Informal Mentoring and Young Adult Employment

Abstract

This study explores the role of informal mentoring (i.e., developing an important relationship with a non-parental adult) in the transition to full time employment among young adults (age 23-28). Multivariate analysis of the Add Health data reveals that mentoring is positively related to the likelihood of full time employment, and the relationship involves both selection and causation processes. Entrance into the world of work facilitates the development of mentoring relationships, especially among youth who identify work-related mentors after adolescence. These relationships have the potential for promoting attachment to the labor force. Mentoring relationships that develop outside of work settings and during adolescence have a positive impact on the odds of full time employment. The receipt of guidance and advice from mentors, as well as access to weak-tied mentoring relationships, teacher mentors, and friend mentors all contribute to the increased odds of employment in young adulthood. However, adolescent mentoring may be less effective among young women than it is among young men.

Having a mentor may enable young people to succeed in the transition to adulthood. Those who identify important non-parental adults in their lives tend to report better psychological well-being, more rewarding relationships with others, academic success, and fewer problem behaviors than their peers (Bryant and Zimmerman, 2003; Greenberger, Chen, and Beam, 1998; Klaw, Rhodes, and Fitzgerald, 2003; Rhodes, Contreras, and Mangelsdorf, 1994; Rhodes, Ebert, and Fischer, 1992; Zimmerman, Bingenheimer, and Notaro, 2002).

Few studies have examined the role of mentors in the lives of young people as they enter the work force. The acquisition and maintenance of stable employment is one of the most important markers of the transition to adulthood, and one of the most difficult transitions to make given current economic conditions. During the recent economic recovery, older workers have experienced an increase in their rate of employment, while youth employment has steadily declined to historic lows (Sum, Khatiwada, McLaughlin, and Palma, 2005). With few institutional mechanisms facilitating the transition to stable employment in the United States (Kerckhoff, 2003), youth must rely on informal help from the important people in their lives or else fend for themselves.

Despite the potential importance of mentors in this transition, investigators have yet to examine in detail the extent to which informal mentoring promotes employment in young adulthood. Research on social networks (Lin, 1999) and on mentoring within work organizations (Ragins, 1999; Russell and Adams, 1997) highlight the importance of significant others in improving career opportunities for the adult workforce. Informal mentoring during the adolescent years is likely to have a positive impact on employment in young adulthood. Since employment experiences in young adulthood often forecast later attainment patterns (Blau and Duncan, 1967), understanding the factors that shape the transition to employment may help to explain why youth follow divergent occupational pathways. In this way, informal mentoring may serve as a hidden mechanism of advantage with long-term consequences.

Using data from the National Longitudinal Survey of Adolescent Health (Add Health), we find that mentors play an important role in the transition to employment in young adulthood. The relationship between mentoring and employment entails both selection and causation. On the one hand, entrance into the world of work can facilitate the development of mentoring relationships, further strengthening attachment to the labor market. On the other, adolescent mentoring has the potential for directly influencing the chances of adult employment. Mentors can expand the labor market information and opportunities of youth and enhance their chances for employment through guidance. However, the influence of informal mentoring on the lives of young people is dependent on its timing and may vary by gender.

Mentors in Young Lives

Research on adolescent mentoring distinguishes between formal and informal mentoring relationships. Formal relationships refer to programs such as the Big Brothers/Big Sisters which attempt to match at-risk youth with adult mentors. These types of programs generally facilitate positive outcomes among youth (DuBois, Holloway, Valentine, and Cooper, 2002), improving academic performance, attitudes, and relationships with friends and relatives as well as reducing problem behaviors (Grossman and Rhodes, 2002; Rhodes, Grossman, and Resch, 2000; Tierney, Grossman, and Resch, 1995).

Informal mentoring relationships occur naturally among youth and the adults with whom they come in contact. These non-parental adults have received a number of labels, such as natural mentors, very important people (VIPs), role models, and significant others (Bö, 1989; Bryant and Zimmerman, 2003; Greenberger, Chen, and Beam, 1998; Rhodes, Ebert, and Fischer, 1992), but all are adults that adolescents perceive to be influential. Regardless of the moniker, informal mentoring generally provides benefits that resemble those of formal mentoring programs, decreasing problem behaviors and improving psychological well-being, academic performance, and relationships with others (Bryant and Zimmerman, 2003; Greenberger, Chen, and Beam, 1998; Klaw, Rhodes, and Fitzgerald, 2003; Rhodes, Contreras, and Mangelsdorf, 1994; Rhodes, Ebert, and Fischer, 1992; Zimmerman, Bingenheimer, and Notaro, 2002).

Some have claimed that adolescents rarely have meaningful relationships with adults outside of their immediate family (Steinberg, 1999), but a recent study reveals that four out of five adolescents report having an informal mentor in their lives (Beam, Chen, and Greenberger, 2002). Therefore, informal mentoring can be seen as a normative process that occurs in the lives of most young people rather than a process that is important only for at-risk youth. Unfortunately, research on the effectiveness of informal mentoring has been confined almost exclusively to at-risk populations—such as pregnant teenagers and lower-income African Americans, Hispanics, and urban youth (Klaw and Rhodes, 1995; Sanchez and Reyes, 1999; Taylor, 1990; Zimmerman, Bingenheimer, and Notaro, 2002). As a result, little is known about the overall impact of adolescent mentoring in the general population.

Much remains unknown about the instrumental benefits of informal mentoring relationships (Darling, Hamilton, and Niego, 1994; Darling, Hamilton, and Shaver, 2003), since most studies focus exclusively on their emotional impact. Mentors do much more than merely serve as cheerleaders for their mentees. They may also enhance the development of essential skills, provide a variety of opportunities, and serve as their sponsors (Hamilton and Hamilton, 2004).

Employment is a major area of neglect in studies of mentors. The transition to employment is tenuous for many young people, especially given the lack of institutionalized structures in the U.S. linking education and employment (Kerckhoff, 2003; Rosenbaum, 2001). Young people must take an active role in shaping their passage to employment. The resources needed to make a successful transition are developed in socialization contexts that promote planning and provide access to social capital (Heinz, 2003). Such contexts are not common among disadvantaged youth who need the most help in obtaining and maintaining work.

With these issues in mind, informal mentoring has a potentially important role to play in the transition from school to work and in the maintenance of stable patterns of work life. Prior research has documented the importance of personal relationships for educational (e.g., Kelly, 2004; Sandefur, Meier, and Campbell, 2005; Sun, 1999) and employment outcomes (e.g., Caspi, Wright, Moffit, and Silva, 1998; Elliott, 2001). However, few studies have examined the relationship between adolescent mentoring and employment in early adulthood.

The only exception comes from the research of DuBois & Silverthorn (2005a; 2005b), who used data from Add Health to explore the effects of informal mentoring on employment. Their results show, overall, a significant positive effect of mentoring on the odds of employment (DuBois and Silverthorn, 2005b), although among those who report having a mentor, specific characteristics of the mentoring relationships were not associated with employment (DuBois and Silverthorn, 2005a). In their study, employment was examined alongside 17 other dependent variables related to educational attainment, problem behavior, psychological well-being, and physical health. As such, they could only offer a cursory analysis of the relationship between mentoring and employment.

We provide a more thorough examination of this relationship, improving on the research of DuBois and Silverthorn in a number of ways. First, we analyze the full Add Health sample rather than the smaller public-use sub-sample, which provides us with greater statistical power and allows the sample to be subdivided. Second, we examine employment only among respondents who are age 23-28, since non-employment during the first years of adulthood is largely unrelated to success in later life (see the next section for more details on this point). Third, we provide a more comprehensive set of control variables in order to assess the indirect effects of mentoring on employment. Finally, we explore the role of both selection and causation processes involved in the relationship between mentoring and employment. The following research questions guide our analysis.

First, is mentoring positively related to the successful transition to young adult employment? Most prior research on adolescent mentoring has focused either on emotional outcomes or on more proximate instrumental outcomes, such as education. We maintain that adolescent mentoring is likely to be associated with a variety of outcomes including employment.

Second, what about the timing of the mentoring? According to life course theory, the impact of any particular event is dependent on the life stage at which it is experienced (Elder, 1998). In this way, the timing of mentoring relations should have important implications for employment. Adolescence is a crucial developmental period in which young people develop the skills, planfulness, and self confidence necessary to succeed in their later work careers (Mortimer, 2003). Mentors can help young people develop these skills and relationships at this timely phase in life. However, youth with later mentoring may find that the social world has passed them by, leaving them unprepared for the world of work. Therefore, we expect the relationship between adult employment and mentoring during adolescence to be stronger than that for mentoring in early adulthood.

Third, in what ways are selection and causation effects expressed in the relationship between mentoring and employment? That is, does mentoring boost a person’s chances of obtaining and maintaining steady employment or does the work environment provide young people with access to influential mentors? From a life course perspective, both selection and causation processes are important for understanding the pathways young people follow in obtaining steady employment (George, 2003). On the one hand, mentors have a causal impact on development, particularly through attachment to the labor market (Levinson, Darrow, Klein, Levinson, and McKee, 1978). On the other hand, employment exposes young people to new relationships with employers and coworkers (Mortimer, 2003). In this way, self-selection into employment may lead to the development of mentoring relationships. These working relationships can provide encouragement and facilitate the learning of skills and competence necessary for success in work settings (Hamilton and Hamilton, 2004). Mentoring through work may therefore help people maintain steady employment, minimizing the risk of both voluntary and involuntary separation from employment. In the end, both causation and selection processes are relevant for understanding the role of informal mentors in employment and our analysis is sensitive to both processes.

Fourth, is the effect of mentoring on employment direct or does it operate through other related processes? Mentors may help young people become attached to the labor market through direct intervention. For instance, they can discuss the importance of employment, identify potential places to work, help fill out applications, and provide transportation to work. Status attainment research, however, suggests that the influence of non-parental adults on employment may also follow an indirect path (Sewell, Haller, and Portes, 1969). Mentors provide behavioral modeling and guidance that directly impacts young people’s attitudes. Over time, this influence helps them develop a sense of self that matches the expectations their mentor has for them, or at least their perception of those expectations. This process is particularly important in the formation of educational aspirations and its relation to the labor market and steady employment. So while we anticipate that adolescent mentoring has a direct impact on the employment opportunities of young adults, we also expect to find an indirect impact of mentoring on employment through improvements in self-esteem, educational attainment, and the acquisition of work-related skills.

Fifth, what are the mechanisms by which mentoring can enhance an individual’s chances for stable employment in young adulthood? Relationships with adult mentors extend the social networks of young people and thereby help them bridge the gap between the adolescent world and the adult world. Therefore, it is important to consider the closeness of the mentoring relationship—i.e., the strength of the tie. Weak ties are well suited for this bridging function, as they provide greater access to non-redundant information about employment (Granovetter, 1973). In other words, connections with adults that operate outside the young person’s close-knit social circle are more likely to provide new information about opportunities. Since weak ties are associated with the receipt of this non-redundant information (Burt, 1992), we expect that young people who maintain relatively weak relationships with their mentors will have the greatest access to labor market information and opportunities. This would, in turn, enhance their chances of being employed in young adulthood. Similarly, connections with non-kin mentors are also likely to provide superior access to labor market information by expanding opportunities beyond the family circle.

Mentors also vary in the types of support that they provide, termed mentoring functions. Some mentors offer guidance to young people that may help them set personal goals and gain direction in their lives. Prior research has noted the importance of connections to adult mentors in the community who can ease the transition to employment by offering guidance, advice, and encouragement (Jarrett, Sullivan, and Watkins, 2005; Zippay, 1995). Others give emotional support which enables them to manage stress in the workplace. Mentors also function as role models who serve as positive worker examples for young people. All of these functions are likely to have a positive impact on employment prospects.

Finally, does the effectiveness of mentoring vary by gender? Bogat & Liang (2005) note that little is known about the relative effectiveness of mentoring for young women and men. Prior research suggests that mentoring relationships for women are more likely to serve psychosocial functions while mentoring of males is more likely to involve instrumental functions (Bogat and Liang, 2005). Meta-analysis of formal mentoring programs suggests that gender is unrelated to mentoring effectiveness (DuBois, Holloway, Valentine, and Cooper, 2002). Research on natural mentoring and psychological distress has found no gender differences in mentoring effectiveness (Zimmerman, Bingenheimer, and Notaro, 2002). However, other studies have documented important gender differences among youth in labor market mobility and job search (e.g., Keith and McWilliams, 1999). Furthermore, on average, work occupies a more central role in the lives of young men than young women. Women are more likely than men to pursue higher education and are less likely to be employed. As a result, we expect the impact of mentoring on employment to be greater for men than for women.

Data & Methods

In the Add Health study, students from grades 7-12 were randomly selected from 134 schools (high schools and middle schools, primarily) in 80 communities throughout the United States to participate in the survey. Wave I of the survey was administered in 1995, Wave II followed one year later, and Wave III was conducted in between August 2001 and April 2002 when respondents were age 18 to 28 years old. Wave III contains a module of questions on informal mentoring relationships with non-parental adults. Add Health’s large sample size and representativeness make it a unique dataset with which to study adolescent mentoring.

We focus our analyses on employment immediately after the initial period of labor market entry. Initial entry into the labor market following schooling has been characterized as a period of “churning” and “milling about” (Gardecki and Neumark, 1998). During this period, young people are often shopping for careers and experimenting before settling into a long term job (Lynch, 1993), plus employers are often reluctant to hire youth into career-ladder jobs without much prior work experience (Lewis, III, Shipley, and Madzar, 1998). Research has shown that it takes most young men until they are about 23 years old before they are employed in a job that they will hold for three or more years, regardless of their educational achievement (Klerman and Karoly, 1994). Employment histories during this initial entry period have minimal impact on the long-term labor market prospects of young adults (Ellwood, 1982; Gardecki and Neumark, 1998; Lynch, 1993; Osterman, 1980). Therefore, we restrict our analysis to young people age 23-28 in the sample.

Wave III of the Add Health survey contains 15,197 respondents. Of these, 14,322 have valid sample weights. Selecting the respondents who were age 23 or older at the time of the interview reduces the sample to 6,015. We employed listwise deletion to handle missing data on all variables of interest (with a few exceptions noted below), resulting in a final sample of 5,740 respondents. Table 1 displays the weighted sample means for the variables used in the analysis. Full time employment status serves as the dependent variable for all of our analyses. Sixty-four percent of the respondents at Wave III said that they were working for pay for at least 35 hours a week. Gender is measured by a dummy variable for male. Race is measured by dummy variables for Black, Hispanic, Asian, and Native American with White as the reference category. As mentioned earlier, the ages range from about 23 to 28, but 98 percent of the respondents are between 23 and 25 years old. To measure family socioeconomic status, we include a measure of the highest educational level of the respondents’ parents. The variable ranges from 1 to 4 (no high school degree, high school degree, some college, college degree). For the cases lacking such information, we set the parent’s education variable to the median (3). For a proxy measure of intelligence, we used the Add Health Picture Vocabulary Test (PVT), which is an abbreviated version of the Peabody Picture Vocabulary Test administered in Wave I. Missing values are set to the mean. Additional analyses (where embedded dummy variables were included in the regression models and where missing cases were excluded) reveal that substituting the mean and median values for these variables did not alter the interpretation of the findings.

Table 1.

Descriptive Statistics

N=5740	Mean	Std. Err.	Range
Full time employment (35+ hours per week)	.64	.01	0-1
*Male	.53	.01	0-1
*Black	.17	.03	0-1
*Hispanic	.12	.02	0-1
*Asian	.04	.01	0-1
*Native American	.02	.00	0-1
Age	23.79	.02	23-28
*Parent’s education	2.78	.05	1-4
*Picture Vocabulary Test (PVT)	99.89	.76	10-128
In school	.22	.01	0-1
Married	.26	.01	0-1
Child under age 6	.22	.01	0-1
*Employed at Wave I	.71	.02	0-1
Self-esteem	12.91	.05	0-16
No high school degree	.08	.01	0-1
Some college	.27	.01	0-1
Two-year college degree	.09	.01	0-1
Four-year college degree	.21	.02	0-1
Prior work experience	3.87	.07	0-6
Mentor	.74	.01	0-1
Age when mentor became important
Early (age 0-17)	.49	.01	0-1
Late (age 18-25)	.26	.01	0-1
Strength of relationship
Very close	.39	.01	0-1
Not so close	.30	.01	0-1
Missing closeness	.06	.00	0-1
Social role
Relative	.27	.01	0-1
Friend	.14	.01	0-1
Teacher	.17	.01	0-1
Community	.17	.01	0-1
Functional role (not mutually exclusive)
Guidance	.45	.01	0-1
Emotional support	.31	.01	0-1
Role model	.09	.01	0-1

^*Variables measured at Wave I. All others measured at Wave III.

Additional controls include dummy variables for whether or not the respondent was enrolled in school, married, or had a child under age 6 at the time of the survey. We also include a control for employment status at Wave I: respondents were asked whether they worked for pay outside the home during the last four weeks. Self-esteem is measured with an index constructed by summing the responses to four questions about how respondents feel about themselves (e.g., “Do you agree or disagree that you have many good qualities?”). The alpha reliability coefficient for these items is .78. Education level is measured categorically, with 4 dummy variables for no high school degree, some college, two-year college degree, and four-year college degree (reference category=completed high school or GED but did not attend college). We also include a measure of prior work experience prior to the Wave 3 interview, measured as the number of years (during 1995-2000) in which respondents worked throughout the entire year.

The mentoring module of Wave 3 begins with the following question:

• Some young people know adults, other than their parents, who make an important positive difference in their lives. Some do not. Other than your parents or step-parents, has an adult made an important positive difference in your life at any time since you were 14 years old?

Almost three-quarters of the respondents in the sample reported that they had such an important non-parental adult. This confirms previous assertions (Beam, Chen, and Greenberger, 2002) that informal mentoring relationships are part of the typical life course of adolescents. Respondents were prompted to think about the mentor who had been the “most influential” and answer a series of questions about their relationship. One of these questions asks respondents how old they were when their mentor first became important in their lives. About half of respondents reported having mentors who became important before they turned age 18. The vast majority of these mentors remained important in the lives of the respondents at the time of the Wave III interview.

The Add Health youth were asked several additional questions about their relationship with their mentor. First, they were asked to explain how close they feel to their mentor. Response categories were recoded as relationships that were very close (“very close” and “quite close”) and not so close (“somewhat close”, “only a little close”, and “not close at all”). About 6 percent of people who identified mentors (n=307) had missing values for this variable (primarily due to the mentor having passed away prior to Wave III). We coded these cases zero for the very close and not so close variables and included a separate embedded dummy variable in the analysis (1=missing, 0=not missing) to account for these missing cases (Hardy and Reynolds, 2004). Second, respondents were asked about the social role of their mentor (Darling, Hamilton, and Shaver, 2003). These social roles have been grouped into four categories: relatives (grandparents, aunts, uncles, siblings, and spouses), friends, teachers (teachers and coaches), and community members (friends’ parents, neighbors, clergy, employers, coworkers, and others). Finally, the respondents were asked an open-ended question about what their mentors did to help them. We recoded the responses to these questions into functional role categories, which include providing guidance and advice, emotional support, and serving as role models. Unlike the social role categories, the functional categories are not mutually exclusive.

Our analyses are all based on logistic regression models of the likelihood of being employed for at least 35 hours per week at the time of the third wave interview. All regressions are weighted and adjusted for sample clustering by schools. Because our expectations are directional (i.e., mentoring is positively related to employment), we rely on one-tailed tests of statistical significance. Our analysis begins with a baseline assessment of the association between mentoring and employment in young adulthood. Second, we examine the role of the timing of mentoring for work-related and non-work mentors, which allows us to explore both selection and causation effects. Third, we examine the potential indirect influence of mentoring on employment through self-esteem, educational attainment, and prior work experience. Finally, we examine the mechanisms by which non-work mentors might directly influence the odds of full time employment.

Findings

Table 2 provides the baseline estimates of the relationship between mentoring and the likelihood of being employed full time. Model 1 suggests that, net of gender, race, age, family background, PVT score, school enrollment, and family status, young people who report having mentors are 25% more likely than those without mentors to be employed full time at Wave III. This suggests that, in general, mentoring is positively related to the transition to full time employment in young adulthood.

Table 2.

Baseline mentoring effects: odds ratios from logistic regressions predicting full time employment

	Model 1	Model 2	Model 3	Model 4
Male	1.562 ***	1.560 ***	1.491 ***	1.630 ***
Black	.735 **	.731 **	.814 *	.956
Hispanic	1.235	1.234	1.336	1.305
Asian	.741	.752	.824	.796
Native American	.919	.922	.898	1.046
Age	1.027	1.029	1.021	.929
Parent’s education	1.052	1.055	1.039	1.026
PVT score	1.009 *	1.009 *	1.009 *	1.002
In school	.241 ***	.241 ***	.241 ***	.206 ***
Married	1.170	1.169	1.143	1.004
Child under age 6	.749 **	.743 **	.735 **	.869
Employed at Wave 1			1.567 ***	1.156
Self-esteem				1.052 *
No high school degree				.755
Some college				1.139
Two-year college degree				1.698 **
Four-year college degree				2.284 ***
Prior work experience				1.423 ***
Mentor	1.252 **
Early mentor (0-17)		1.318 **
Late mentor (18+)		1.128
Early work mentor			1.495	1.107
Late work mentor			1.699 **	1.330
Early non-work mentor			1.274 **	1.213 *
Late non-work mentor			.830	.810 *
Pseudo R-square	.079	.080	.091	.172
N	5740	5740	5740	5740

^*p<.05,

^**p<.01,

^***p<.001;

one-tailed test

In order to better understand the role of mentoring in this process, it is important to examine its timing. Model 2 reveals that young people who experience mentoring during adolescence or earlier are significantly more likely to hold a full time job in young adulthood than people who do not report having mentors. Developing a mentoring relationship after adolescence is positively related to employment, but the effect is not statistically significant.

One alternative interpretation is that these findings reflect duration effects and not timing effects. In other words, the positive effect of early mentoring could be driven by the fact that most early mentoring relationships last longer than most late mentoring relationships. Duration and timing are highly correlated (r = -.83), since the vast majority of respondents with mentors reported that their mentors remained important in their lives at Wave III. Because of multicollinearity between the two variables, we cannot assess the net influence of timing while controlling for duration in a single regression model. However, to further explore this alternative explanation, we estimated separate detailed models of duration and timing effects (available upon request). Two patterns support our timing interpretation. First, the results showed that relationship between duration and employment is curvilinear rather than positive, with its greatest impact in the middle of the distribution (duration = 6-15 years) and not among those with the longest mentoring relationships. Second, the impact of mentoring is greatest when experienced during the height of adolescence (ages 15-17), a period when youth are beginning to take an active role in shaping their own occupational trajectories. This suggests that the timing of the relationship is likely to be a more important factor than the duration of the relationship.

Early and late mentoring represent fundamentally distinct processes that can be revealed once we begin to parse out selection and causation forces. We do this in Model 3 by first including a control for whether the respondents were working at Wave I, which serves as a proxy measure of an individual’s orientation toward work. Not surprisingly, employment at the first wave is a strong predictor of full time employment in the third wave. Second, we identify the respondents who 1) met their mentors through work or 2) reported that their mentoring relationships became important either during the same year or after the start of their current job. Both instances imply that the mentoring relationship began after their job began, perhaps resulting from selection into employment. Respondents with these “work-related mentors” are distinguished from “non-work mentors” in Model 3.

The results suggest that the relationship between mentoring and employment involves both selection and causation processes. Both the early and late work mentor variables reveal positive effects on the odds of employment, but only respondents with late work-related mentors have odds of full time employment that are significantly greater than those individuals without mentors. These coefficients suggest that entry into the world of work can facilitate the development of mentoring relationships, especially among those who develop work-related mentoring relationships after adolescence. Cultivating mentoring relationships with coworkers and supervisors can also encourage young people to develop useful work habits and to maintain their current jobs. Such a process involves a feedback loop, where selection into employment increases the chances of developing a mentoring relationship and mentors in turn increase a young person’s labor force attachment.

The effects for non-work mentors are positive and significant only among respondents who developed mentors before adulthood. This suggests that the causal influence of non-work mentoring on employment is largely confined to mentoring that occurs prior to adulthood. Youth who develop non-work mentors during adulthood are actually less likely than those without mentors to maintain a full time job, though the difference is not significant. This finding should be interpreted with caution. To some extent, these late non-work mentors might be steering young adults away from employment, but they also might be helping to promote positive outcomes in other important spheres of life, like those related to higher education or relationship quality. Of course, these mentors could have other potentially negative influences, such as friend mentors who promote criminal activity or spouse/romantic partner mentors who encourage economic dependency, thereby reducing labor force attachment. However, the relationship remains essentially unchanged even after friend, spouse, and partner mentors are removed from the analysis, suggesting that these situations are not driving the effects.

As mentioned earlier, a number of factors may be involved in the relationship between mentoring and employment, including self-esteem, educational attainment, and prior work experience. These factors could affect the chances of mentoring relationship formation, which could in turn impact the odds of full time employment. Or they could mediate the relationship in instances where the development of mentoring relationships impacts self-esteem, educational attainment, and work experience. Controlling for these factors in Model 4 allows us to assess the extent to which mentoring directly and/or indirectly influences the odds of employment.

Self-esteem, education, and work experience are all positively and significantly related to employment. The inclusion of these variables substantially increases the explanatory power of the regression model. Their inclusion also substantially reduces the effect of work-related mentors on employment, to the point where the effect for late work-related mentors is no longer significant. Additional analyses reveal that this reduction in the effects is almost entirely due to the inclusion of the prior work experience variable. These findings lend support to the notion that work-related mentoring involves a feedback process, whereby work experience facilitates the development of work-related mentoring relationships, which in turn increase attachment to the labor force. At the same time, controlling for self-esteem, education, and work experience lead to a modest reduction in the effects for the non-work mentors. The impact of early non-work mentors remains positive and significant even after controlling for these potentially indirect influences. This suggests that the influence of early non-work mentors is largely direct.

The remainder of the analysis explores in greater detail the two remaining questions we outlined: 1) the direct causal influence of non-work mentoring on young adult employment, focusing on the characteristics (closeness, social roles, and functional roles) of these mentoring relationships and 2) the gender differences in the effectiveness of mentoring. For the work-related mentors, we include a table in the appendix that shows the zero-order associations between the work-related mentoring characteristics and full time employment. We decided against examining the characteristics of these relationships in a multivariate framework for both empirical and theoretical reasons. First, the number of work-related mentors is quite small (especially for the early work mentors) compared to the non-work mentors. Splitting these mentors into distinct categories for closeness, social roles, and functional roles further reduces the cell sizes to the point where a number of the regression coefficients become unstable. Second, as demonstrated above, selection and causation processes are intimately involved in relationship between work-related mentoring and employment opportunities. Endogeneity of the independent variable with the dependent variable makes interpretation of the findings dubious. Multivariate analyses of work-related mentoring characteristics can reveal which types of mentors young people tend to develop in work settings, but say little definitively about the causal influence of work-related mentors on employment.

Since the non-work mentor categories include only those respondents who met their mentors outside of work and before they started their current jobs, these variables are exogeneous to the dependent variable. Thus, we are able to isolate the direct causal impact of non-work mentoring on the odds of employment. An examination of the characteristics of non-work mentoring relationships (closeness, social roles, and functional roles) provides insights into the mechanisms by which mentors help young people gain and maintain full time employment. Furthermore, we examine males and females separately in order to detect gender differences in these processes.

The four sets of regression models summarized in Table 3 are estimated separately. All of the models control for the variables included in Model 4 of Table 2. The first three sets of mentoring dummy variables—the main mentoring effects, mentoring closeness, and social role characteristics—all reference respondents who did not have mentors. The functional role characteristics, however, are not mutually exclusive so they each reference those individuals who did not receive the specific functions (from their mentor or because they had no mentor). The results from Model Set A suggest that, on average, young men who developed non-work mentoring relationships in adolescence were about 40% more likely to be full time employed than those who did not develop mentoring relationships. Late non-work mentoring is essentially unrelated to the odds of employment for young men. For the females, early non-work mentoring is positively but insignificantly related to employment, while late non-work mentoring is associated with a 28% decrease in the odds of employment. Overall, the positive influence of mentoring on employment tends to be greater for men than for women, but this difference is not itself statistically significant.

Table 3.

Mentoring mechanisms: odds ratios from logistic regressions predicting full time employment

	Males (N=2849)		Females (N=2891)
	Early Non-work	Late Non-work	Early Non-work	Late Non-work
Model Set A
Mentoring1
Mentor	1.411 *	.986	1.116	.716 *
Model Set B
Closeness1
Very close	1.187	1.103	1.129	.714
Not so close	1.994 ***	.790	1.027	.706
Missing closeness	.850	3.080	1.313	.765
Model Set C
Social role1
Relative	1.247	1.516	1.033	.580
Friend	1.466	1.589	1.566 *	.795
Teacher	1.609 *	.669	1.220	.714
Community	1.593	.677	.953	.838
Model Set D
Functional role
Guidance	1.351 *	.764	1.238 †	.842
Emotional support	.995	1.141	1.016	.971
Role model	1.218	1.055	1.094	.518

^†p=.051,

^*p<.05,

^**p<.01,

^***p<.001;

one-tailed test

All models include controls listed in Table 3.

¹Reference category=no mentor

Model Set B examines the closeness of the mentoring relationship, revealing a strong positive effect among young men for early non-work mentoring relationships that respondents characterize as not very close. Maintaining these weak-ties to mentors nearly doubles a young male’s odds of employment. For young women, relationship closeness is unrelated to employment. Analysis of the social role characteristics in Model Set C reveals that males who identify teachers as mentors during adolescence are significantly more likely to be employed in young adulthood. Women who develop mentors with adult friends during adolescence are 57% more likely to be employed in young adulthood. The remaining social roles are not significantly related to the odds of employment.

The findings for the young male sample support the strength of weak ties hypothesis. However, it remains possible that the results are driven entirely by the importance of teacher mentors, since only 15% of young people with mentors who are teachers report being very close with them. To explore this possibility, we interacted the closeness and social role variables for the early non-work mentors of young men (see Figure 1). The results show that weak ties to mentors are generally more effective than strong ties; the difference between the coefficients is significant for the relative and friend mentors and insignificant for teachers. Strong ties to community mentors appear to be somewhat more effective than weak ties to community mentors, though the difference in the coefficients is statistically insignificant. In other words, weak ties to relative and friend mentors can help young men to expand knowledge of labor market opportunities beyond redundant information passed through close-knit social circles. The positive impact of having a teacher mentor on the odds of employment is largely independent of the strength of the relationship.

Figure 1.

Odds of full time employment for social role*closeness interaction for males with early mentors

Model Set D highlights the importance of guidance from mentors during adolescence. Young men and young women who receive guidance from mentors are more likely to be employed than those who did not report receiving guidance. The effects are significant for males and marginally significant for females (p=.051). Emotional support and role modeling are not significantly related to the odds of being employed full time for either males or females.

Discussion

The transition to employment in the United States can be treacherous for many young people. This is particularly true in the current economic climate, where they have borne the brunt of the recent “jobless recovery” (Sum, Khatiwada, McLaughlin, and Palma, 2005). Now more than ever, young people need guidance to help them make a successful transition from school to work. This study demonstrates that informal mentoring plays an important role in this transition. Prior research has tended to focus on the relationship between mentoring and emotional outcomes. But mentoring also has the potential to promote tangible benefits for young people in the labor market.

The relationship between mentoring and employment involves both causation and selection processes. Mentoring can 1) influence a young person’s odds of gaining full time employment (causation) and 2) employment can provide young people with access to work-related mentors (selection), which in turn can help promote labor force attachment (causation). The character of the relationship between mentoring and employment is largely linked to the timing of mentoring. Mentoring that occurs prior to adulthood tends to have a causal influence on employment opportunities later in life. Youth who develop non-work mentors during adolescence are significantly more likely than those without mentors to be employed during early adulthood. Mentoring relationships that develop after adolescence involve selection processes that can feedback into causation processes. Gaining employment facilitates the development of significant relationships that expand social capital resources (Mortimer, 2003). The development of these work-related mentoring relationships may help to protect young people from dropping out of the labor market, increasing attachment to full time employment.

Mentoring relationships that develop outside of work settings and during adolescence have a significant causal influence on the odds of employment during young adulthood, even after controlling for self-esteem, educational attainment, and prior work experience. Young men who maintain weak relationship ties to their mentors and those who develop mentoring relationships with teachers have the greatest chance of being employed during young adulthood. Young women who develop friend mentors are the most likely to employed full time. Regardless of gender, young people who receive guidance and advice from their mentors are significantly more likely to be full time employed than youth who do not receive guidance and advice.

Tapping into the social capital resources of adults can help young people succeed in the transition to adulthood. The findings confirm the notion that weak ties are especially important for employment (Granovetter, 1973). Young men with weak ties to their mentors have the best chances at employment. Weak ties are valuable because they can increase the likelihood of receiving non-redundant information that expands knowledge of labor market opportunities (Burt, 1992). In this way, weak ties to mentors play a particularly important role in the transition from school to work. Prior research has documented the way that teachers often rely on their own personal contacts to provide students with employment opportunities in the community (Rosenbaum, 2001; Royster, 2003).

The results presented here suggest that non-work mentoring may be less effective among young women than it is among young men. Future research should assess why this might be the case. One possibility is that the mentors for young females offer fundamentally different types of support. For example, the women in the Add Health study were more likely than the men to receive emotional support, a factor that was unrelated to the odds of employment. At the same time, the women were less likely than men to receive guidance from their mentors, a factor that significantly predicts employment. Young women’s networks may also be qualitatively different from the networks of young men. In other words, the people that become important in men’s lives may be better able to access valuable resources, such as employment opportunities (Campbell, 1988; Moore, 1990). Males tend to have less instability across their work careers than women (Williams and Han, 2003) and therefore have access to more and better information about job opportunities than their female counterparts. The gender homophilous character of mentor-mentee matches suggests that young women would have access to relatively fewer employment opportunities.

Many other possible explanations exist. Young women report particularly high levels of discrimination in the labor market (Antecol and Kuhn, 2000). Such labor market discrimination could dampen the positive effects of mentoring for women. In young adulthood, women may be less likely than men to rely on their social resources when attempting to gain employment (McDonald, 2005), using their educational credentials more than their social contacts to help them get their jobs. Finally, men and women, in the aggregate, maintain different preferences and expectations for paid employment. Virtually all young men are expected to work in society, while some young women view employment as an option rather than as a necessity. In this way, employment can take on varying levels of importance in people’s lives. This difference in importance may reflect itself in the choice of “important adults”, such that some women may be more likely to identify adults that help them navigate other (non-work) arenas of life.

Future research should also explore the extent to which mentoring promotes equity versus inequality in society. Mentors may enable disadvantaged youth to “catch up” with their more fortunate peers, supplementing their lack of skills and resources with social capital. Conversely, the advantages of mentoring may be additive rather than supplemental, further enhancing the opportunities of the most advantaged youth. Preliminary analyses from Add Health suggest that mentoring relationships are least likely to form among young people who need mentors the most—disadvantaged youth. Others have identified racial differences in teacher assistance that place blacks and whites on divergent occupational trajectories (Royster, 2003). On the other hand, findings from the High School and Beyond survey suggest that job placement through teachers and schools occurs most often among women and minorities, while students who received help from relatives are more likely to come from high SES backgrounds (Rosenbaum, DeLuca, Miller, and Roy, 1999). This suggests that different mentoring processes can simultaneously promote and reduce inequality.

Furthermore, future research should explore the impact of mentoring on occupational trajectories. It seems likely that the employment benefits of adolescent mentoring in young adulthood could place youth on a path of advantage in their work careers. Little is known about the long-term consequences of informal mentoring since most studies rely on cross-sectional data (Zimmerman, Bingenheimer, and Behrendt, 2005). Given its longitudinal research design, the Add Health study will be a useful tool for assessing these kinds of long-term trajectories of mentoring effectiveness. Such assessments should also be expanded by examining how mentoring is related to job quality. Future studies might consider the relationship between mentoring and other job quality indicators, such as job satisfaction and wages.

A number of limitations of this study are worth noting. First, selective recall is a potential problem, as the Add Health respondents were asked to identify an important non-parental adult in their lives. One would expect that successful youth would be more likely to recall a mentor than someone who has not fared as well. Such a bias would result in the overestimation of the effectiveness of mentoring in the analysis. Second, the respondents were asked to describe only their “most influential” mentoring relationship. We are unable to account for the possibility of multiple mentoring relationships, either simultaneously or at different stages in life. Therefore, it remains possible that young people who provided information on a mentor relationship that developed after age 17 may or may not have had a mentor during adolescence as well. Finally, we are limited in our ability to isolate selection effects from causal effects, since selection and causation processes often operate simultaneously.

A number of policy implications are suggested by this study. While mentors can have a positive influence in the lives of young people, those who do not have mentors may be left to their own devices. Young people who lack exposure to positive role models and guidance are likely to have relatively few career opportunities (Forum on Adolescence, 1999). In recent years there has been a rapid expansion in the number of formal school-to-work programs for high school students (Linnehan, 2003). In these programs, students are assigned adult mentors from community organizations that provide them with employment opportunities (Linnehan, 2001). Programs such as these are likely to facilitate entry into work by fostering connections between youth and adults in the community.

Future interventions would be well-served by focusing on the areas that need it most, such as disadvantaged neighborhoods where access to both mentoring relationships and quality employment is limited (Forum on Adolescence, 1999). These programs should also work to incorporate young women as well as young men. Informal mentoring does not consistently offer the kinds of contacts, guidance, and support that are necessary to improve the labor market opportunities of women. Formal mentoring might begin to fill this gap. Given the relatively high levels of poverty among women, it is most important that young women develop the kinds of mentoring relationships that enhance their labor market opportunities. Also, interventions that aim to increase employment among youth should be tailored to different age groups in order to mirror naturally-occurring mentoring processes. In other words, adolescent mentoring that occurs outside the world of work can have a positive influence on employment in adulthood. After adolescence, mentoring programs that take place within the context of work are likely to have the greatest impact on future employment patterns.

Finally, the mentor-mentee dyad is embedded within the context of a larger constellation of social relationships (Higgins and Thomas, 2001). The informal influence of mentoring has remained a hidden mechanism of advantage that has yet to be examined in depth alongside that of parents and peers. Research on adolescent mentoring and the transition to adulthood opens new opportunities for understanding how social relationships influence the life trajectories of young people, launching some on a pathway of advantage, while leaving others to fend for themselves.

Appendix

Table A1.

Bivariate Correlations between Full Time Employment and Work-Related Mentoring Characteristics

Early work mentoring	Full time employment
	Male N=61	Female N=40
Mentoring
Mentor	.072 *	-.040
Closeness
Very close	.052	-.069 *
Not so close	.066 *	.023
Social Role
Relative	.051	-.081 *
Friend	.043	.003
Teacher	-.005	.014
Community	.050	-.018
Functional Role
Guidance	.063 *	-.021
Emotional support	.057 *	-.036
Role model	-.005	-.026
Late work mentoring	Full time employment
	Male N=369	Female N=276
Mentoring
Mentor	.107 **	.106 ***
Closeness
Very close	.092 **	.108 ***
Not so close	.044	.016
Social Role
Relative	.105 ***	.042
Friend	.014	.048
Teacher	-.041	-.016
Community	.093 ***	.081 **
Functional Role
Guidance	.078 **	.077 **
Emotional support	.036	.058 *
Role model	.015	.032

^*p<.05,

^**p<.01,

^***p<.001