بررسی کمی پژوهش مشاوره: آزمون یک مدل

Over the past 25 years, numerous researchers have studied the effects of mentoring on work outcomes. However, several reviewers have noted that many of the observed relationships between mentoring and its outcomes are potentially spurious. To summarize this widely dispersed literature, a quantitative research synthesis was conducted focused on estimating multivariate analytical paths between mentoring and several career outcomes, while holding constant correlates of mentoring including demographics, human capital, and core self-evaluations. The results demonstrate that mentoring does have substantial effects on job and career satisfaction after holding these covariates constant; yet factors such as core self-evaluations, tenure, and education have stronger effects on objective career outcomes. Potential future directions to enrich the study of mentoring and career success are described.

Although the concept of mentoring dates back to the earliest stages of human civilization, the pioneering qualitative work of Levinson, Darrow, Klein, Levinson, and McKee (1978) and Kram (1983) suggested that mentoring is a powerful influence on success in organizational environments. Despite promising theoretical propositions, several recent qualitative reviews of the literature note that there is considerable ambiguity regarding the outcomes of mentoring, with some studies reporting strong relationships between mentoring and career outcomes, while others find far less support (Noe, 1988a, Ragins, 1999a and Russell and Adams, 1997; Wanberg, Welsh, & Hezlett, 2003). In response to these concerns, a meta-analysis demonstrated that after aggregating across a variety of studies, there are reliable, but small, effects of mentoring on several career outcomes (Allen, Eby, Poteet, Lentz, & Lima, 2004). However, as Shadish (1996) has argued, although meta-analysis is a valuable statistical technique, a limitation of univariate meta-analysis is that “the statistical models used in most meta-analyses have probably been very poor approximations to any reasonable theoretical models about the causal structures that give rise to meta-analytic data” (p. 50). To remedy this substantive limitation, the use of causal modeling techniques based on meta-analytic data has been advocated (Viswesvaran & Ones, 1995). Such models are especially useful in fields where there is a question of the contribution of several related variables to a common outcome like performance or success, as demonstrated in the literatures on staffing (Schmidt & Hunter, 1998), self-efficacy (Judge, Jackson, Shaw, Scott, & Rich, 2007) and training (e.g., Colquitt, LePine, & Noe, 2000). The current study provides a quantitative synthesis of the mentoring literature in hopes of resolving issues related to: (a) the definition and functions of mentoring, (b) assessment of the effects of mentoring in multivariate models, and (c) an examination of the influence of mentoring on markers of career success relative to other related constructs. The structural model used in the study appears in Fig. 1. 1.1. Definition and dimensions Consistent with previous meta-analytic work, we compare how various measures of mentoring might relate to career outcomes (Allen et al., 2004). Researchers often provide research participants with a definition of mentoring and then ask a single question about whether respondents have such a relationship. As an example, Allen, Poteet, Russell, and Dobbins (1997) told respondents that, “Mentors are persons usually considered as more experienced, who support, train, ‘teach the ropes to’ or sponsor others as they pursue their career goals. Although your boss, manager, and/or supervisor can be a mentor, usually a mentor does not have to involve a day-to-day formal supervisory relationship” (p. 9). Similar definitions for mentoring can be found in numerous other studies (e.g., Chao, 1997, Dreher and Cox, 1996 and Ragins and Cotton, 1991). This measurement strategy allows for an examination of the effect of having a mentor, yet treats all mentors as equally effective. To deal with the quality of mentoring, researchers also have employed continuous indices of mentoring quality. These indices answer a completely different research question involving whether different mentors are differentially effective—in other words, all participants have mentors, and the question for researchers involves which types of mentors are most effective. One of the most direct methods for achieving this end is the use of aggregated scales of mentoring quality (e.g., Dreher and Ash, 1990, Feldman et al., 1999, Gilbert and Ivancevich, 1999, Hollingsworth and Fassinger, 2002, Kahn, 2001 and Mullen, 1998). Mentoring relationships have also been described in terms of two broad categories of functions supposedly provided by mentors based on both qualitative and quantitative data (Kram, 1983, Noe, 1988a and Tepper et al., 1996). Career functions include actions such as providing the protégés with human capital enhancement opportunities and links to powerful individuals in the organization. Psychosocial functions include counseling the protégé about anxieties and uncertainty, providing friendship and acceptance, and role modeling. Previous meta-analytic work showed differential relationships between mentoring functions and outcomes, but all of the relationships were in the same direction and many were of similar magnitude (Allen et al., 2004). Unfortunately, the correlation between mentoring methods was not investigated, leaving the dimensionality of mentoring open to question, especially since confirmatory factor analysis has suggested the two factor model explains the data little better than a single factor model (Tepper et al., 1996). Our model attempted to answer this question. 1.2. Antecedents of mentoring At the same time that writers observed that mentors may serve important career functions for many protégés, it was recognized that some individuals are more likely to receive mentoring than others. One proposition is that women and minorities may encounter more barriers to obtaining a mentor than White men and may, therefore, be less likely to have a mentor or receive quality mentoring (Noe, 1988b, Ragins, 1999a and Ragins, 1999b). Self-report data suggest that women perceive that there are more barriers to gaining a mentor than do men (Ragins & Cotton, 1991), but the empirical evidence that women and minorities end up with fewer mentors or receive less mentoring is inconclusive (Ragins, 1999a and Wanberg et al., 2003). Another potential antecedent of mentoring is human capital, in the form of education and organizational tenure. There is considerable evidence that mentors select protégés based on their expected productivity. Allen, Poteet, and Burroughs (1997) found in their qualitative interviews with 27 mentors that mentors seek out competent, motivated individuals to serve as protégés. Similar findings have been reported in subsequent studies (Allen, 2004 and Allen et al., 2000). Mentors deliberately seek capable individuals to act as protégés under the expectation that these protégés will be the best able to reciprocate the mentor’s assistance by giving information and providing the mentor with power in the organization (e.g., Mullen and Noe, 1999 and Ragins, 1997; Ragins & Scandura, 1994). Individuals who are more educated and experienced therefore may be more likely to attract mentors. Similar to education and tenure, a person’s self-image has dual effects by both leading to job performance and career rewards, and also making the individual more attractive to potential mentors. Evidence from studies of core self-evaluations show that these variables related significantly to motivation, job performance, and job satisfaction (Erez and Judge, 2001, Judge and Bono, 2001 and Judge et al., 2003). Since motivation and performance have been described as characteristics sought by mentors, and satisfaction and income are the proposed outcomes of mentoring, it appears that core self-evaluations might be the ideal dispositional trait as a control in studies of mentoring. 1.3. Correlates of mentoring Because performance is associated with mentoring (Allen et al., 1997 and Allen et al., 2000), and because it relates to extrinsic career success in the form of pay and promotions (see Gerhart & Milkovich, 1992; for a review), it is important that estimates of the effect of mentoring take performance into account. Because it is not possible to determine the extent to which productivity is a cause or an effect of mentoring, we treat these variables as having a noncausal association. Another possible confound in studies of mentoring is the fact that mentoring status may be the result of variables that are well-known antecedents of positive career outcomes such as tenure and education (Wanberg et al., 2003). 1.4. Outcomes of mentoring Effectiveness in one’s career is traditionally assessed with a combination of subjective perceptions as well as attitudes towards ones’ job and career progress (Hall, 2002 and Judge et al., 1999). Extrinsic success has generally been defined in terms of a person’s current salary and either the number of promotions one has received over a time period or in terms of a person’s rank in an organization’s hierarchy. The effect of mentoring on protégé intrinsic success should come through two distinct pathways. First, mentor effects on extrinsic career success should have a subsequent effect on protégé intrinsic success (Judge, Cable, Boudreau, & Bretz, 1995). This is shown in Fig. 1 as the paths between mentoring and salary and promotions, with additional paths from these extrinsic success factors and intrinsic outcomes. Because mentors also provide their protégés with psychological support and opportunities for development, it is likely that they will also directly contribute to the general satisfaction of protégés above and beyond the extrinsic rewards they can secure for their protégés (Russell & Adams, 1997).

The current research provides an opportunity to consider what we have learned from the study of one of the most focal work relationships—the mentoring relationship—to date. Our analyses included numerous potential antecedents and covariates of mentoring, including gender, race, core self-evaluations, tenure, education, and job performance which were not included in Allen et al. (2004) meta-analysis. Because of the examination of these correlates, we also determined how the purported effects of mentoring are modified once these variables were taken into consideration. This strengthens our appreciation of mentoring, because even with personality and other career related variables held constant, mentoring remains an important predictor of many career outcomes. However, when assessed relative to the effect sizes for variables like tenure and education (in predicting salary) and core self-evaluations (in predicting performance, job satisfaction, and career satisfaction), it appears that the benefits of mentoring are modest. Our structural model allows us to assess the distinct contribution of career and psychosocial mentoring when both are taken into account. The finding that career mentoring is considerably more important in a multivariate model was not previously demonstrated. 4.1. Implications for mentoring Overall, the results suggest that the effects on mentoring on career outcomes range from moderate to weak. When the effect of the mentoring variables was studied in the context of a path analysis, the results did not change appreciably, although the effects of mentoring tended to be considerably smaller than the effects of the covariates. Thus, although mentoring may not be properly labeled a useless concept to careers, neither can it be argued to be as important as the main effects of other influences on career success such as ability and personality (Judge et al., 1999), human capital (Judge et al., 1995), networks (Marini & Fan, 1997), or even the demographic variables included here. Several alternative explanations for the relatively small effect sizes for mentoring can be offered. First, there is evidence to suggest that individuals who do not have mentors will seek out more information from their co-workers than individuals who do have mentors (Ostroff & Kozlowski, 1993). Thus, the advantage conveyed by mentors may be offset by the use of alternative information sources by those who do not have mentors as resources. Second, it may be that the effect of mentoring on career success is moderated by other such as mentor gender, and gender similarity (e.g., Scandura & Williams, 2001). Other moderating influences, such as the ability and motivation of the protégé to implement whatever benefits the mentor bestows, may be operative. This last possibility seems especially worth examining, because mentors prefer protégés they perceive to be higher in willingness and ability to learn (Allen, 2004). In addition to the modest nature of the average relationships, surprisingly, there was not much variability in effect sizes by the definition of mentoring. Regarding how mentoring was defined, the results demonstrated few substantial differences in observed effect sizes. In the multivariate models, career mentoring related significantly and positively to every single outcome, but psychosocial mentoring is either not significantly related, or negatively related to the outcomes. One possible interpretation is that multicollinearity between the mentor functions influenced the results (Cohen, Cohen, West, & Aiken, 2003). The lack of distinction between mentoring forms corresponds to the high correlations among the mentoring measures (View the MathML source.59⩽ρˆ⩽.73—see Table 1). In considering the implications of these strong correlations, it is worth noting that the correlations among the mentoring measures is quite similar to the average incorrelation among the dimensions of organizational citizenship behavior (View the MathML sourcer¯c=.67) (LePine, Erez, & Johnson, 2002). Given the intercorrelations observed here with respect to the different conceptualizations of mentoring, like OCBs, the different conceptions of mentoring may be less distinct than has been assumed. Future research should investigate more explicitly whether it makes sense to distinguish among these various conceptions and measures of mentoring. Another noteworthy finding was the weak relationship between mentoring and demographic variables. Our results are inconsistent with the assertion that a primary reason for gender-based earnings disparities is the differential availability of mentoring (e.g., Glass Ceiling Commission, 1995). The correlational data show that men are not more likely to report having mentors, and that both men and women find that their mentors provide similar mentoring. This coincides with prior research demonstrating few differences between outcomes for male and female protégés (Ragins, 1999a). Additionally, the multivariate results show that race and gender are substantial correlates of career outcomes even after mentoring is held statistically constant. This however, does not discount the possibility that there are different variables in men’s and women’s mentoring relationships that may explain differences in career outcomes. Some might wonder if placing performance in our structural model eliminates the primary mechanism through which mentoring has its effects. Our goal in including performance in the structural model was not to pit performance versus mentoring as leading to specific outcomes, but rather, to show the influence of mentoring after performance as a potential mediator was taken into account. Additionally, because there is good reason to suspect that mentoring is more likely to be provided to individuals who are higher performers, it is not possible to attribute a relationship between mentoring and career outcomes unambiguously to mentoring without taking performance into account. There are reasons to suspect that mentoring might have performance-independent influences that should be examined in future research, as shown in our data. Ragins (1999b) notes that mentoring relationships provide resource and power for the protégé, and this is reflected in career enhancing measures that discuss factors like “protection” and “running interference” that would seem to enhance mentor work outcomes without necessarily improving their task performance. 4.2. Implications for future research The strongest correlations found in the current study were between attitudinal descriptions of mentoring and attitudinal measures of satisfaction with one’s job or career. The lack of correspondence between objective and subjective measures highlights a need for studies that obtain measures of mentoring from other sources. Wanberg et al. (2003) noted that individuals with high negative affectivity may be more likely to report that they have not been helped with their careers and that they dislike their jobs, which would artificially generate a positive correlation between mentoring and job satisfaction. An analysis that combines more objective indicators of career success (promotions and salary) with more affective, subjective indicators (job and career satisfaction) might help to explain this result. Research starting from a classical training approach, with pre- and post-mentoring measures of work outcomes would be very instructive in this regard (e.g., Goldstein, 1993 and Hellervik et al., 1992). Because mentoring is a dynamic process that unfolds in time (Wanberg et al., 2003), growth curve modeling studies would be especially instructive in this regard. Given the fairly weak relationships between mentoring and most career outcomes, the time may have passed for research that investigates mentor functions as a primary determinant of the success of mentoring. Several observers suggest that it may be more profitable to consider the mentor’s position within the organization as an explanatory variable or by examining the social network that makes up the organization (Podolny and Baron, 1997, Siebert et al., 2001 and Wanberg et al., 2003). In short, it may not be especially helpful to one’s career success if a relatively powerless or naı¨ve mentor comes to one’s assistance, no matter how helpful he or she may try to be. It is also possible that research will benefit from considering developmental relationships with multiple mentors simultaneously rather than concentrating on the behavior of a single influential individual (Higgins & Kram, 2001). In either event, it appears that if researchers wish to explain career success, they may increasingly have to turn away from mentor functions scales and towards a more detailed understanding of the mentor process as organizationally embedded.