Having Friends In High Places: The Effects Of Structural Characteristics Of Mentoring Dyads On Protégé Career Outcomes

Size: px
Start display at page:

Download "Having Friends In High Places: The Effects Of Structural Characteristics Of Mentoring Dyads On Protégé Career Outcomes"

Transcription

1 University of Miami From the SelectedWorks of Terri A. Scandura January 1, 1992 Having Friends In High Places: The Effects Of Structural Characteristics Of Mentoring Dyads On Protégé Career Outcomes Terri A Scandura, PhD, University of Miami Chester A Schriesheim, University of Miami Available at:

2 Having Friends in High Places 1 Having Friends In High Places: The Effects Of Structural Characteristics Of Mentoring Dyads On Protégé Career Outcomes Terri A. Scandura School of Business Administration University of Miami Chester A. Schriesheim School of Business Administration University of Miami (Running Head: Having Friends in High Places) Address correspondence to: Terri A. Scandura Department of Management 414 Jenkins Building School of Business Administration University of Miami Coral Gables, Florida (305) Scandura, T. A. & Schriesheim, C. A. (1992). Structural effects of mentoring relationships and protégé career outcomes. Presented at the Academy of Management meetings, Las Vegas, NV.

3 Having Friends in High Places 2 Abstract Three structural characteristics of mentoring dyads (the mentor s hierarchical level, duration of the mentoring relationship, and how long the protégé has been without a mentor) were proposed as having main and possibly interactive effects with three mentoring functions (social support, career coaching, and role modeling) on five protégé outcome variables (career expectations, commitment to the firm, number of hours worked during peak season, number of direct reports supervised, and current annual salary). Data from a sample of Certified Public Accountants (CPAs; N=786) were employed to first assess the distinctiveness of the proposed three mentoring functions by maximum likelihood confirmatory factor analyses. Then, hierarchical moderated multiple regression analyses were conducted for the structural characteristics and mentoring functions. Controlling for covariates, results indicated that the three mentoring functions are separate constructs and that each structural characteristic significantly correlated with protégé career outcomes in addition to infrequently acting as moderators for the mentoring functions. Implications for future research on mentorship in organizations are discussed.

4 Having Friends in High Places 3 Young and relatively inexperienced employees often report that they have learned a great deal from a mentor--an older and more experienced employee who advises, counsels, and otherwise enhances their career development (Kram, 1985). This older-younger, experiencedinexperienced dyadic relationship promotes processes resulting in beneficial outcomes for both protégé and mentor. Mentoring has been associated with enhanced job success, increased job satisfaction, increased professional promotions, improved organizational commitment and lowered intention to leave (e.g., Baugh, Lankau, & Scandura, 1996; Dreher & Ash, 1990; Fagenson, 1989; Kram, 1985; Scandura, 1992). Organizations advocate mentoring (both formal and informal programs) to capitalize on these gains improving organizational communication, more effective management development, and productivity (Burke & McKeen, 1989; Hunt & Michael, 1983; Murray, 1991; Ostroff & Kozlowski, 1993; Ragins, Cotton, & Miller, 2000). Mentoring is thus generally seen as enhancing the career outcomes of protégés, and additional research on mentoring would prove useful in advancing knowledge about the potential of this important organizational practice. Mentoring At Work: A Brief Literature Review The processes which occur in mentoring dyads have been discussed in a number of sources, but perhaps best-known is the in-depth analysis of mentor-protégé pairs reported by Kram (1985). Her research identified two basic dimensions of mentoring at work: coaching ( vocational ) and social support ( psycho-social ) functions. The coaching functions of mentoring include sponsorship, exposure and visibility, protection, and the provision of challenging assignments to the protégé (Kram, 1985). Social support functions on the other hand involve acceptance and confirmation, counseling, and personal friendship (Kram, 1985). The

5 Having Friends in High Places 4 mentor may also function as a role model demonstrating and reinforcing appropriate behavior. The existence of these mentoring functions has been replicated in a variety of settings (Burke, 1984; Douglas & Schoorman, 1988; Dreher & Ash, 1990; Fagenson, 1989; Fagenson-Eland, Marks, & Amendola, 1997; Noe, 1988; Olian, Carroll, Giannatonio & Feren, 1988; Schockett & Haring-Hidore, 1985; Scandura, 1992, 1997; Williams, 1999). Dreher and Ash (1990) used a global measure of mentoring functions to examine mentoring among technical, managerial, and professional positions noting the need for future research in the dimensionality of mentoring. This study explores structural characteristics of mentoring dyads which may directly effect protégé career outcomes, in addition to possibly moderating relationships between social support, coaching, and role modeling and these outcomes. The specific potential moderators examined include the mentor s level in the organizational hierarchy, the duration of the mentoring relationship, and any lapse in mentoring (i.e., how long the protégé has been without a mentor). As reviewed below, these characteristics are suggested in the literature as potentially important, yet there have been few attempts to empirically examine their effects on career outcomes for protégés. Most of the mentoring literature have focused attention on the nature of the mentoring relationship concentrating on relationship quality in formal and informal mentoring and attitudinal responses (e.g., Chao, Walz, & Gardner, 1992; Noe, 1988; Ragins, et al., 2000; Scandura & Williams, 1998) neglecting possible effects of mentor hierarchical level, mentoring duration, and mentoring lapses. LISREL 8 and maximum likelihood confirmatory factor analysis (Joreskog & Sorbom, 1993) is used to establish the viability of the mentoring functions (social support, coaching, and role modeling) confirming their use in this study and appropriateness for future studies. Structural Characteristics of Mentoring Dyads: Review and Hypotheses

6 Having Friends in High Places 5 Mentor Level Kram (1985) and Kram and Isabella (1985) examined and reported differences in mentoring relationships based upon whether the mentor was a peer or a hierarchical superior. The effect of differences in supervisor versus peer mentoring have been further investigated by Douglas and Schoorman (1988). Briefly, these studies found that supervisors provide more vocational mentoring than psycho-social mentoring. Further research by Allen and colleagues (Allen, Maetzke, & Russel, 1994; Allen, McManus, & Russell, 1999; Allen, Russell, & Maetzke, 1997) and Ensher, Thomas, and Murphy (2001) supports the peer relationship as providing both vocational and psycho-social mentoring. However, beyond peer and supervisory relationships, research has been lacking on the magnitude of the effect of hierarchical level when the mentor is an immediate supervisor versus a higher-level person in the organization. Thus, one unanswered question concerns the effect of hierarchical level of non-peer mentors. In this regard, Burke (1984) discusses the implications of having a mentor who is at a high level in the hierarchy, suggesting that when the mentor is a high-level person, the protégé gains more from the mentoring relationship. Ragins (1997) considers the asymmetrical power relationship inherent in diversified mentoring dyads. Power as defined by Ragins (1997) is sociologically rooted in group membership. While the issue of power plays a central role, Ragins focused on group status and organizational power with members of minority groups restricted in their capacity to wield power. The suggestions of Ragins support Burke s implications of the rewards of greater power. Thus, it might be expected that one important structural characteristic of non-peer mentoring dyads is the hierarchical level of the mentor. Higgins and Kram (2001) investigate a reconceptualization of mentoring as a network of developers throughout the protégé s career. In this developmental network perspective, the

7 Having Friends in High Places 6 central issues of network diversity and strength of relational ties strongly influence outcomes received. However, from a power perspective, it is reasonable to infer that the composition of the mentoring constellation would benefit from a high level mentor. The higher in the organization is the mentor s position the stronger the overall effect of mentoring as well as the stronger the effect of each separate mentoring function thereby improving the effectiveness of the developmental network. Duration and Lapse of Mentoring Kram (1985) discussed phases of the mentoring relationship, indicating that mentoring relationships cycle through initiation, cultivation, separation, and redefinition. Kram's research placed general time boundaries on these phases of development, noting that the initiation phase occurred during a period of six months to one year, the cultivation phase spanned a period from two to five years, the separation phase ranges from six months to two years, and an indefinite period was specified for the redefinition phase. From this, it seems clear that time is an important factor in mentoring relationships. However, time, as a key variable has not been empirically examined. For this reason, the current research examines duration of the mentoring relationship in the expectation of finding it to have a direct effect on protégé career outcomes, as well as possibly moderating relationships between mentoring functions and career outcomes. Lapse in mentoring (the period of time that the protégé has been without a mentor) will also be examined as a time variable. Persons having a lapse in mentoring may be expected to suffer with respect to career outcomes vis-à-vis protégés who currently have on-going mentoring relationships. Therefore, lapses in mentoring relationships should have negative direct on protégé career outcomes, as well as possibly moderating relationships between social support, coaching, and role modeling and protégé career outcomes.

8 Having Friends in High Places 7 Summary Based upon the discussion above, it is hypothesized that the level of the mentor, the duration of the mentoring relationship, and any lapse in mentoring may moderate relationships between the three mentoring functions and protégé career outcomes. However, these structural characteristics should have direct effects as well. Having a higher-level mentor should enhance protégés outcomes, as should mentor-protégé relationships of longer duration. Finally, any lapse in mentoring is expected to have a detrimental impact on protégé career outcomes. Although a number of career outcome variables could be examined, five were selected for the current study based upon their theoretical importance and use in prior research on mentoring and in the literature on careers and (cf., Feldman, 1988; Hall, 1976; Hunt & Michael, 1983; Kram, 1983, 1985; Stumpf & London, 1981). The five items reflect beneficial outcomes of successful mentoring relationships and include the protégé s: (1) expectations for career advancement within the firm (Career Expectations) (Chao et al., 1992; Scandura, 1992), (2) intent to pursue an accounting career entirely within the current employer (Commitment to the Firm) (Colarelli & Bishop, 1990; Scandura & Viator, 1994), (3) reported mean hours worked in the peak (busy) season (a measure of motivation or behavioral commitment) (Baugh et al., 1996), (4) number of direct reports supervised (a measure of job responsibility) (Dreher & Ash, 1990), and (5) annual salary (Dreher & Ash, 1990). METHOD Sample A random sample of Certified Public Accounting professionals (CPAs) was obtained using mailing lists provided by the American Institute of Certified Public Accountants (AICPA). The sample involved staff accountants from only large CPA firms and only those holding staff-

9 Having Friends in High Places 8 level positions. Potential respondents were mailed a survey packet that included a cover letter (identifying the study as AICPA-sanctioned but university- conducted), a survey, and a postagepaid envelope. Participant anonymity was guaranteed, and they were not asked to identify themselves. The overall response was 34.1% (1024 respondents), a reasonably good rate, particularly since, Mail surveys with response rates over 30 percent are rare. Response rates are often only about 5 or 10 percent (Alreck & Settle, 1985, p. 45). Of the 1024 respondents, 236 indicated that they had not had a mentor during their careers. Since this study dealt with mentoring relationships, only respondents who have had or were currently involved in a mentoring relationship were considered. Removing 2 missing data cases on the question regarding experienced mentoring relationship resulted in a usable sample of 786 for the current study. Sample demographics are reported in table 1. As shown in Table 1, the sample was 68% male, the average age of the respondents was 30 years and the majority of the respondents were white (88%). Average tenure in their current position was 2.2 years. Their average salary was $46,198 and, on the average, they supervised around 10 direct reports. With respect to who the mentors were, 59% were partners, 36% were managers, while 5% were peers. Perhaps not surprisingly, almost all of the mentors were male (91%). Mentor race was not collected. The average duration of the mentoring relationship was 4.3 years Insert Table 1 about here Measures Mentoring functions. To assess the three mentoring functions of this research, a preliminary pool of 20 items was developed, based upon the mentoring literature (Burke, 1984;

10 Having Friends in High Places 9 Hill, Rouner & Bahniuk, 1987; Kram, 1985). These items were then pilot-tested on a sample of 30 CPAS in two large Midwest CPA firms; several items were changed as a result of this pilot test sample s comments. The revised pool of 20 items was next examined for consonance with the literature s treatment of social support, coaching and role modeling and the best 4 items selected for each dimension. These items were included on the survey questionnaire administered the sample, asking the respondents to indicate their agreement with each of the 12 statements about their mentors. The response categories were Strongly Disagree, Disagree, Neutral, Agree, and Strongly Agree, scored 1 to 5, respectively. The three mentoring scales were next examined for their empirical distinctiveness by using LISREL 8.3 maximum likelihood confirmatory factor analysis (Joreskog & Sorbom, 1996) to statistically assess the goodness-of-fit of five rival models to the current data. Following convention (cf. Joreskog & Sorbom, 1996; Widaman, 1985), these analyses used common factor models with correlated factors and uncorrelated error. The first model was a single-factor model, assuming that the three mentoring dimensions are not separate and distinct. The second was a two-factor model with the social support and coaching items forming one factor and role modeling items forming a second factor. The third model had two factors, with social support and role modeling comprising one factor and coaching forming the second factor. The fourth model had a first factor consisting of the coaching and role modeling items and a second factor comprised of the social support items. Finally, a three-factor model was examined, with the social support items loading on only the first factor, the coaching items loading on the second factor, and the role modeling items loading on the last factor. For all these models, no crossloadings or error correlations were estimated to inflate model fit (MacCallum, 1986).

11 Having Friends in High Places 10 The five models were assessed by the Root Mean Square Error of Approximation (RMSEA), the Bentler-Bonnett Non-Normed Fit Index (NNFI), the Bentler Comparative Fit Index (CFI), and the Goodness-Of-Fit Index (GFI) provided by the LISREL output. (e.g., Marsh, Balla, & McDonald, 1988; Mulaik, James, Van Alstine, Bennett, Lind, & Stillwell, 1989). Comparisons between the models were undertaken using chi-square likelihood test for nested models described by Bentler and Bonett (1980) and others (e.g. Bollen, 1989; Joreskog & Sorbom, 1993; Kline, 1998; Widaman, 1985), as well as by examining differences in RMSEA, NNFI, CFI and GFI. Though theory supports a three-factor mentoring model, the fit of the simpler single factor model was tested against the three-factor model to validate the ability to test more complex models. If the simpler model could not be rejected, there would be no need to go forward with further models since the observed variables would appear to only measure one domain. The confirmatory analysis and rival model comparisons revealed that the three-factor model was clearly superior to all four rival models. Its GFI (.94) (AGFI =.91) were substantially better than those of the one-factor model (GFI =.82; AGFI =.74). Furthermore, the chi-square difference test yielded a highly significant value (X 2 = 538.4, df = 3, p <.001), and the Non- Normed Fit Index for the single factor model is less than.80 (NNFI =.58) further indicating its inadequacy in accounting for the observed covariances (Kline, 1998). The statistics for the second (GFI =.86; AGFI =.80), third (GFI =.88; AGFI =.82), and fourth (GFI =.88; AGFI =.83) models (with two factors each) likewise indicated their inferiority to the three-factor model. For each of these rival models, their chi-square difference tests all similarly indicated the superiority of the three-factor model (these figures were X 2 = , , and , respectively; all df s = 2, all p s <.001).

12 Having Friends in High Places 11 Table 2 presents the obtained three-factor model, with the top part showing factor loadings and model fit summary and the bottom part giving the factor intercorrelations. As shown in Table 2, the factor loadings are generally good for each of the three factors. All except Mentor15 ( I respect this person s knowledge of the accounting profession ) exceed the.40 level commonly considered meaningful in factor-analytic investigations (Ford, MacCallum, & Tait, 1986) and all significant (p <.05) Insert Table 2 about here The factor intercorrelations shown in Table 2 are also all significant, as might be expected with a correlated (oblique) factor analysis (cf. Joreskog & Sorbom, 1996). However, these are estimated disattenuated relationships (corrected for measurement error), so that while there is significant shared variance among the three constructs underlying the scales, it is estimated to not exceed 29% (the largest factor correlation,.54 squared). This clearly supports the use of the three mentoring scales, as well as their treatment as empirically (as well as theoretically) separate dimensions. Covariates. Studies of mentoring need to statistically control for relevant exogenous variables so as to reduce the plausibility of extraneous factors as determinants of any obtained results (Fagenson, 1989; Ragins & McFarlin, 1990; Scandura, 1992). Thus, for the present study, protégé gender, protégé race, protégé age, protégé number of dependent children, and mentor gender, were statistically controlled in the analyses (marital status was omitted due to its redundancy with the number-of-children variable). Furthermore, the protégé s level in the organization (junior or senior staff) and the protégé s years of tenure in the organization were

13 Having Friends in High Places 12 also used as covariates. These control variables were measured by single-item demographic questions on the survey, and they were selected for inclusion based on prior research, which suggests that each may be independently capable of impacting on career outcomes (cf., Colarelli & Bishop, 1990; Dreher & Ash, 1990; Feldman, 1988; Hall, 1976; Johnson & Scandura, 1994; Levinson, 1978; Schein, 1971). Structural characteristics of mentoring dyads. Mentor level was operationalized as the protégé s indication of whether their mentor was a peer (i.e., the same level, typically a junior associate), a manager (i.e., a direct reporting relationship) or a partner in the firm (i.e., a higherlevel mentor). The duration of the mentoring relationship was measured by subtracting the protégé s report of when the relationship started from when the relationship ended. Similarly, lapse in mentoring was assessed by the length of time since the protégé reported that the mentoring relationship had ended. Career outcomes. Career outcomes for protégés were selected for study based on the literature on careers in organizations (Feldman, 1988; Hall, 1976). Thus, for the current research, a four-item measure of Career Expectations was employed, assessing the degree to which the protégé felt that he/she would be promoted in the current firm. The exact items employed were, I expect to be promoted at this firm, To get ahead, I will have to change firms, (reversedscored), I expect that I will be a partner at this firm, and Currently my career is at a stalling point (reversed-scored). The response categories employed were the same as those used for the mentoring items. Commitment to the firm was assessed by asking the protégés two questions, The position I would like to hold in 5 years is, and The position I desire at the highest point in my career is ; seven options were listed for the respondents to chose. For the current analysis, these were scored to indicate intent to pursue a career only within the current employer.

14 Having Friends in High Places 13 The scoring was as follows: 2 = Current Position or Higher Position With This Firm, and 1 = Similar Position With Other Firm, Higher Position With Other Firm, Position in Industry, Sole Partnership, or Other. Since this measure had only two dichotomous items, coefficient alpha was not used to estimate internal consistency but, rather, the.61 correlation between the two items was adjusted by the split-half formula (cf., Nunnally, 1978). This split-half reliability estimate was.76. For the present sample, three additional career outcomes were examined using data provided by the respondents. The Mean Number of Hours Worked in the Peak Season was employed as an index of motivation and behavioral commitment to the organization, while the Number of Direct Reports Supervised was used as a measure of responsibility entrusted to the protégé. Both of these variables were assessed by single-item demographic questions on the survey, as was current Annual Salary. Analyses The internal reliabilities of all multi-item measures were first computed, using coefficient alpha for all but the estimate of Commitment to the Firm (see discussion above). As shown in Table 1, all of the obtained reliability estimates appeared acceptable (cf. Nunnaly, 1978). The intercorrelations among the main study were next computed, to examine direct relationships without controlling for the study covariates. Finally, hierarchical moderated regression analysis procedures were used to control for the covariates and to test the effects of mentoring functions and structural characteristics on career outcomes for the protégés (Cohen & Cohen, 1975; Cohen et al., 2003). First, the five outcome variables (Career Expectations, Commitment to the Firm Mean Hours Worked in Peak Season, Number of Direct Reports Supervised, and Annual Salary) were regressed separately on the covariates (block 1). Next, the

15 Having Friends in High Places 14 three mentoring functions were entered separately in each of the regression equations (block 2). Then the structural characteristics (Mentor Level, Mentor Duration, and Lapse in Mentoring) were entered separately (block 3). Finally the interaction of each mentoring function and each structural variable was entered (e.g., Social Support x Mentor Level)(block 4). Standardized regression coefficients were computed for each variable as well as the increment in explained variance (R 2 Change). Results Table 3 shows the intercorrelations among the main study variables the structural characteristics, mentoring functions, and career outcomes. As shown in Table 3, the structural characteristics of mentoring dyads have significant relationships with the mentoring function and career outcome variables. Specifically, Mentor Level is significantly and negatively related to Social Support, indicating that higher-level mentors are not seen as providing as much support as are lower level mentors. Mentor Level is, however, significantly and positively associated with Career Expectations, Firm Commitment, Number of Direct Reports Supervised, and Annual Salary. Duration of Mentoring is positively related to Coaching, Firm Commitment, Number of Direct Reports Supervised by the protégé, and the protégé s Annual Salary. As expected, lapse in Mentoring is negatively associated with Career Expectations; however, unexpectedly, it is also positively related to the protégé s Annual Salary (perhaps an artifactual result of both being related to other time variables, such as age and tenure; such effects are statistically controlled in the regression analyses reported below). Social Support has a positive relationship with Mean Hours Worked in Peak Season. Coaching has positive relationships with Career Expectations, Commitment to the Firm, and Mean Hours Worked in Peak Season. Role Modeling has a positive relationship with Career Expectations.

16 Having Friends in High Places Insert Table 3 about here Results for Mentor Level The results of the hierarchical regression analyses for Mentor Level as a main effect and as a moderator are shown in Table 4. After controlling for the covariates, Social Support accounts for additional variance in Career Expectations an in Hours Worked. On the other hand, Mentor Level accounts for additional significant variance in all of the outcome variables except for Hours Worked. For the Coaching (vocational) mentoring function, Table 4 shows that after controlling for the covariates, the coaching function accounts for additional significant variance in three of the five career outcomes: Career Expectations, Commitment to the Firm, and Hours Worked. The Mentor Level variable produces similar effects to those noted above for the analysis with Social Support. The results for the Role Modeling mentoring function are also shown in Table 4. After controlling for the covariates, Role Modeling accounts for significant variance in Career Expectations. As in the two analyses discussed above, Mentor Level is significant in four of the equations, indicating that the level of the mentor clearly affects career outcomes. Mentor Level does not act as moderator in any of the equations indicating no moderating effect for the Social Support-, Coaching-, and Role Modeling x Mentor Level interaction Insert Table 4 about here

17 Having Friends in High Places 16 Results for Duration of the Mentoring Relationship The results of the hierarchical regression analyses for Mentor Duration as a main effect and as a moderator are shown in Table 5. Controlling for the covariates, Social Support adds significant incremental variance for Career Expectation and Hours Worked, while Mentoring Duration adds significant variance to all protégé outcomes except for Hours Worked. With respect to mentor Coaching, this vocational mentoring function contributes incremental variance to three of the five regression equations for Career Expectations, Commitment to the Firm, and Hours Worked by the protégé. Mentoring Duration adds significant variance for Career Expectations, Firm Commitment, Number of Direct Reports Supervised, and Annual Salary. Role modeling shows a significant main effect for Career Expectations, while Mentoring Duration adds significant incremental variance for four of the five career outcomes: Career Expectations, Commitment to the Firm, Number of Direct Reports Supervised, and Annual Salary. No significant interaction exists for Social Support, Coaching, and Role Modeling with Mentoring Duration in predicting the outcome variables Insert Table 5 about here Results for Lapse in Mentoring The hierarchical regression results for the Lapse in Mentoring variable are shown in Table 6. As can be seen there, Social Support from the mentor adds significant variance for the Career Expectations and Hours Worked variables. The Lapse in Mentoring variable adds to the explained variance in Career Expectations. A significant interaction is also obtained for Social

18 Having Friends in High Places 17 Support x Lapse in Mentoring for the Expectations outcome variable. Both Social Support and Lapse in Mentoring had significant main effects as well as an interaction effect when linearly partialed out. However, the opposite signs of the regression coefficients signify a buffering interaction where the impact of one predictor increases in value while the impact of the other predictor is diminished (Cohen et al., 2003). Coaching from the mentor shows a direct effect, as reported in Table 6, for Career Expectations, Commitment to the Firm, and Mean Number of Hours Worked in Peak Season. Lapse in Mentoring is significant for Career Expectations and no significance is found for the interaction term, Coaching x Lapse in Mentoring. With respect to Role Modeling, it has a main effect for significant additional variance in Career Expectations, while Lapse in Mentoring has a main effect on Career Expectations. None of the interactions are significant Insert Table 6 about here Discussion Result of the LISREL 8 maximum likelihood confirmatory factor analyses provides support for the continued use of the tripartite treatment of mentoring functions. Social Support (psycho-social), Coaching (vocational), and Role Modeling emerged as separate and distinct dimensions of the mentoring process. The results of the present study confirms the appropriateness of conceptualizing, measuring, and examining these dimensions separately,

19 Having Friends in High Places 18 particularly given that the three mentoring functions were differentially related to the five protégé career outcomes (see Tables 3, 4, 5 and 6). The results of this investigation also support the structural and temporal characteristics of mentoring dyads and the use of these factors in the study of mentoring at work. It appears that mentor level, duration, and lapse in mentoring have significant effects on career outcomes of proteges since these characteristics had main effects on the five career outcome dependent variables, even after controlling for exogenous variables (covariates) and the three mentoring functions. These results, then, appear to support suggestions by Burke (1984), and Kram and Isabella (1985) that the mentor s hierarchical level can have a critical impact on the career outcomes of proteges. Mentor s hierarchical may reflect power and access to greater levels of organizational resources indirectly benefiting the protégé. Ragins et al. (2000) included mentor level as a study variable however, their analysis focused on attitudinal outcomes and not career effects. They conclude that the dominating factor in a mentoring program may be the degree of satisfaction with the mentor. Attitudinal responses may reflect psychological processes. In this study, the higher the level of the mentor, the better the career benefits. These findings also have important implications for the study of mentoring and indicate that future studies should include questions regarding the hierarchical level of the mentor, since it appears that who the mentor is may, at least sometimes, be as or more important than what the mentor does. With respect to the time dimension of mentoring, it should be noted that Kram, (1983) outlines a phase model in which mentoring relationships change over time. In this regard, the results of the present study further support the importance of time as a factor in mentoring relationships since the duration of mentoring and having a lapse in mentoring had significant effects on career outcomes. In particular, duration of the mentoring relationship showed

20 Having Friends in High Places 19 significant and positive associations with all of the career outcomes measured, except for the number of hours worked, after controlling for covariates and the mentor functions. These main effects indicate that longer duration relationships generally enhanced the career outcomes of proteges. Noting from Table 1 that the average length of these relationships was 4.3 years suggests that many of these relationships were probably in the cultivation stage identified by Kram (1983). These effects indicate that longer duration relationships generally enhanced the career outcomes of protégés. Duration may also be related to quality of mentoring relationship. Unsuccessful relationships may end earlier than successful relationships (Ragins & Scandura, 1997; Scandura, 1998). Thus, satisfaction with the mentor may be reflected in the duration of the relationship with more successful relationships enjoying a higher satisfaction level thereby prolonging the parties desire to remain in the mentorship. These results further support the suggestion that the mentoring phase (as indicated by duration) is an important aspect of the mentor-protégé dyad. Moreover, from Tables 3,4,5 and 6 it appears that the duration effect was perhaps stronger and more important than the mentoring functions themselves, indicating that this critical variable should be included in future studies of mentoring. The second temporal dimension investigated in this study was lapse in mentoring as reported by the protégés (how long the protégé had been without an organizational mentor). This may roughly correspond to the redefinition phase in the Kram (1983) model, following the separation and breakup of the mentor-protégé dyad. Lapse showed significant negative effects on the attitudinal career outcome variables of Career Expectations, after controlling for the covariates and the mentor functions in the hierarchical regression analyses. However. Lapse in mentoring was positively associated with the Number of Hours Worked During the Peak Season, indicating that mentoring was associated with a reduced workload reported by the protégés. This

21 Having Friends in High Places 20 may be due to mentors assignment of less time-consuming tasks to protégés, or perhaps helping the proteges perform assigned tasks more efficiently. In any event, this is clearly an area for future research on mentoring because it suggests that having a mentor may buffer protégés from the effects of work-related stress (indicated here by workload during peak season) as supported by Allen et al. (1999). Although interaction terms were also tested in the present study so as to determine if the structural characteristics of mentoring dyads had moderating effects on the mentoring functions, such interactions were significant in only one of the forty-five instances tested (Social Support x Lapse in Mentoring). Though significant at the.05 level, caution should be exercised in interpreting this interaction since 5% of the interaction would be expected to be significant by chance alone. Thus, the mentor level, duration, and lapse variables do not seem to be strong moderators of mentoring function effects, although they have their own significant direct or main effects. Before concluding, it should perhaps be mentioned that since the data of this investigation were collected using the same instrument, concern might arise as to whether the obtained results are merely common-method artifacts. Although such effects cannot be ruled out, their plausibility can be reduced. Here, it should be noted that a principal factor analysis conducted on the measures used did not reveal anything remotely resembling a general factor, reducing the likelihood that the results are strictly artifactual (Podsakoff & Dalton, 1987). Also, since relatively "objective" measures (pay, number of direct reports, and peak hours worked) formed a majority of the dependent variables, common-method bias does not seem particularly problematic (respondents are probably less likely to distort their responses to such measures; Nunnally, 1978).

22 Having Friends in High Places 21 In conclusion, the results reported in this study of the structure and function of mentoring dyads indicates that structural characteristics of mentoring dyads have important effects on career outcomes for protégés. The mentor's level in the organizational hierarchy, the duration of the mentoring relationship, and the length of lapse in mentoring all showed main effects on outcomes, in addition to the mentoring effects discussed above. Few investigations of mentoring have tested for contextual and temporal characteristics of the mentoring dyads studied, despite theoretical frameworks and qualitative research which suggests their importance (Kram, 1985; Kram & Isabella, 1985). This may explain the sometimes tenuous relationships found between mentoring and career outcomes reported in the literature on mentoring (Scandura, 1992). Future studies of mentoring dyads should therefore include the structural characteristics examined here and continue to specify and test other structural characteristics and their potential moderating effects on mentoring functions. Hopefully, such research should further add to our knowledge about mentoring and its effects on proteges in organizations.

23 Having Friends in High Places 22 REFERENCES Allen, T. D., McManus, S. E., & Russell, J. E Newcomer socialization and stress: Formal peer relationships as a source of support. Journal of Vocational Behavior, 54: Allen, T. D., Maetzke, S. B., & Russell, J. E Formal peer mentoring: Factors related to protégé s satisfaction and willingness to mentor others. Paper presented at the Southern Management Association Meeting, New Orleans, LA. Allen, T. D., Russell, J. E., & Maetzke, S. B Formal peer mentoring: Factors related to protégé s satisfaction and willingness to mentor others. Group & Organization Management, 22: Alreck, P. A., & Settle, R. B The survey research handbook. Homewood, IL: Irwin. Baugh, S. G., Lankau, M. L., & Scandura, T. A An investigation of the effects of protégé gender on responses to mentoring. Journal of Vocational Behavior, 49: Bollen, K. A Structural equations with latent variables. New York: John Wiley & Sons. Burke, R. J Mentors in Organizations. Group and Organization Studies, 9: Burke, R. J., & McKeen, C. A Developing formal mentoring programs in organizations. Business Quarterly, 53: Cao, L. T. & Buchanan, P. G A profile of the woman management accountant. The Woman CPA, April, Carmer, S. G. & Swanson, M. R An evaluation of ten pairwise multiple comparison procedures by Monte Carlo methods. Journal of the American Statistical Association, 68:

24 Having Friends in High Places 23 Chao, G.T., Walz, P. M., & Gardner, P. D Formal and informal mentorships: A comparison of mentoring functions and contrast with nonmentored counterparts. Personnel Psychology, 45: Cohen, J., Cohen, P., West, S. G., & Aiken, L. S Applied multiple regression/correlation analysis for the behavioral sciences, 3 rd ed. Mahwah, NJ: Lawrence Erlbaum Associates. Colarelli, S. M., & Bishop, R. C Career commitment: Functions, correlates, and management. Group & Organization Studies, 15: Douglas, C. A. & Schoorman, F. D The impact of career and psychosocial mentoring by supervisors and peers. Paper presented at the annual Academy of Management Convention. Dreher, G. F. & Ash, R. F A comparative study of mentoring among men and women in managerial, professional, and technical positions. Journal of Applied Psychology, 75: Ensher, E. A., Thomas, C., & Murphy, S. E Comparison of traditional, step-ahead, and peer mentoring on protégés support, satisfaction, and perceptions of career success: A social exchange perspective. Journal of Business and Psychology, 15: Fagenson, E The mentor advantage: Perceived career/job experiences of proteges versus non-proteges. Journal of Organizational Behavior, 10: Fagenson-Eland, E. A., Marks, M. A., & Amendola, K Perceptions of mentoring relationships. Journal of Vocational Behavior, 51: Feldman, D. C Managing careers in organizations. Glenview, IL: Scott, Foresman. Ferris, G. R. & Kacmar, K. M Perceptions of organizational politics. Journal of Management, 18:

25 Having Friends in High Places 24 Ford, J. K., MacCallum, R. C., & Tait, M The application of exploratory factor analysis in applied psychology: A critical review and analysis. Personnel Psychology, 39: Hall, D. T Careers in organizations. Pacific Palisades, CA: Goodyear Publishing. Hill, S. E., Rouner, D., & Bahniuk, M Mentoring and other communication support systems in the academic setting: Development of an instrument. Paper presented at the annual International Communication Association Convention. Huitema, B. E The analysis of covariance and alternatives. NY: John Wiley Interscience. Hunt, D. M. & Michael, C Mentorship: A career training and development tool. Academy of Management Review, 8: Jennings, E. E Routes to the executive suite. NY: McGraw-Hill. Johnson, N. B., & Scandura, T. A The effect of mentorship and sex-role style on malefemale earnings. Industrial Relations, 33: Joreskog, K. G., & Sorbom, D LISREL 8: User s reference guide. Chicago: Scientific Software International, Inc. Kline, R. B Principles and practice of structural equation modeling. New York: Guilford Press. Kram, K. E Phases of the mentoring relationship. Academy of Management Journal, 26: Kram, K. E Mentoring at work. Glenview, IL: Scott, Foresman. Kram, K. E. & Isabella, L. A Mentoring alternatives: The role of peer relationships in career development. Academy of Management Journal, 28: Levinson, D The seasons of a man s life. New York: Knopf

26 Having Friends in High Places 25 MacCallum, R Specification searches in covariance structure modeling. Psychological Bulletin, 100: Manz, C. C. & Sims, H. P Vicarious learning: The influence of modeling on organizational behavior. Academy of Management Review, 6: Murray, M Beyond the myths and magic of mentoring: How to facilitate an effective mentoring program. San Francisco: Jossey-Bass. Noe, R. A An investigation of the determinants of successful assigned mentoring relationships. Personnel Psychology, 41: Nunnally, J. C Psychometric theory, 2nd Ed. NY: McGraw-Hill. Olian, J. D., Carroll, S. J., Giannantonio, C. M., & Feren, D. B What do proteges look for in a mentor? Results of three experimental studies. Journal of Vocational Behavior, 33: Ostroff, C., & Kozlowski, S The role of mentoring in the information gathering processes of newcomers during early organizational socialization. Journal of Vocational Behavior, 42: Podsakoff, P. M. & Dalton, D. R Research methodology in organizational studies. Journal of Management, 13: Ragins, B. R Diversified mentoring relationships in organizations: A power perspective. Academy of Management Review, 22: Ragins, B. R., Cotton, J. L., & Miller, J. S Marginal mentoring: The effects of mentor, quality of relationship, and program design on work and career attitudes. Academy of Management Journal, 43:

27 Having Friends in High Places 26 Ragins, B. R., & McFarlin, D Perceptions of mentor roles in cross-gender mentoring relationships. Journal of Vocational Behavior, 37: Ragins, B. R., & Scandura, T. A Burden or blessing/ Expected costs and benefits of being a mentor. Journal of Organizational Behavior, 20: Ragins, B. R., & Scandura, T. A The way we were: Gender and the termination of mentoring relationships. Journal of Applied Psychology, 82: Ragins, B. R., & Scandura, T. A Gender differences in expected outcomes of mentoring relationships. Academy of Management Journal, in press. Roche, G Much ado about mentors. Harvard Business Review, 57: Scandura, T. A Dysfunctional mentoring relationships and outcomes. Journal of Management, 24: Scandura, T. A Scandura, T. A Mentorship and career mobility: An empirical investigation. Journal of Organizational Behavior, 13: Scandura, T. A., & Viator, R Mentoring in public accounting firms: An analysis of mentor-protégé relationships, mentorship functions, and protégé turnover intentions. Accounting, Organizations & Society, 19: Scandura, T. A., & Williams, E. A Initiating mentoring: Contrasting the reports of protégés in assigned and informal relationships. Proceedings of the Southern Management Association, New Orleans, LA. Schein, C. K The individual, the organization, and the career. Journal of Applied Behavioral Science, 1:

28 Having Friends in High Places 27 Schockett, M. & Haring-Hidore, M Factor analytic support for psychosocial and vocational mentoring functions. Psychological Reports, 57: Shelton, C. K The relationship of mentoring and behavioral style to selected job success variables. Unpublished doctoral dissertation, Northern Illinois University. Stevens, J Applied multivariate statistics for the social sciences. Hillsdale, NJ: Lawrence Erlbaum Associates. Stumpf, S. A. & London, M Management promotions: Individual and organizational factors influencing the decision process. Academy of Management Review, 6: Widaman, K. F Hierarchically nested covariance structure models for multitraitmultimethod data. Applied Psychological Measurement, 9: Williams, E. A

29 Having Friends in High Places 28 Table 1 Means and Standard Deviations for Study Variables Variable Mean SD Reliability Covariates Protégé Age (Years) (N=784) Protégé Gender (1=Male, 2=Female) 1.32 (N=785) Protégé Race (1=Nonwhite, 2=White) 1.88 (N=784) Protégé Org. Tenure (Years) 2.29 (N=769) Protégé No. of Children.69 (N=786) Protégé Level (1=Junior, 2=Senior) 1.88 (N=778) Mentor Gender (1=Male, 2=Female) 1.09 (N=783) Relationship Attributes Mentor Level (1=Peer, 2=Manager, 3=Partner) 2.54 (N=783) Duration of Mentoring (Years) 4.37 (N=781) Lapse in Mentoring (Years).61 (N=768) Mentoring Functions Social Support (Range 4-20) (N=784) Coaching (Range 4-20) (N=786) Role Modeling (Range 4-20) (N=783) Career Outcomes Career Expectations (Range 4-20) (N=771) Commitment to the Firm (Range 2-4) 3.12 (N=769) a Mean Hours Worked in Peak Season (N=717) Number of Direct Reports Supervised 9.77 (N=685) Annual Salary ($U.S.) 46, (N=677) 15, a Split-half coefficient; all others are coefficient alphas.

30 Having Friends in High Places 29 TABLE 2 Mentoring Functions Confirmatory Factor Analysis Results Item Support Coaching Modeling 06. I share personal problems with this person I socialize with this person after work I exchange confidences with this person I consider this person to be a friend This person takes a personal interest in my career This person has placed me in important assignments This person gives me special coaching on the job This person has devoted special time and consideration to my career 12. I try to model my behavior after this person I admire this person s ability to motivate others I respect this person s knowledge of the accounting profession 17. I respect this person s ability to teach others Root Mean Square Error of Approx. (RMSEA) = Non-Normed Fit Index (NNFI) = 0.85 Comparative Fit Index (CFI) = 0.89 Goodness of Fit Index (GFI) = 0.94 Adjusted Goodness of Fit Index (AGFI) = 0.91 Factor Intercorrelations Factor Support Coaching Modeling Coaching Modeling Note. The non-zero parameters which are shown were the only ones that were estimated; all of these are statistically significant (p<.05).

31 Having Friends in High Places 30 TABLE 3 Intercorrelations of Main Study Variables Variable Mentor Level Duration of.254** Mentoring 3. Lapse in Mentoring ** Social Support -.277** Coaching * ** Role Modeling **.337** Career Expectations.128** ** **.097** Commit. to the Firm.235**.201** ** ** Hours *.091* * No. Supv..143**.222** **.112* Annual Salary.373**.430**.178** *.263** ** Note: 2-tailed significance test; pairwise treatment N=( ) *p<.05 **p<.01

32 Having Friends in High Places 31 TABLE 4 Moderated Hierarchical Regression Analysis for Mentor Level as Moderator Expectations Firm Commitment Hrs. Worked No. Supervised Annual Salary a Covariates (R 2 ).088***.063*** ***.366*** Beta +R 2 Beta +R 2 Beta +R 2 Beta +R 2 Beta +R 2 Mentor Level as Moderator SS.007*.082* *.086* L.016***.136***.027***.180*** *.102*.038***.215*** SS x L C.059***.245***.011**.103**.008*.086* L.016***.136***.027***.180*** *.102*.038***.215*** C x L RM.016***.127*** L.016***.136***.027***.180*** *.102*.038***.215*** RM X L a Covariates: Protégé gender, protégé race, protégé age, protégé number of children, protégé level, protégé organizational tenure, mentor gender. SS- Social Support, C- Coaching, RM- Role Modeling, L- Mentor Level *p <.05; **p <.01; ***p <.001