet many mentors expressed concerns that the cost of weekly activities was burdensome or that finding fun, yet inexpensive, activities was difficult. One mentor said, “If Sarah was not expecting me to blow big bucks on her all the time, I could think of all kinds of things to do that don’t take much money, but she doesn’t want to do most of those things.”

Mentors also expressed concerns about the child viewing the relationship as primarily instrumental in nature and concerns that the child may have viewed the mentor as “someone who buys him something.” Some children seemed to view the mentor as an adult who provided material goods, rather than viewing the mentor as an adult who provided emotional or social support. One mentor said, “I’m disheartened that Liz doesn’t seem enthusiastic about seeing me except for when I spend money on her.” Additionally, mentors expressed concerns about feeling pressure to provide monetary support to the children and their families in an effort to ameliorate some of the families’ financial hardships. Families may have seen the mentor as someone who was able to provide these supports. One mentor expressed these concerns stating, “I get the impression that Lexi’s grandmother is prompting her to ask me for money.”

Finally, participants cited incompatibility or the child not wanting a mentor as reasons for match dissolution. Often matches dissolved because the mentor and child did not get along or did not enjoy the same activities. One caregiver said, “They just didn’t ‘click’ – their personalities were just different.” One child said, “We didn’t do anything fun … she wasn’t any fun.” Another child said, “I didn’t want a mentor in the first place.”

The Nature of Match Contact

Quantitative analysis

On average, during the first 6 months of program participation, mentors and children met three times per month (M = 2.91, SD = 0.93) and spent approximately 2.5 hours (M = 2.57, SD = 0.76) together per meeting. During one month, matches spent an average of 8.3 hours together (SD = 4.08). We also examined associations between match contact and mentors’ reports of the relationship on the Mentor Survey. As a result of missing data and match termination, however, the number of Mentor Surveys was smaller than anticipated. Therefore, we examined bivariate correlations rather than conducting regression analyses. Bivariate correlations revealed no significant associations between match contact variables and Mentor Survey total scores (see Table 1). We did find, however, that children spent more time with their mentors (r = .28, p = .04) and met with their mentors more frequently (r = .31, p = .03) when the children experienced more sociodemographic risk factors.

Match experiences: Activities and relationship development

Through interviews with mentors, caregivers, and children, it was evident that each mentor-child relationship experienced a variety of strengths and challenges during the first 6 months of the mentoring program. Through analysis of interviews, categories emerged reflecting matches’ activities, relationship development, and increased feelings of closeness, trust, and disclosure.

Matches engaged in a variety of activities such as going to movies, participating in seasonal activities (e.g., sledding), shopping, going out to eat or making a meal together, working on projects (e.g., knitting), or working on academic activities (e.g., homework). Matches that completed a 6-month visit reported participating in activities that both the mentor and the child enjoyed. Further, when the mentor and the child were both engaged and equally invested in the activities, participants reported more positive feelings about the relationship. For instance, after attending a pottery class together, one girl said, “Kay is really nice and she does good choices.” An older boy said, “We go to the gym and work out. I like the things Mark plans.”

Many participants experienced difficulties at the beginning of the relationship, including issues related to children’s early discomfort and emerging trust in the relationship. One mentor said, “It took Max the first four months to really open up.” Another mentor said, “Beth has been more apprehensive lately, possibly because her dad’s out of jail. I just get the sense that Beth isn’t comfortable with me. I asked her if she wanted a new mentor, but she said no … I’m comfortable with Beth, but she seems apprehensive.” In some cases, children’s early challenges with the mentoring relationship were related to their experiences of parental incarceration and disruption in their close relationships. One caregiver said, “He has a history of disappointment,” referring to the repeated disruptions in the child’s relationship with his incarcerated mother.

Themes related to the development of the mentoring relationship also emerged, especially after children had continued in the program for four or more months. One mentor said, “I can see her getting more comfortable with me as the weeks go by.” Another mentor said, “I think Ciarra feels very close to me. If we’re out somewhere, she always waits for me.” When asked what was going well with the child, one mentor said, “Jayla seems to trust people more, is more social, less moody, better at asking and speaking her mind.” One caregiver said, “She trusts Alison, enjoys her and looks forward to seeing her.” It was uncommon for children to directly report feeling closer or more trusting of their mentors. Instead, children were more likely to discuss enjoyment of the activities and characteristics of the mentor. One child said, “Rachel is the best. She is fun and nice.” After 6 months of participation, another child said, “Max gives me advice when I need it. He’s funny and nice to hang out with.”

Match Contact and Children’s Behavior Problems

Quantitative analyses

As a result of match termination and attrition, our sample size at 6 months was considerably smaller than anticipated. Thus, the following analyses are exploratory and should be interpreted with appropriate caution. To examine associations between frequency of match contact and children’s behavior problems, we conducted partial correlations between match contact and children’s externalizing and internalizing behavior problems. In both analyses, initial behavior problems and the risk index were entered as control variables. More frequent contact during the first 6 months in the program was associated with fewer externalizing symptoms at 6 months (as rated by caregivers), even after controlling for initial symptoms and sociodemographic risks (partial r = −.60, p = .02). Further, more frequent contact with mentors was associated with caregivers’ ratings of fewer internalizing symptoms at 6 months, after controlling for initial symptoms and sociodemographic risks (partial r = −.58, p = .03). To assess potential changes in children’s behavior during the first 6 months of participation, we also examined mean differences in caregivers’ reports of internalizing and externalizing symptoms via paired sample t-tests at intake and 6 months. These analyses revealed no differences in symptoms during children’s first 6 months of participation.

Changes in children’s behavior problems

During interviews, caregivers often reported noticing changes in children’s behaviors (despite our quantitative finding that mean levels of internalizing and externalizing symptoms did not change between intake and 6 months). Most often, caregivers attributed these changes to the child’s continued participation in the program. For instance, one caregiver said, “John is helping Jay find a job and he’s a father figure, Jay really looks up to him. John is encouraging Jay to see his dad and that’s good. I’ve noticed an attitude change for the better.” Another caregiver said, “Dayna’s behavior has really improved. She used to cry every time she was away from me, but since the program, she’s ‘grown up’ and doesn’t cry about that anymore.” Mentors also reported changes in children’s behavior over the course of their relationship. One mentor said, “We’ve worked on Janella’s reading, which has improved. She also seems less fearful of certain things since we started meeting.” Several children reported doing well in school. One child said, “Joe really helps me with school and band and it’s going really well.” A teenage boy said, “I’m learning to control my anger.”


Despite the growing number of children affected by parental incarceration and their heightened risk for adverse outcomes, few studies have focused on intervention programs targeting this population. In this mixed method study focusing on children of incarcerated parents in the context of a mentoring program, three main findings emerged. First, nearly one-third of the matches terminated within the first 6 months of participation in the program, although children and mentors in terminated matches did not differ from successful matches on intake characteristics other than family sociodemographic risk. Several common reasons for match termination emerged through qualitative analysis, however, including scheduling conflicts, family issues, residential mobility, mentors’ underestimation of the commitment required, and match incompatibility. The study’s second finding was that children experienced a number of sociodemographic risks and family transitions that created challenges for mentors. Finally, although initially difficult to establish, many children developed feelings of trust and closeness toward their mentors after continued participation in the program. Further, children who continued to participate in the program for 6 months and who met with their mentors more frequently exhibited fewer internalizing and externalizing symptoms.

The high rate of early termination in this mentoring program is a cause for concern. Although the national evaluation of BBBS (Tierney et al., 2000) did not report a termination rate at 6 months, 39% of matches in their sample were no longer meeting at 18 months. In the Amachi sample (Jucovy, 2003), 13% of matches terminated before 6 months and an additional 25% of matches terminated between 6 months and 1 year after the start of the mentoring program. Whereas the consequences of termination have been acknowledged and studied in other significant adult-child relationships (e.g., therapeutic relationships; see Bembry & Ericson, 1999), scholars have paid minimal attention to this dynamic in the context of mentoring relationships. In our analyses, children in terminated matches did not differ on any intake characteristics other than family sociodemographic risk when compared to children whose matches continued. Previous research, however, has found that older children and children with a history of psychological treatment or educational remediation are at higher risk for match termination (Grossman & Rhodes, 2002). We did not find age or behavior problems at intake to be risk factors for early termination, although we did not measure all aspects of children’s psychosocial adjustment.

Other research with this sample has documented the complexities of family relationships, contact with incarcerated parents, and children’s behavioral outcomes (Shlafer & Poehlmann, in press). Although attachment theory posits that children with less than optimal relationships with their incarcerated parents and their current caregivers may have a more difficult time forming and maintaining relationships with mentors, our analyses did not support this explanation, as there were no significant associations between IPPA scores and match termination. There are methodological problems with using self-reports of attachment security, however, (Kerns, Schlegelmilch, Morgan, & Abraham, 2005), and the IPPA may not have adequately captured the complex nature of children’s close relationships in this sample. In addition, the IPPA was only completed by children aged 9 and older; disruptions may have a more substantial effect on younger than on older children. Further, in our qualitative analysis of interview data, we found that caregivers discussed concerns related to children’s difficulties forming and maintaining trusting relationships, consistent with concerns reported by foster parents in Rhodes and colleagues’ (1999) sample. Researchers and practitioners targeting this population must consider how children’s past relationship experiences contribute to their current functioning, particularly in the context of a program that seeks to improve children’s behavioral outcomes through a relationship-based intervention. Practitioners should be mindful of children’s histories and how these experiences potentially inhibit children’s progress in the mentoring relationship.

In our sample, caregivers and mentors expressed concerns about children’s trust and comfort in the new relationship at the start of the program; however, over time these concerns lessened. After several months of participation in the program, caregivers and mentors discussed how mentoring relationships progressed and deepened over time. These themes are consistent with previous research describing the development and characteristics of mentoring relationships (e.g., Rhodes, 2002), suggesting that the processes through which mentoring relationships develop may not be specific to children of incarcerated parents. Challenges in children’s close relationships with caregivers and incarcerated parents, however, (Shlafer & Poehlmann, in press) may complicate the development of the mentoring relationship for children with incarcerated parents (e.g., behavioral reactions to changes in caregiving environments).

Our quantitative analyses revealed that the frequency of match contact in our sample was consistent with data from other mentoring programs targeting children of incarcerated parents (Jucovy, 2003). In the Amachi sample, mentors spent an average of 7.3 hours per month with their mentees, and they met on average twice per month. Similarly, participants in MC met less than the required one meeting per week, but they met for longer than the required 1 hour per meeting, possibly to make up for missed weeks. Further, children who continued participation in the program and who met with their mentors more frequently exhibited fewer internalizing and externalizing behavior problems, even after controlling for children’s behavior problems at intake, although mean levels of behavior problems across children remained unchanged. These analyses represent only those children who continued to participate in the program, however, as we were not able to follow matches post-termination. It is unknown whether children whose matches terminated experienced changes in their behaviors as a result of the termination (e.g., acting out, increased sadness). To ensure that program participation has not caused more harm than good, it is critical for researchers to examine children’s outcomes beyond termination.

Although our analyses revealed that children and mentors in terminated matches did not differ from successful matches on the majority of intake characteristics, matches were less likely to terminate if children’s families experienced more cumulative sociodemographic risk factors. This was somewhat surprising given the qualitative themes that emerged regarding challenges that mentors experienced working with this high-risk population. One possible explanation for this finding is that mentors may have been more committed to continuing relationships with the most high-risk children because mentors believed that these children and their families needed the most support. Families experiencing more sociodemographic risks may also have been more committed to continuing mentoring for their children because of higher levels of need. Indeed, children from families with more sociodemographic risks also spent more time with their mentors and met with their mentors more frequently. In addition, program staff may have responded differently to the highest-risk families, potentially resulting in selection effects. Program staff may have matched the most competent and committed mentors with the highest-risk children or provided more support to these matches. Future research should consider how these and other participant and program characteristics relate to the development and sustainability of mentoring relationships with children of incarcerated parents.

The contemporary and delayed correlates of early termination may be especially salient for children of incarcerated parents because of their history of disrupted relationships and, thus, more research should explore this side of mentoring. Although mentoring could potentially have a positive impact, a relationship that terminates early may reinforce messages that these children cannot depend on or trust adults for love and support. National studies estimate that the benefits of mentoring relationships become evident after 12 months; however, these same studies estimate that nearly 40% of matches terminate before 18 months (Tierney et al., 2000). These factors beg the question: How should mentoring programs be planned and implemented so that they promote positive outcomes while avoiding potential harm? Although children with incarcerated parents could benefit from a well-organized and consistent mentoring relationship, comprehensive services that address the risk factors experienced as a result of parental incarceration across the multiple ecological systems (e.g., limited financial resources, mental health, barriers to contact with inmates) may be more beneficial to children and their families (Arditti, 2005). Despite their popularity, mentoring programs with high rates of match termination among the most vulnerable children could be both fiscally and socially irresponsible.

Interviews with participants suggested that structural and logistical factors were critical to match termination, including family instability, scheduling conflicts, family problems, and match incompatibility. In addition, some mentors underestimated the commitments required to work with this population, and many families had inaccurate expectations of the instrumental support that mentors could provide. Mentoring programs targeting this population should be especially aware of the challenges these families experience and provide additional services to help mentoring relationships be more successful (i.e., provide additional training to mentors regarding families’ needs, provide monetary support to help cover activities and meals). Our findings highlight areas that could be improved through implementation of best practices in mentoring outlined by DuBois and colleagues (DuBois, Holloway, Valentine, & Cooper, 2002), including parent involvement, providing structured activities, providing ongoing training, monitoring program implementation, and choosing mentors who have a background in a helping role.

Implementation of these best practices could address some of the issues that led to termination, including inaccurate expectations of mentors’ roles, scheduling difficulties, and mentor stress. Programs need to allocate funds to provide specialized training for mentors (e.g., how to handle issues related to child-incarcerated parent contact, social stigma related to parental incarceration) and ongoing support for matches (e.g., strategies for maintaining consistent contact despite no-shows and scheduling difficulties). In addition, site-based (e.g., school- or clinic-based) mentoring programs could facilitate scheduling meetings with a population that is characterized by instability and could also provide additional support (e.g., academic tutoring or counseling) to children with incarcerated parents. Socioeconomic differences between mentors and families may have contributed to some of the challenges mentors experienced about the expenses related to their participation. More training could help mentors understand the risks incurred by children of incarcerated parents, as well as how mentors might deal with difficult situations (e.g., when caregivers ask for money, family transitions related to incarceration).

Bowlby (1979) wrote that humans are “happiest and able to deploy their talents to best advantage when they are confident that, standing behind them, there are one or more trusted persons who will come to their aid should difficulties arise” (p. 103). Although he was not referring to mentors, recent applications of attachment theory suggest that mentors may be able to fulfill some attachment-related functions. Mentors who create long-lasting, meaningful relationships with children and cultivate feelings of trust and safety can be considered attachment figures (Ainsworth, 1989). Although programs strive to create this type of relationship with each match, few reach the level of closeness and commitment characteristic of an attachment. Even if the relationships do not become attachments, however, future research should use attachment theory to help illuminate the processes through which mentoring relationships develop and end.


The results of this study should be considered exploratory and viewed with caution for several reasons. Because we did not include a comparison group, only within-group variation was explored. Additionally, high rates of attrition and match termination limit the study’s findings in a number of ways. Although it was not possible to rule out potential selection bias as a result of match termination and attrition, all of the individuals who were matched in the program agreed to participate in the study. Moreover, although there was only one statistically significant difference in intake characteristics between matches that participated at 6 months and matches that did not, there were a number of unmeasured characteristics that could potentially relate to continued participation in the program (e.g., caregiver’s mental health). Finally, as a result of attrition and match termination, we were unable to make full use of the longitudinal data in our quantitative analyses. In addition to match termination and attrition, data were missing for other reasons. At intake, several CBCLs were missing because the BBBS case manager did not administer the measure, reflecting some of the difficulties of conducting research with community agencies serving high-risk families.


This study is one of the first to explore mentoring children with incarcerated parents. Federal initiatives to extend mentoring to high-risk children are a first step, but programs must do more to understand and address these children’s specific needs. Our conclusions echo those of other researchers who have cautioned against the unfettered expansion of mentoring programs (e.g., Rhodes & DuBois, 2006). To ensure that mentoring programs are most effective, they should be theoretically grounded, rigorously evaluated using randomized controlled designs, and assess children’s relationships with mentors and family members. Although many policymakers have embraced mentoring, scholarly work has not been commensurate with this enthusiasm. We are hopeful that this study can provide some initial insights for researchers and practitioners.


Support for this work was provided by DHHS (Grant # HHS-2004-ACF-ACYF-CU-0001) and a traineeship through NIMH (Grant # 5T32MH015755).


  • Achenbach TM. Manual for the Youth Self-Report and 1991 profile. University of Vermont, Department of Psychiatry; Burlington, VT: 1991.
  • Ainsworth M. Attachments beyond infancy. American Psychologist. 1989;44:709–716. [PubMed]
  • Arditti JA. Families and incarceration: An ecological approach. Families in Society: The Journal of Contemporary Social Services. 2005;86:251–260.
  • Arditti JA, Lambert-Shute J, Joest K. Saturday morning at the jail: Implications of incarceration for families and children. Family Relations. 2003;52:195–204.
  • Armsden GC, Greenberg MT. The Inventory of Parent and Peer Attachment: Relationships to well-being in adolescence. Journal of Youth and Adolescence. 1987;16:427–454. [PubMed]
  • Bembry J, Ericson C. Therapeutic termination with the early adolescent who has experienced multiple losses. Child & Adolescent Social Work Journal. 1999;16:177–189.
  • Bowlby J. The making and breaking of affectional bonds. Tavistock; London, England: 1979.
  • Bozeman B, Feeney MK. Toward a useful theory of mentoring: A conceptual analysis and critique. Administration and Society. 2007;39:719–739.
  • Creswell JW, Plano Clark VL, Gutmann ML, Hanson WE. Advanced mixed methods research designs. In: Tashakkori A, Teddlie CB, editors. Handbook of mixed methods social and behavioral research. Sage; Thousand Oaks, CA: 2003. pp. 209–240.
  • Deklyen M, Greenberg MT. Attachment and psychopathology in childhood. In: Cassidy J, Shaver PR, editors. Handbook of attachment. 2nd ed Guilford Press; New York: 2008. pp. 637–665.
  • DuBois DL, Holloway BE, Valentine JC, Cooper H. Effectiveness of mentoring programs for youth: A meta-analytic review. American Journal of Community Psychology. 2002;30:157–197. [PubMed]
  • Furano K, Roaf P, Styles M, Branch A. Big Brothers/Big Sisters: A study of program practices. Public/Private Ventures; Philadelphia: 1993.
  • Glaze LE, Maruschak LM. Parents in prison and their minor children (NCJ 222984) Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics; 2008.
  • Grossman JB, Rhodes JE. The test of time: Predictors and effects of duration in youth mentoring relationships. American Journal of Community Psychology. 2002;30:199–219. [PubMed]
  • Jucovy L. Amachi: Mentoring children of prisoners in Philadelphia. Public/Private Ventures; Philadelphia: 2003.
  • Kerns KA, Schlegelmilch A, Morgan TA, Abraham MM. Assessing attachment in middle childhood. In: Kerns KA, Richardson RA, editors. Attachment in middle childhood. Guilford Press; New York: 2005. pp. 46–70.
  • LaRossa R. Grounded theory methods and qualitative family research. Journal of Marriage and Family. 2005;67:837–857.
  • Murray J, Farrington DP. Parental imprisonment: Effects on boys’ antisocial behaviour and delinquency through the lifecourse. Journal of Child Psychology and Psychiatry. 2005;46:1269–1278. [PubMed]
  • Poehlmann J. Children’s family environments and intellectual outcomes during maternal incarceration. Journal of Marriage and Family. 2005a;67:1275–1285.
  • Poehlmann J. Representations of attachment relationships in children of incarcerated mothers. Child Development. 2005b;76:679–696. [PubMed]
  • Rhodes J, DuBois D. Understanding and facilitating the youth mentoring movement. Social Policy Report. 2006;20:3–19.
  • Rhodes JE. Stand by me: The risks and rewards of mentoring today’s youth. Harvard University Press; Cambridge, MA: 2002.
  • Rhodes JE, Haight WL, Briggs EC. The influence of mentoring on the peer relationships of foster youth in relative and nonrelative care. Journal of Research on Adolescence. 1999;9:185–201.
  • Rhodes JE, Reddy R, Roffman J, Grossman JB. Promoting successful youth mentoring relationships: A preliminary screening questionnaire. Journal of Primary Prevention. 2005;26:147–167. [PubMed]
  • Sameroff AJ, Gutman LM, Peck S. Adaptation among youth facing multiple risks: Prospective research findings. In: Luthar SS, editor. Resilience and vulnerability: Adaptation in the context of childhood. Cambridge University Press; New York: 2003. pp. 364–391.
  • Shlafer RJ, Poehlmann J. Attachment and caregiving relationships in families affected by parental incarceration. Attachment and Human Development. (in press) [PMC free article] [PubMed]
  • Strauss AL, Corbin JM. Grounded theory in practice. Sage; Thousand Oaks, CA: 1997.
  • Tierney JP, Grossman JB, Resch N. Making a difference: An impact study of Big Brothers/Big Sisters (re-issue 1995 study) Public/Private Ventures; Philadelphia: 2000.
  • Walker G. Mentoring, policy, and politics. Public/Private Ventures; Philadelphia: 2007.