mentor functions and outcomes: a comparison of men …€¦ · mentor functions and outcomes: a...

Journal of Applied Psychology1999, Vol. 84, No. 4, 529-550

Copyright 1999 by the American Psychological Association, Inc.0021-9010/99/S3.00

Mentor Functions and Outcomes: A Comparison of Men and Women inFormal and Informal Mentoring Relationships

Belle Rose RaginsUniversity of Wisconsin—Milwaukee

John L. CottonMarquette University

The authors examined the effects of the type of mentoring relationship and the gendercomposition of the relationship on mentoring functions and career outcomes reported by 352female and 257 male proteges. Proteges of informal mentors viewed their mentors as moreeffective and received greater compensation than proteges of formal mentors. Proteges withinformal mentors also received more career outcomes than nonmentored individuals, but nosignificant differences were found between nonmentored and formally mentored individuals.The gender composition of the relationship affected mentoring functions and outcomes, andprotege gender interacted with the type of relationship to affect mentoring functions.

Mentoring relationships are a critical career resource foremployees in organizations. Mentors are individuals withadvanced experience and knowledge who are committed toproviding upward support and mobility to their protege'scareers (Hunt & Michael, 1983; Kram, 1985a). Mentorshelp their proteges by providing two general types of be-haviors or functions: career development functions, whichfacilitate the protege's advancement in the organization, andpsychosocial functions, which contribute to the protege'spersonal growth and professional development (Kram,1985a). The presence of a mentor is associated with an arrayof positive career outcomes: Proteges receive more promo-tions (Dreher & Ash, 1990; Scandura, 1992), have higherincomes (Chao, Walz, & Gardner, 1992; Dreher & Ash,1990; Whitely, Dougherty, & Dreher, 1991), and reportmore mobility (Scandura, 1992) and career satisfaction (Fa-genson, 1989) than nonproteges. Mentoring has also beenfound to have a positive impact on organizational socializa-tion (Ostroff & Kozlowski, 1993), job satisfaction (Koberg,Boss, Chappell, & Ringer, 1994), and reduced turnoverintentions (Viator & Scandura, 1991).

Belle Rose Ragins, School of Business Administration, Univer-sity of Wisconsin—Milwaukee; John L. Cotton, Department ofManagement, Marquette University.

An earlier version of this article was presented at the meeting ofthe National Academy of Management, August 1998, San Diego,California, and received the Best Applied Paper Award in the Ca-reers Division. This study was funded by a 1991 grant from theWomen's Bureau of the U.S. Department of Labor.

Correspondence concerning this article should be addressed toBelle Rose Ragins, School of Business Administration, P.O. Box742, University of Wisconsin, Milwaukee, Wisconsin 53201. Elec-tronic mail may be sent to [email protected].

Many organizations recognize the important benefits ofmentoring and have attempted to replicate informal men-toring relationships by creating formal mentoring programs(Burke & McKeen, 1989; Geiger-DuMond & Boyle, 1995;Zey, 1985). One key difference between formal and infor-mal mentoring relationships is that informal mentoring re-lationships develop spontaneously, whereas formal mentor-ing relationships develop with organizational assistance orintervention—usually in the form of voluntary assignmentor matching of mentors and proteges. A second distinctionis that formal relationships are usually of much shorterduration than informal relationships (Douglas, 1997).

Although it is clear that formal and informal mentoringrelationships differ in how they are formed and the length ofthe relationship, there is little research on whether formaland informal relationships differ in the functions mentorsprovide or the career outcomes proteges obtain during thementoring relationship. Many organizations simply assumethat formal relationships are as effective as informal rela-tionships and implicitly offer their employees formal rela-tionships as a substitute for informal mentoring relation-ships (Keele, Buckner, & Bushnell, 1987; Kram & Bragar,1992). Moreover, formal mentoring programs are beingimplemented across the nation: It is estimated that a third ofthe nation's major companies have a formal mentoringprogram (Bragg, 1989), and this figure is expected to con-tinue to increase (Murray, 1991). These formal mentoringprograms are being developed without the benefit or guid-ance of empirical research.

This situation has particular relevance for women, whoface greater barriers to developing informal mentoring re-lationships than men (Ragins & Cotton, 1991) and maytherefore be more likely to seek formal relationships as a

529

530 RAGINS AND COTTON

substitute for informal mentoring relationships. Addition-ally, many organizations target women for formal mentor-ing programs in the attempt to help them advance in theorganization and break through the "glass ceiling" (Cata-lyst, 1993; Kerry, 1994; Kram & Hall, 1996; Scott, 1992).However, these organizations may not be helping theirfemale employees if formal mentors are less effective thaninformal mentors.

Accordingly, the first and primary purpose of this study isto compare the mentoring functions and career outcomesassociated with formal and informal mentoring relation-ships. A second purpose of this study is to understandwhether the gender composition of the relationship affectsthe relationship's functions and outcomes and whether thiseffect varies by the type of mentoring relationship. Asdiscussed earlier, many organizations develop formal men-toring programs that focus either exclusively or primarily ontheir female employees. The assignment of a male or femalementor may produce different outcomes from the relation-ship (Ragins, 1989) and may therefore be an importantfactor to consider in the development of a formal mentoringprogram. This study's objective, therefore, is to providetimely information for organizations seeking to developmentoring programs and to contribute to the knowledgebase necessary for emerging mentorship theory.

Literature Review and Hypotheses

Mentor Functions

According to Kram's mentor role theory (1985a), men-tors can provide two broad categories of mentor functions.First, they provide career development functions, whichhelp proteges learn the ropes and facilitate the protege'sadvancement in the organization. Kram (1985a) theorizedthat mentors can provide five specific career developmentfunctions: sponsoring promotions and lateral moves (spon-sorship); coaching the protege (coaching); protecting theprotege from adverse forces (protection); providing chal-lenging assignments (challenging assignments); and in-creasing the protege's exposure and visibility (exposure).

Psychosocial functions compose the second broad cate-gory of mentor functions. These behaviors address interper-sonal aspects of the mentoring relationship and enhance theprotege's sense of competence, self-efficacy, and profes-sional and personal development. Career development func-tions depend on the mentor's power and position in theorganization, whereas psychosocial functions depend on thequality of the interpersonal relationship and the emotionalbond that underlies the relationship. Career developmentfunctions focus on the organization and the protege's career,whereas psychosocial functions affect the protege on a morepersonal level and extend to other spheres of life, such as theprotege's personal development. Kram (1985a) theorized

that mentors may provide four psychosocial functions: help-ing the protege develop a sense of professional self (accep-tance and confirmation), providing problem-solving and asounding board (counseling), giving respect and support(friendship), and providing identification and role modeling(role modeling). It is important to note that mentoring is notan all or none phenomenon; a given mentor may provide allof just some of these functions.

Formal and Informal Mentoring Relationships

There are distinct differences between formal and infor-mal mentoring relationships that may impact the mentor'sfunctions and the career outcomes of the relationship. Thesedifferences involve the way the relationship is initiated, thestructure of the relationship, and the processes involved inthe relationship.

Initiation of relationship. Informal mentoring relation-ships develop on the basis of mutual identification and thefulfillment of career needs. Mentors select proteges who areviewed as younger versions of themselves, and the relation-ship provides mentors with a sense of generativity, or con-tribution to future generations (Erikson, 1963). Mentors areusually in mid-career stages that involve reassessment oflife accomplishments (Erikson, 1963; Kram, 1985a; Levin-son, Darrow, Klein, Levinson, & McKee, 1978), and gen-erativity helps mentors avoid stagnation and allows them toprogress to the next life stage. Proteges select mentors whoare viewed as role models. Proteges are in early careerstages that involve developing a sense of professional iden-tity, and role modeling helps proteges advance through thisstage. This mutual identification leads to the often-citedintensity of the informal relationship and the parallels drawnbetween mentoring and parent-child relationships.

Informal mentoring relationships also develop on thebasis of perceived competence and interpersonal comfort(Allen, Poteet, & Burroughs, 1997; Kalbfleisch & Davies,1993; Kram, 1983, 1985a; Olian, Carroll, & Giannantonio,1993; Olian, Carroll, Giannontonio, & Feren, 1988). Men-tors tend to select high-performing proteges who are con-sidered rising stars or even diamonds in the rough. Simi-larly, proteges select mentors with desired expertise.Members of informal mentoring relationships select part-ners they enjoy working with and often report a mutualattraction or chemistry that sparks the development of therelationship (Kram, 1983, 1985a).

In contrast, members of formal mentoring relationshipsare typically assigned to one another by a program coordi-nator on the basis of application forms submitted by thepotential mentor and protege (Douglas, 1997; Gaskill, 1993;Murray, 1991). In many cases, the mentor and protege donot even meet until after the match has been made. Thus, incontrast to informal relationships, identification, role mod-eling, and interpersonal comfort do not play a role in the

MENTOR FUNCTIONS AND OUTCOMES 531

development of formal relationships. It is therefore reason-able to expect that the psychosocial functions of role mod-eling, friendship, and counseling may be less in formal thaninformal mentoring relationships. Formal mentoring rela-tionships are also less likely to be founded on mutualperceptions of competency and respect. Formal mentors areselected on the basis of their competency, but this judgmentis made by the program coordinator rather than the protege(Gaskill, 1993; Murray, 1991; Phillips-Jones, 1983). Addi-tionally, formal mentors may view their proteges as at-riskperformers who enter the program because they need reme-dial attention (Ragins, 1997a). It is reasonable to expect thatthe acceptance and confirmation mentor functions, whichare founded on respect and perceived competency, will beless in formal than informal mentoring relationships.

Structure of relationship. Formal and informal mentor-ing relationships differ in the length and formality in therelationship. Informal relationships last between 3 and 6years (Kram, 1985a), whereas formal relationships are usu-ally contracted to last between 6 months and 1 year (Mur-ray, 1991; Zey, 1985). Members of informal relationshipsmeet when desired, but the mode, frequency, and location ofcontact for formal relationships are often specified in acontract signed by both parties (Murray, 1991; Zey, 1985).The goals of formal relationships are specified at the start ofthe relationship and are screened by the program coordina-tor. In contrast, the goals of informal relationships evolveover time and adapt to the career needs of the individuals.

There are three potential outcomes of these differentstructures. First, informal mentoring relationships havemore time to build psychosocial and career developmentfunctions. Mentoring may not have an immediate effect oncareer outcomes; it may take time for the benefits of men-toring to materialize. Kram (1985a) theorized that the ben-efits of mentoring extend beyond the duration of the rela-tionship, and Chao (1997) found that the advantages ofbeing mentored continue over time. This time-lag effect isparticularly relevant when considering promotions andcompensation, which usually change only once a year.Therefore, informal mentors' career interventions may havemore time to reach fruition than formal mentors' careerinterventions.

Second, proteges in formal relationships may perceivethat their mentors spend time with them because of a com-mitment to the mentoring program and the organization,rather than because of personal commitment to the protege,or because the mentor believes in the protege's potential.Both parties recognize that the relationship is short-term andthat the mentor may be assigned to another protege after therelationship is over. These factors may restrict the develop-ment of trust and emotional closeness in the relationship andthe provision of psychosocial functions.

Third, formal relationships are contracted to focus oncareer goals that are short-term and relate to the protege's

current position (Geiger-DuMond & Boyle, 1995; Gray,1988; Murray, 1991). In contrast, informal mentors areconcerned with the long-term career needs of their proteges;in some cases the protege's needs may take precedence overthe needs of the organization, and the mentor may recom-mend that the protege move to another organization. Infor-mal mentoring relationships are therefore more aligned withthe view that careers no longer unfold within a singleorganization but instead become "boundaryless" in span-ning many different organizational settings (cf. Arthur &Rousseau, 1996; Hall & Mirvis, 1996).

Processes in relationship. There are four additionalprocesses that may lead to different functions in formal andinformal mentoring relationships. First, formal mentors maybe less motivated to be in the relationship than informalmentors (Ragins & Cotton, 1991). As discussed earlier,formal mentors may not identify with their proteges. Inaddition, formal mentors may enter the relationship to begood organizational citizens rather than because of theirown developmental needs. Although formal mentors mayreceive more organizational recognition than informal men-tors, they may not be ready to be mentors, and they may beless likely to receive the internal rewards associated withmentoring (cf. Ragins & Scandura, in press). It is thereforereasonable to expect that formal mentors may not be asmotivated as informal mentors to provide career develop-ment and psychosocial functions.

Second, formal mentors may have less effective commu-nication and coaching skills than informal mentors (Kram,1985b, 1986). Proteges select informal mentors with strongcommunication and coaching skills (Kalbfleisch & Davies,1993; Olian et al., 1988). Formal mentors may be viewed ashaving good communication skills by program coordina-tors, but if the protege does not share these perceptions, therelationship may become strained and less effective.

A third factor that may limit the effectiveness of formalmentoring relationships is that many programs match mem-bers from different departments or functional units in theattempt to avoid charges of favoritism (Douglas, 1997;Murray, 1991). This practice may impede the formal men-tor's ability to intervene on the protege's behalf and provideexposure, protection, sponsorship, and challenging assign-ment functions. Moreover, informal relationships oftenevolve on the basis of mutual interests, job functions, andcareer paths. Formal mentors who are in different depart-ments or functional areas than their proteges may also havedifferent career paths and may therefore provide less effec-tive career counseling and role modeling for their proteges.

Finally, because formal mentors are more visible thaninformal mentors, they may be more self-conscious aboutengaging in career development behaviors that may beconstrued as favoritism by others in the organization. Infor-mal mentors generally engage in such behaviors with im-punity; informal mentors have been found to sponsor their


proteges into upwardly mobile positions, give them chal-lenging "stretch" assignments, and buffer them from ad-verse forces in the organization (Kram, 1983; Scandura,1992). Because formal mentoring relationships are publicrelationships that are monitored by program coordinators,formal mentors may be less likely than informal mentors tointervene on their protege's behalf.

Review of Research

As the above theoretical review indicates, formal mentorscan be expected to provide less of each of the nine careerdevelopment and psychosocial functions than informalmentors. Only two studies directly investigated the relation-ship between type of mentor and mentor functions. On thebasis of existing theory, both studies proposed greater ben-efits for informal than formal mentoring relationships. Al-though the results of these studies were complementary withexisting theory, only partial support for theoretical predic-tions was found. In a study of 212 informal and 53 formalproteges, Chao, Walz, and Gardner (1992) found that pro-teges in informal mentoring relationships reported morecareer development functions and higher salaries than pro-teges in formal relationships. However, no support wasfound for their hypothesis that proteges in informal rela-tionships would report more psychosocial functions thanproteges in formal relationships. Fagenson-Eland, Marks,and Amendola (1997) found that informal proteges reportedmore psychosocial benefits than formal proteges in theirstudy of 16 informal and 30 formal proteges employed attwo merging organizations. However, informal and formalproteges did not significantly differ in reports of careerdevelopment and role modeling functions.

These studies, although groundbreaking, used only thebroad categories of mentoring functions and did not providean in-depth investigation of the nine specific mentoringfunctions theorized by Kram (1985a). Chao et al, (1992)used Noe's (1988) 21-item mentor functions instrument,which measures Kram's (1985a) two broad categories ofcareer development and psychosocial functions. The careerdevelopment scale consisted of 7 items that measure theprotection, challenging assignment, and exposure functions,but do not include Kram's coaching and sponsorship func-tions. Psychosocial functions were measured using a 14-item scale that omitted the friendship function and includedthe coaching function. The coaching function is a careerdevelopment function (Kram, 1985a) but is loaded on thepsychosocial factor in Noe's scale (Noe, 1988). The instru-ment is conceptually limited in that it does not allow for ananalysis of the nine individual mentor functions becausesingle items are used to measure many functions, and somefunctions are not represented in the instrument. Fagenson-Eland and her colleagues (1997) used the Scandura andKaterburg (1988) 18-item mentoring functions question-

naire, which collapsed Kram's (1985a) nine mentor func-tions into three broad categories: career development, psy-chosocial, and role modeling. However, like Noe's (1988)instrument, this instrument assesses only the broad catego-ries of mentor functions and therefore does not allow for afull assessment of Kram's mentor role theory by investigat-ing the effects of each of the nine individual mentorfunctions.

One objective of the present study, therefore, was toextend this prior research by providing a more fine-grainedanalysis of the relationship between type of mentor and eachof Kram's (1985a) mentor roles. Toward this aim, we usedthe Ragins and McFarlin (1990) 33-item mentor role instru-ment, which allows for a separate analysis of the effects ofeach of Kram's (1985a) nine mentor functions, as well astwo additional functions.

Another objective of our study was to assess the relation-ship between formal and informal mentoring and careeroutcomes, such as promotion rate and compensation. Therehas been a lack of research investigating these relationships.As discussed earlier, Chao et al. (1992) explored the rela-tionship between the type of mentoring relationship andcompensation. She reported that individuals with informalmentors had greater compensation than individuals withformal mentors, but she did not investigate the relationshipbetween type of mentor and promotion rate. We would liketo replicate and extend her study by investigating the rela-tionship between formal and informal mentoring, compen-sation, and promotion.

Our study also attempts to follow Chao et al.'s (1992)lead in recognizing that the effects of mentoring on careeroutcomes may not be immediate, but may take place overtime. This is particularly relevant when investigating formalmentoring relationships. As discussed earlier, because for-mal mentoring relationships usually last a year or less(Murray, 1991; Zey, 1985), static measures, although suit-able for capturing current perceptions of the mentor's be-haviors or functions, fall short of capturing the career out-comes of the relationship; the effects of a formal mentoringrelationship may not be realized for 1 or 2 years followingthe relationship's termination. Chao and her colleagues(1992) recognized this issue and wisely included duration ofmentoring relationship as a control variable in their study.They also tested for differences between proteges withcurrent or recent mentors and proteges who had relation-ships that ended 2 years prior to being surveyed. They foundno differences between these groups, but they made thevaluable point that it is important to examine historicaleffects and control for the duration of the relationship wheninvestigating the relationship between type of mentoringand career outcomes. Toward that end, we controlled for theduration of mentoring relationship and included retrospec-tive measures of the history of mentoring relationships inour analyses.


Hypotheses of Formal and Informal MentoringRelationships

On the basis of the theory and research reviewed above,we hypothesized that proteges in informal mentoring rela-tionships would report that their mentors provide morecareer development and psychosocial functions than pro-teges in formal relationships.

Hypothesis 1: Proteges in informal mentoring relationshipswill report that their mentors provide more career develop-ment functions (sponsorship, coaching, protection, challeng-ing assignments, exposure) than proteges in formal mentoringrelationships.

Hypothesis 2: Proteges in informal mentoring relationshipswill report that their mentors provide more psychosocialfunctions (acceptance and confirmation, counseling, friend-ship, and role modeling) than proteges in formal mentoringrelationships.

We also expected that informal proteges would reportmore overall satisfaction with the performance of theirmentor than proteges in formal relationships.

Hypothesis 3: Proteges in informal mentoring relationshipswill report greater satisfaction with their mentors than pro-teges in formal mentoring relationships.

We also expected a positive relationship between thehistory of prior mentoring relationships and career out-comes. In particular, proteges who had primarily informalrelationships in the past should report more compensationand a higher promotion rate than proteges with a history ofprimarily formal relationships.

Hypothesis 4: Proteges with a history of informal mentoringrelationships will receive more compensation and promotionsthan proteges with a history of formal relationships.

The next section explores the conjoint effects of thegender composition of the relationship and the type ofmentor on mentor functions and career outcomes.

Gender Composition of Relationship

Existing mentorship theory holds that the gender compo-sition of the mentoring relationship is a critical factor af-fecting mentoring functions and outcomes (cf. Ragins,1997a). Ragins (1997a, 1997b) observed that social identitytheory (Tajfel, 1978), the similarity-attraction paradigm(Byrne, 1971), and the relational demography perspective(Tsui, Egan, & O'Reilly, 1992; Tsui & O'Reilly, 1989) allpredict more perceived similarity, identification and role-modeling in same-gender as compared with cross-genderrelationships. Carrying this logic to the mentoring arena,Ragins (1997a) predicted that psychosocial functions,which incorporate similarity, identification, and role mod-

eling, should also be stronger in same-gender as comparedwith cross-gender mentoring relationships.

Hypothesis 5: Proteges in same-gender relationships will re-port more psychosocial functions (acceptance and confirma-tion, counseling, friendship, and role modeling) than protegesin cross-gender mentoring relationships.

Ragins (1997a) also proposed that mentors' power in theorganization influences their ability to provide their pro-teges with such career development functions as sponsoringtheir proteges to high-ranking positions, protecting themfrom adverse forces, and giving them needed exposure. Sheproposed that because majority mentors (i.e., male mentors)generally have more power in organizations than minoritymentors (i.e., female mentors), they should be better able toprovide career development functions and organizationaloutcomes. This proposition was partially supported in arecent study by Dreher and Cox (1996), which found thatproteges with male mentors received greater compensationthan proteges with female mentors. However, their studydid not investigate career development functions or promo-tion rates, or the impact of the history of mentoring rela-tionships on compensation and promotion. We thereforewanted to build on their study and test existing theory byproposing the following hypotheses:

Hypothesis 6a: Proteges with male mentors will report more-career development functions (sponsorship, coaching, protec-tion, challenging assignments, exposure) than proteges withfemale mentors.

Hypothesis 6b: Proteges with a history of male mentors willreport more compensation and promotions than proteges witha history of female mentors.

Expanding on this framework, Ragins (1997a) proposedthat compared with other gender combinations, male pro-teges with male mentors should receive the most benefitsfrom the mentoring relationship; these proteges should bethe most satisfied with their relationships and should receivemore psychosocial and career development benefits fromtheir relationships than any other gender combination. Cor-respondingly, she proposed that relationships involving mi-nority mentors (i.e., female mentors) and majority proteges(i.e., male proteges) should be the most limited in providingmentoring functions; they are limited in providing careerdevelopment functions because of the female mentor's rel-ative lack of power, and they are limited in providingpsychosocial functions because the relationship is cross-gender. Existing theory (Kram, 1985a) and research (Chaoet al., 1992) indicates that psychosocial and career devel-opment mentor functions are inter-related and may syner-gistically build on one another, thereby making the differ-ences between these dyads even more salient (Ragins,1997a).


This theoretical perspective was tested by the followinghypotheses:

Hypothesis 7a: Male proteges with male mentors will reportmore career development and psychosocial functions andmore satisfaction with their mentors than any other gendercombination of mentoring relationship.

Hypothesis 7b: Male proteges with female mentors will reportless career development and psychosocial functions and lesssatisfaction with their mentors than any other gender combi-nation of mentoring relationship.

We also expected that male proteges who had primarilymale mentors in the past would receive greater career out-comes than any other gender combination. Similarly, maleproteges with a history of primarily female mentors shouldreport less career outcomes than any other gender combi-nation. These expectations are based on the theory discussedabove (Ragins, 1997a), as well as on the finding of positiverelationships between compensation and psychosocial andcareer development mentoring functions (Chao et al., 1992;Scandura, 1992).

Hypothesis 8a: Male proteges with a history of male mentorswill report greater compensation and promotion than anyother gender combination.

Hypothesis 8b: Male proteges with a history of female men-tors will report less compensation and promotion than anyother gender combination.

Finally, we wanted to explore whether the gender com-position of the relationship interacts with the type of rela-tionship in influencing the mentoring relationship. Becausethere was no theory or research to provide direction, wetested the following research question: Does the gendercomposition of the relationship moderate the relationshipbetween type of mentoring (formal vs. informal) and reportsof mentor functions and satisfaction with the relationship?

To date, only one study investigated the impact of gendercomposition of mentoring relationships on mentor func-tions, and there has been no research investigating therelationship between the historical composition of the rela-tionship and career outcomes. In a survey of 181 proteges ininformal mentoring relationships, Ragins and McFarlin(1990) found that same-gender proteges reported engagingin more social activities with their mentors than cross-gender proteges and that female proteges with female men-tors reported more role modeling than any other gendercombination. However, one limitation of this study was thata restricted sample of female mentors prevented a compar-ative analysis of all four gender combinations. Because thissituation is relatively common in male-dominated organi-zations and occupations (Noe, 1988; Ragins, 1989), weobtained a sample of individuals from male-typed, female-typed, and gender-integrated occupations to obtain adequate

sample sizes for each of the four gender combinations ofmentoring relationships.

In a related study, Ensher and Murphy (1997) examinedthe effects of race similarity among 104 minority studentswho were assigned to formal, same-gender mentoring rela-tionships that lasted for 8 weeks. The researchers found thatminority proteges assigned to same-race mentors reportedmore career development functions than minority protegesassigned to different race mentors but found no support fortheir prediction that psychosocial support would be greateramong same-race mentors or that female mentors wouldprovide more psychosocial support than male mentors.However, because proteges were assigned to same-gendermentors, the researchers were unable to examine the effectof the gender composition of the relationship on mentoringfunctions.

In this study, we examined the effects of gender compo-sition and the type of mentoring relationship on mentoringfunctions and outcomes. We extended prior research andtested new theory by measuring specific mentor roles andby investigating the impact of history of mentoring relation-ships on career outcomes.

Method

Procedure and Respondents

Sampling procedure. One goal of our study was to investigatethe effects of the gender composition of the mentoring relationshipon mentor functions and outcomes. Because male-dominated oc-cupations have a shortage of higher-ranking, female mentors (Ra-gins, 1989), and this study called for a sample of male protegeswith female mentors, we obtained a sample of men and women inmale-typed, female-typed, and gender-integrated occupations.Three occupations were selected on the basis of labor statistics andresearch on occupational gender-typing (Beggs & Doolittle, 1993;U.S. Department of Commerce, 1996): engineering (male-dominated), social work (female-dominated), and journalism(gender-integrated). Formal mentoring programs are used in allthree of these occupations (Paine, 1986; Smith, Chase, & Byrd,1986; Taibbi, 1983).

We obtained a national random sample by using mailing lists ofnational professional associations representing these three occupa-tions. To obtain a gender-balanced sample, equal numbers of maleand female names were randomly selected from each mailing list.A total of 3,000 surveys were mailed; 1,000 (500 to men, 500 towomen) were sent to each of the three occupations. Follow-upsurveys and reminder letters were sent according to a modifiedversion of the Dillman mail survey method (Dillman, 1978). Atotal of 1,258 surveys were returned, for a response rate of 42%.Self-employed and retired employees were excluded from analy-ses. Relatively complete data for analyses were available for 1,162respondents.

Respondents. Respondents consisted of 654 women and 500men; 8 respondents did not report their gender. The occupationalbreakdown of the respondents consisted of 362 journalists(31.2%), 414 social workers (35.6%), and 386 engineers (33.2%)..


We used an established definition of mentor (Ragins, 1989) inour survey: "A mentor is generally defined as a higher ranking,influential individual in your work environment who has advancedexperience and knowledge and is committed to providing upwardmobility and support to your career. Your mentor may or may notbe in your organization and s/he may or may not be your imme-diate supervisor." Formal mentoring was defined as follows: "Inorder to assist individuals in their development and advancement,some organizations have established formal mentoring programs,where proteges and mentors are linked in some way. This may beaccomplished by assigning mentors or by just providing formalopportunities aimed at developing the relationship. To recap: For-mal mentoring relationships are developed with organizationalassistance. Informal mentoring relationships are developed spon-taneously, without organizational assistance." To ensure that re-spondents had a clear understanding of the distinction betweenformal and informal mentors, we asked respondents to describetheir formal mentoring program immediately following the ques-tion asking them to identify whether their mentor was formal orinformal. We also asked the respondents to describe the history oftheir mentoring relationships over the last 10 years, starting withtheir most recent mentor. For each relationship, we asked therespondents to give information on the mentor's gender, the typeof relationship (formal/informal), the mentor's position (supervi-sory/nonsupervisory), and the duration of the relationship.

On the basis of these definitions, 510 respondents (43.9%)reported having an informal mentor, 104 respondents (9%) re-ported having a formal mentor, and 548 (47.2%) did not have amentor. The final sample was composed of these 614 proteges,which consisted of 352 female proteges, 257 male proteges, and 5who did not report their gender. Although men and women did notsignificantly differ in having a mentor, ) f ( l , N = 1,135) = 0.36,ns, men were nearly twice as likely to be in a formal mentoringrelationship: 22% of the men reported currently being in a formalrelationship, compared with 12% of the women in the sample,/(I, N = 608) = 10.39, p < .001. Men were also significantlymore likely than women to have a history of formal mentoringrelationships: 18.5% of the men reported having primarily formalmentors in their past, compared with 9.7% of the women, ^(l,N = 579) = 9.37, p < .01. One implication of these genderdifferences is that if the type of mentoring relationship does affectmentoring functions and outcomes as proposed, the type of men-toring relationship should be included as a control variable whentesting the gender composition hypotheses.

The sample involved 348 (57.1%) individuals in same-genderrelationships and 261 (42.9%) in cross-gender relationships. Amore fine-grained breakdown revealed 233 male proteges withmale mentors, 115 female proteges with female mentors, 24 maleproteges with female mentors, and 237 female proteges with malementors. The median length of the current mentoring relationshipwas 4.0 years, and the average length was 6.7 years. The currentmentoring relationship for male proteges was significantly longerthan for female proteges, r(426) = —2.77, p < .01, indicating thatthis variable should also be used as a control variable in theanalysis. Fifty-three percent of the proteges reported that theirmentors were also their supervisors, and there were no significantgender differences on this variable, x*(l,N = 611) = 1.06, ns. Theaverage age of the proteges was 46 years, and 92% were Cauca-

sian. Most were married (70%) and had completed (63%) orpursued (12%) graduate degrees.

Measures

History of mentoring relationships. We asked our respondentsto give information about each mentoring relationship they hadover the last 10 years, as well as their current mentoring relation-ship. They were asked to report whether their prior relationshipswere formal or informal, the gender of their mentors, their men-tors' positions, and the duration of the relationships.

The history of prior relationship type was measured by dividingthe number of prior relationships that involved formal mentors bythe total number of prior formal and informal mentoring relation-ships. Higher values therefore reflect a greater proportion of for-mal than informal mentoring relationships. The history of priorgender of mentors was measured by dividing the number of priorrelationships involving male mentors by the total number of rela-tionships involving both male mentors and female mentors. Highervalues thus reflect a greater proportion of male than female men-tors. The history of relationships involving specific gender com-positions was computed in a similar manner.

Mentor functions. The Mentor Role Instrument (MRI) wasused to measure mentor functions (Ragins & McFarlin, 1990). The33-item instrument was developed via confirmatory factor analy-sis, and independently measures each of Kram's (1985a) ninementor roles. The instrument also assesses two additionalpsychosocial-related roles: parent and social interactions. Accord-ing to Kram (1985a), these roles may emerge in response to genderissues in mentoring relationships; proteges may seek to avoidsexual issues in cross-gender relationships by viewing their mentoras a parent figure or by avoiding informal, after-work socialinteractions.

The MRI has proven reliability and preliminary evidence ofvalidity (Ragins & McFarlin, 1990). The 33-item instrument has 3items per mentor role and was measured on a 7-point Likert scalewith responses ranging from 1 (strongly disagree) to 7 (stronglyagree). The instrument may be found in the Appendix. The coef-ficient alphas for the eleven mentor roles ranged from .63 to .91and are listed on the diagonal of the correlation matrix in Table 1.

Mentor satisfaction. The protege's satisfaction with the men-tor was measured by a 4-item scale. The items, which are listed inthe Appendix, used the same 7-point Likert scale used in the MRI.The coefficient alpha for the satisfaction scale was .83.

Career outcomes. Promotion rate and compensation wereused to measure mentor outcomes. Using established definitions ofpromotions (cf. Whitely et al., 1991), promotions were defined asinvolving two or more of the following criteria: significant in-creases in annual salary, significant increases in scope of respon-sibility, changes in job level or rank, or becoming eligible forbonuses, incentives or stock plans. Given the above definition,respondents were asked how many promotions they received overthe last 10 years. Respondents were also asked to report theircurrent annual compensation, which included salary, bonuses,commissions, stock options, and profit sharing.

Control variables. Six variables that may be related to promo-tion rate and compensation were considered as potential covari-ates: organizational rank, organizational tenure, position tenure,number of career interruptions, education, and occupation (engi-


tfj

3"i"

^.0

-S%>

£y-SJ•5

"z*a„

a-c-S-

s:•2aS

Q"P

IC

i§

NOCM

CM

CM

C->CM

CM

CM

OCM

ON

oo

r-

NO

'"'

S

cr>

~~*

-

O^^

ON

oo

r-

SO

m

^

en

CM

— '

Q

^

Var

iabl

e

O O O —> <- ' C\

m m oJ \o — oo o o o — ' c

\ \ I

ooo -« c\ \

CO —i — -3- ON

o o o — • o c

—> o o —> <-< c!

O rn — -i p-.. -rj- \j

— 00 — — C\

r- o <N oo fi TO — O O O

\ \ \

^ ___ ^- ^H

o o o o o r\ \ \ \

— ON fi NO fM yO O O O O

1 ! 1

cs t — r~- so ^^ ^o o o o o c

\ \

— \o — i — ' -*t ro o o o o

\

ooo o o c1 \

O CM O — i O C1 1 1

c*"i oo in c — r- -^- H O — o o c

1

CM O ON CM NO t;0 <N 0 — — CI \

u~i ON o cM ^- r— O O O O C

I I I I

— < o o o o c1 I

S — c ~-c — c t i r— o — o c

1

^- O — -i ON in ••;O — • 0 O O C

1 I

— i — i oo -^f- CM r:O O O CN — v

!

o o o iS S1 1 1

oo r- CM ON

--0 * |

ON ON CM

T ~ T1 1 '

m o

°7 1m7 1i '

oo m oo \o -*f wO •* O \O CM C

CO — '

o oo •* r- "* v

CO -<

obg^ Cl. •""'>"»

O -^- ex O• w .™ oo

3 5 « C a>

gill IIIlll lllS 1 u 5 1 1 d I-CMm^J- «DNO

•> r- NO m CM r

1 I

rj ol -H _i -^ f.3 O O O O C

I |

1 CO CS CM Tj- -3 O — — O C

1 1 1 i

S r*"> cO t — CO O O C

1 1

3 •<*• O ON NO C1 O — O O -

1 ! 1

1 1 7

NO rn -H ON r-

1

H -H oo r^ " r*^ -H o O O C

1

D CO O — < Tt f- O — — O C

I I

£ S - o5i1 1 1

•a NO oo r- NO v0 -, _ , o C

I 1

H \o — i CM CM c*5 O — — ' O C

i i

i tn " in r*i c*3 O O O O C

1

S S - 22?1

o en — r- -5 _ _ — [ ( M e *

•a m CM •<*• — '3 O O O CO

i 1

t —> CS -"3-3 O O O

1 1

•1 — CO2 CM OO

oo2 — < 1

t1 1

5 — | O O O C

1 ON r- ON. r- v•} -H -^ o r- c-> -H o O O C

u

II1 o ^ S-°'E s. a. ° a g =u 5*".c:«> IS £ c

i 1 1 3 1 11 5 1r- oo ON o —

S o o o g - S ^ S ^ f t S S o S m S

" '

- C - H — i n c M m c M N O - — i — " O < n c M o os o o o ^ o c M c n — mcMNomcMr^

1 " '

'—'

3 r - c M m o o o c M O i n o o N O o o

- g3gooo^^oo§

^>_HC4 r-.oor-,oooNooinv

? O - H — (Mf-- inNo>nco

T ^ ^ c ^ ^ - ^ ^ o C T3 O O - ^ m < n T f T r o N

^ o o o — <r-r i -oo_1 ""

3 o o — « co r r-^

^2 O — i — ' CM CO

"""'^_^

3 O O O 00

§ -*-H 1

ss

1

rru- iocNcNr-nm — m m ^ o o m c M

Ow

^ ^ ^ S ^ S m S ^ ^ ^ ^ S S ^

m

O

1j ^

JljL.llf,,!!,£5lll-<3lsl'll(£lll2222SE :22S?3SSSSS

a u3c o*- tT

.t:0 "

T3 ^

^ -uJ3 Cfa— *§)

S o

.1 "i"•§1x. s s""* .=: ^^- o.c co —

11£ 5ex g•a —

^ 0

t — i -0 0" C!! S= 'I

•I a.

l|

S S

^ 1«n ^

" c«[I <L>

11 —

*""' >

l.-f

.S>-c

^ s 1 is.,1 S"u c3 --^x "o E5u a gm o S

•" £ 1S ^ _ur- 2 "«S 1 s

6 S oO •— c

"o S "•£QJ O O

§ * Sw» " °"£ O W

S g |"o-n

M

*- £ 5J "2 §5 S E.S o H87 s0 5c >

= -i -^"«5 .1 "-S

fc ™ °

£ 1 jj 52 c t3 2

i^ ^ i1 1 § j -i.s ! v'3 -a o ft,aj <u 'C

2 l"| A

i 1 £ £


neering, social work, and journalism). We also considered sixother mentoring variables that have been found to be related tomentor roles and outcomes in other studies (cf. Burke & McKeen,1997; Chao et al., 1992; Ragins & McFarlin, 1990; Whitely et al.,1991). These variables included whether the mentor is the pro-tege's supervisor, the length of the mentoring relationship, theaverage number of hours spent in the relationship per month, thenumber of prior mentoring relationships, the protege's age, and theprotege's socioeconomic background.

To preserve power, covariates were selected that had significantcorrelations with dependent variables but low intercorrelations(Neter & Wasserman, 1974). Tests were made of all controlvariables by independent interaction terms to test homogeneity ofregression assumptions fundamental to covariance analyses. Onthe basis of these criteria, the protege's occupation (engineering,social work, journalism), the duration of their current mentoringrelationship (converted to months), and whether their current men-tor was their supervisor were selected as covariates for analysesthat involved reports of the current mentor's functions and satis-faction with the current mentor. For analyses assessing the rela-tionship between the prior history of mentoring relationships andcareer outcomes, covariates included occupation, position tenure,the number of career interruptions, the duration of prior mentoringrelationships, and the number of those relationships involvingsupervisory mentors.

Results

Analyses

The correlations, means, standard deviations, and coeffi-cient alphas for the study variables are displayed in Table 1.Hierarchical multiple regression analyses and a prioriplanned group contrasts were used to test the study'shypotheses.

Separate hierarchical regression analyses were conductedfor each of the dependent variables. For hierarchical anal-yses involving the dependent variables of mentoring func-tions and satisfaction with the mentor, independent vari-ables were entered in the following order: (a) the controlvariables; (b) type of relationship, that is, whether the cur-rent relationship was formal or informal (Hypotheses 1, 2,and 3); (c) the current mentor's gender (Hypothesis 6a) andthe protege's gender; (d) the two-way interactions involvingmentor gender, protege gender, and the type of relationship;and (e) the three-way interaction involving mentor gender,protege gender, and the type of relationship (ResearchQuestion). Hypotheses that assessed the relationship be-tween the history of the mentoring relationship and careeroutcomes (promotion and compensation) were tested in athree-step hierarchical analysis in which the control vari-ables were entered in the first step, the history of the type ofmentoring relationship in the second step (Hypothesis 4),and the history of the gender composition of prior relation-ships in the third step (Hypothesis 6b).

Duncan's Multiple Range tests were used to test hypoth-

esized differences between the adjusted group means of thefour gender combinations of mentoring relationships (Hy-pothesis 5, Hypotheses 7a & 7b, Hypotheses 8a & 8b). Weused hierarchical regression analyses to obtain betas for thefour covariates and the type of mentoring relationship (for-mal vs. informal); we then used these betas to compute theadjusted means.

Mentoring Control Variables

Both of the mentoring control variables were signifi-cantly related to the dependent variables. Mentors who weresupervisors provided more career development functionsthan nonsupervisory mentors, but no significant differenceswere found for psychosocial functions. The length of thementoring relationship was positively related to compensa-tion and psychosocial functions but was unrelated to careerdevelopment functions.

Comparisons of Informal and Formal MentoringRelationships

As displayed in Table 2, the significant negative beta forthe type of relationship term indicates full support for Hy-potheses 1, which predicted that proteges with informalmentors would report more career development functions(sponsoring, coaching, protection, challenging assignments,and exposure) than proteges with formal mentors. Hypoth-eses 2 predicted that proteges with informal mentors wouldreport more psychosocial functions than proteges with for-mal mentors. As shown in Table 3, this hypothesis receivedsupport for four of the six psychosocial functions. Protegeswith informal mentors reported that their mentors providedmore psychosocial functions involving friendship, socialsupport, role modeling, and acceptance than proteges withformal mentors. However, no significant differences werefound on reports of the parent and counseling functions. Fullsupport was also received for our prediction (Hypothesis 3)that proteges with informal mentors would report greatersatisfaction with their mentors than proteges in formalrelationships.

Partial support was received for Hypothesis 4. In supportof our prediction, the hierarchical regression analyses re-vealed that individuals with a history of informal mentorshad significantly greater compensation (M = $56,629) thanindividuals with a history of formal mentors (M =$48,107). Although those with a history of informal men-tors also had more promotions over the last 10 years(M = 2.55) than those with a history of formal mentors(M = 2.04), the differences were not significant whencontrolling for differences in position tenure, number ofcareer interruptions, occupation, supervisory status of men-tors, and length of mentoring relationships over the last 10years. The results of these analyses are shown in Table 4.


Table 2Summary of Hierarchical Regression Analysis for Variables Predicting Career DevelopmentMentoring Functions (Hypothesis 1, Hypothesis 6a, and Research Question)

Mentor role and predictor

Sponsor (n = 567)Step 1: Control variables

Supervisory mentorRelationship lengthOccupation

Step 2RT

Step 3MGPG

Step 4MG X PGRT X MGRT X PG

Step 5MG X PG X RT

Coach (n = 567)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

Protect (n = 564)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

Challenge (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

B

1.070.02

-0.03

-2.05

0.520.02

-0.07-0.58

1.29

3.33

-0.37-0.04

0.06

-1.15

-0.03-0.13

2.71-0.53

2.95

4.18

1.31-0.02-0.12

-1.28

0.43-0.10

-0.480.051.07

-1.36

1.800.000.32

-1.05

0.39-0.42

0.40-0.23

2.16

6.52

SE

0.360.030.23

0.47

0.440.37

1.511.231.22

3.23

0.320.020.21

0.42

0.390.33

1.341.091.08

2.85

0.360.030.23

0.48

0.450.38

1.541.251.24

3.29

0.360.030.23

0.48

0.450.38

1.551.251.25

3.28

0

.12**

.03-.01

__!g***

.05

.00

-.01-.04

.07

.16

-.05-.07

.01

-.12**

-.00-.02

.31*-.05

.-- .19**

.24

.15***-.04-.02

-.11**

.04-.01

-.05.004.06

-.007

.20***

.003

.05

-.09*

.04-.05

.04-.02

.12

.32*

Afl2

.03***

.002

.00

.002

.01**

.00

.02**

.003

.01**

.001

.001

.000

.01*

.002

.005

.006*

R2

.01*

.05***

.05***

.05***

.05***

.006

.02*

.02

.04**

.04**

.03**

.04***

.04***

.04**

.04**

.05***

.06***

.06***

.06***

.07***


Table 2 (continued)

Mentor role and predictor SE

Exposure (n = 567)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

1.04-0.03

0.26

-1.60

0.79-0.05

-0.0230.620.01

1.43

0.350.030.22

0.45

0.420.36

1.461.181.18

3.11

.13**-.05

.05

-.15***

.08-.01

-.002.05.00

.07

.02**

.02*** 04***

.006 .05***

.001 .05***

.000 .05***

Note. Dummy coding of predictors: supervisory mentor: 1 = supervisory, 0 = not supervisor. Relationshiptype (RT): 1 = formal mentor, 0 = informal mentor. Mentor gender (MG): 1 = male, 0 = female. Protegegender (PG): 1 = male, 0 = female.*p£.05. **p<.0\. ***p<.00l.

Effect of Gender on Mentoring Relationships

As discussed earlier, because men were significantlymore likely to have formal mentors than women, and thetype of mentoring relationship was found to be related tomentoring and career outcomes, the type of mentoring re-lationship needed to be controlled for when investigatinggender effects in mentoring relationships. Therefore, type ofmentoring relationship was entered hierarchically beforegender terms in the regression analyses, and the groupmeans were adjusted for type of mentor in the plannedgroup contrasts.

Effect of same-gender relationships. We predicted (Hy-pothesis 5) that individuals in same-gender mentoring rela-tionships would report more psychosocial functions thanindividuals in cross-gender relationships. We tested thishypothesis by using the Duncan's Multiple Range test. Theadjusted group means for all gender combinations are dis-played in Table 5. Although the means generally were in thedirection predicted, Hypothesis 5 was not supported. How-ever, there was evidence that the gender composition of therelationship affected two psychosocial functions. Specifi-cally, female proteges with female mentors were signifi-cantly more likely to report engaging in social activitieswith their mentors than female proteges with male mentors.Male proteges with female mentors were significantly lesslikely than all other gender combinations to report that theirmentor provided acceptance roles.

Effect of male mentors. As displayed in Table 2, nosupport was found for Hypothesis 6a, which held that malementors would be associated with more career developmentfunctions than female mentors. Partial support was received

for Hypothesis 6b, which held that proteges with a history ofmale mentors would report more compensation and promo-tions than proteges with a history of female mentors. Whencontrolling for differences in position tenure, number ofcareer interruptions, occupation, length of mentoring rela-tionships, supervisory status of mentors, and the type ofmentor (formal vs. informal), the hierarchical regressionanalyses revealed that proteges with a history of male men-tors received significantly greater compensation (M =$60,140) than proteges with a history of female mentors(M = $41,354). Proteges with a history of male mentorsalso reported more promotions over the last 10 years(M = 2.6) than those with a history of female mentors(M = 2.3), but these differences were not statisticallysignificant. The results of these analyses are shown inTable 4.

Effect of other gender combinations. We used the Dun-can's Multiple Range test to test Hypotheses 7a and 7b, andthe results of these analyses are shown in Table 5. Contraryto our predictions (Hypothesis 7a), male proteges with malementors did not report more mentoring functions or satis-faction with their mentors than any other gender combina-tion. However, some support was found for Hypothesis 7b,which predicted that male proteges with female mentorswould report fewer functions and less satisfaction than anyother gender combination. The general pattern of meansrevealed that male proteges with female mentors reportedless psychosocial and career development functions than theother gender combinations, but these effects were onlysignificant for three of the mentor roles and the mentorsatisfaction variable. Compared with other gender combi-nations, male proteges with female mentors were less sat-


Table 3

Summary of Hierarchical Regression Analysis for Variables Predicting Mentor Satisfaction and

Psychosocial Mentoring Functions (Hypothesis 2, Hypothesis 3, and Research Question)


Mentor satisfaction (n = 568)Step 1: Control variables


Step 2RT

StepSMGPG


Step 5MG X PG X RT

Friendship (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

Social (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

Parent (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

B

0.010.03

-0.14

-1.25

0.23-0.004

2.70-1.47

1.84

0.86

-0.190.08

-0.28

-1.38

-0.22-0.13

2.65-1.74

2.90

-1.90

-0.390.12

-0.66

-1.22

-0.980.95

4.65-0.36

3.38

1.65

0.120.06

-0.23

-0.15

0.390.62

0.63-0.72

1.61

-3.37

SE

0.360.030.23

0.48

0.450.38

1.541.251.24

3.28

0.280.020.18

0.37

0.340.29

1.180.960.95

2.52

0.460.030.29

0.61

0.570.49

1.941.571.56

4.14

0.260.020.16

0.35

0.320.28

1.110.900.89

2.36

|3 AS2

.001

.05-.03

.01**-.11**

.00.02.00

.008.27

-.11.10

.000.04

-.03.15***

-.07.02***

-.15***.001

-.03-.02

.02**.34*

-.17.21**

.000-.12

-.03.14***

-.09*.01*

-.08*.01

-.07.08*

.02*.37*

-.02.15*

.00.06

.02

.13**-.06

.00-.02

.01*.05.10*

.00.08

-.08.12

.00-.23

R2

.004

.02*

.02

.02

.02

.03***

.06***

.06***

.08***

.08***

04***

.04***

.05***

.07***

>07***

.02**

.02*

.04***

.04**

.04**


Table 3 (continued)


Role model (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

Counseling (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

Acceptance (n = 568)Step 1: Control variables


Step 2RT

Step 3MGPG


Step 5MG X PG X RT

B

-0.230.09

-0.14

-1.07

-0.31-0.06

2.55-1.24

2.08

-1.53

0.270.03

-0.11

-0.01

-0.080.08

0.430.343.10

-0.42

-0.100.05

-0.05

-0.88

0.19-0.27

2.37-0.79

0.58

2.09

SE

0.320.020.20

0.42

0.390.34

1.351.101.09

2.89

0.300.020.19

0.40

0.380.32

1.291.051.04

2.76<*-'

0.190.010.11

0.24

0.230.20

0.790.640.64

1.68

J3

-.03.15***

-.03

-.10*

-.03-.01

.29-.11

.13*

-.08

.04

.06-.02

-.00

-.01.01

.05

.03

.21**

-.02

-.03.15***

-.02

_ 1^***

.04-.06

.46**-.12

.06

.20

AS2

.01*

.001

.01*

.004

.000

.000

.017*

.000

.02***

.003

.01**

.002

R2

.03***

.04***

.04***

.05***

.05***

.005

.005

.005

.02

.02

.03**

.05***

.05***

.07***

.07***

Note. Dummy coding of predictors: Supervisory mentor: 1 = supervisory, 0 = not supervisor. Relationshiptype (RT): 1 = formal mentor, 0 = informal mentor. Mentor gender (MG): 1 = male, 0 = female. Protegegender (PG): 1 = male, 0 = female.*/)<.05. **/7<.01. ***/?<.001.

isfied with their mentors and were less likely to report thattheir mentor provided acceptance in their professional de-velopment. Male proteges with female mentors were alsosignificantly less likely to report that their mentors providedchallenging assignments and exposure than female protegeswith male mentors.

Although male proteges with male mentors did not reportmore mentoring functions than other gender combinations,they did report more compensation, as shown in Table 5.

This indicated partial support for Hypothesis 8a. As pre-dicted, male proteges with a history of primarily malementors reported greater compensation than any other gen-der combination. However, male proteges with a history ofmale mentors did not receive more promotions than anyother gender combination. Limited support was found forHypothesis 8b; male proteges with a history of femalementors reported significantly less compensation than maleproteges with male mentors, but predicted differences were


Table 4Summary of Hierarchical Regression Analysis for Variables Predicting Career Outcomes(Hypotheses 4 and 6b)

Predictor and career outcome B SE AR2 R2

Promotions (n = 512)Step 1: Control variables

Position tenureCareer interruptionsOccupationSupervisory mentorsLength of relationships

Step 2: Hypothesis 4Prior relationship type

Step 3: Hypothesis 6bPrior gender of mentors

Compensation (n = 508)Step 1: Control variables

Position tenureCareer interruptionsOccupationSupervisory mentorsLength of relationships

Step 2: Hypothesis 4Prior relationship type

Step 3: Hypothesis 6bPrior gender of mentors

-0.010.050.040.09

-0.01

-0.43

0.09

17.80-5148.51

6044.48219.03211.41

-22552.85

19503.37

0.0010.110.110.070.01

0.50

0.24

17.011723.671722.061169.03

111.82

7523.73

3569.42

-.23***.02.02.05

-.03

-.04

.01

.05-.14**

.16***

.01

.08*

-.13**

^22***

.06***

.001 .07***

.000 .07***

.05***

.01** .06***

Q4*** JJ***

Note. Higher values of prior supervisory mentors indicate greater proportion of supervisory mentors. Highervalues of prior relationship type indicate greater proportion of formal mentors. Higher values of prior gender ofmentors indicate greater proportion of male mentors.*p<.05. **p£.01. ***/><.001.

not found for the other gender combinations. Although maleproteges with a history of female mentors had the lowestpromotion rate of any gender combination, the differencewas only significant when comparing this group with fe-male proteges and male mentors.

An examination of the means in Table 5 reveals otherprovocative findings. Specifically, although female protegeswith a history of primarily male mentors received signifi-cantly more promotions than male proteges with either maleor female mentors, they did not receive more compensation.This suggests that compensation and promotion may beindependent events for some groups. Another interestingresult was that female proteges with a history of malementors reported significantly greater compensation thanfemale proteges with a history of female mentors. Thesewomen also reported significantly greater promotion ratesthan male proteges with a history of male mentors and thosewith a history of female mentors.

Research Question and Other Gender Interactions

The research question was assessed with the three-wayinteraction between mentor gender, protege gender, andtype of mentoring relationship (formal vs. informal). Asignificant three-way interaction was only found for thementoring function of challenging assignments. A plot ofthis interaction revealed that for individuals in cross-gender

relationships, informal mentors provided more challengingassignments than formal mentors, but for individuals withsame-gender relationships, formal mentoring relationshipsprovided more challenging assignments than informalrelationships.

An examination of Tables 2 and 3 reveals that the effectsof formal and informal mentoring relationships varied bythe gender of the protege in 5 of the 11 mentoring functions.The adjusted means for the significant two-way interactionsrevealed a similar pattern for 4 of the 5 mentoring functions.Specifically, the type of mentor influenced reports of men-tor functions for female, but not for male proteges. Com-pared with female proteges with informal mentors, femaleproteges with formal mentors reported less coaching (for-mal M = 13.9, informal M = 15.9), role modeling (formalM = 13.4, informal M = 14.8), friendship (formalM = 14.9, informal M = 16.8), and social interactions(formal M = 6.8, informal M = 9.0). Male proteges, on theother hand, reported equivalent coaching (formal M = 15.0,informal M = 15.2), role modeling (formal M = 14.1,informal M = 14.3), friendship (formal M = 15.8, informalM = 16.1), and social interactions (formal M = 8.8, infor-mal M = 8.9) with formal and informal mentors. Thepattern was somewhat different for the counseling function.Like the other functions, female proteges reported receivingless counseling with formal mentors (M = 14.2) than in-


Table 5Adjusted Means for Study Variables by Gender Combination of Current and Prior Mentoring Relationships(Hypotheses 5, 7a, 7b, 8a, and 8b)

Male proteges, malementors

(n = 208)

Variable

Current mentoring relationshipOverall satisfaction with mentor

SatisfactionCareer development roles

SponsorCoachProtectChallengeExposure

Psychosocial rolesFriendshipSocialParentRole modelCounselingAcceptance

Prior mentoring relationshipCareer outcomes

Compensation (X$l,000)Promotion rate

M

23.6,

14.014.912.515.515.1

16.39.24.3

14.414.617.9a

68.4a.b,2.3b

SD

4.1

4.13.64.14.33.9

3.35.43.23.83.62.2

42.32.2

Female proteges,female mentors

(n = 105)

M

23.6b

13.915.412.415.714.8

16.99-9a3.2

15.114.818.4b

41.7a,d2.5

SD

4.0

4.13.64.34.14.1

3.25.83.33.63.42.1

19.42.1

Male proteges,female mentors

(n = 23)

M

21.4a,b.c

13.513.812.214.0a

13.6a

15.68.34.0

13.614.316.3a.b,c

$46.5b1.3a

SD

6.1

3.94.74.35.24.4

4.94.93.14.94.34.5

13.62.2

Female proteges,male mentors

(n = 220)

M

23.5C

14.415.213.016.1.15.5a

16.78.4a

4.014.614.718.3C

$50.8cd

2.9a,b

SD

4.5

4.23.64.34.14.2

3.45.22.93.83.62.1

23.72.1

Note. Means adjusted for covariates and type of mentoring relationship. Means having the same subscript differ significantly at p < .05.

formal mentors (M = 14.8), but male proteges actuallyreported receiving more counseling from formal (M = 15.5)than informal (M = 14.2) mentors. In short, although pro-teges in formal relationships generally reported less men-toring functions than proteges in informal relationships, thisrelationship was not equivalent for men and women. Al-though the presence of a formal mentor was not associatedwith less mentoring functions for male proteges, formalmentoring was associated with less mentoring functions forfemale proteges.

Significant two-way interactions between mentor genderand protege gender were also found for the coaching, friend-ship, and social roles functions. For coaching, althoughthere were few differences between male and female pro-teges with male mentors (Ms = 15.4 and 15.7, respec-tively), male proteges with female mentors reported signif-icantly less coaching (M = 14.2) than female proteges withfemale mentors (M = 15.8). A similar pattern was foundfor the friendship function: Male and female proteges withmale mentors reported equivalent friendship functions(Ms = 16.2 and 16.5, respectively), but male proteges withfemale mentors reported significantly less friendship func-tions (M = 15.4) than female proteges with female mentors(M = 16.8). A plot of the interaction for social rolesrevealed a somewhat different pattern in that proteges insame-gender mentoring relationships reported more socialinteractions with their mentors than proteges in cross-

gender relationships. The social functions means for womenand men in same-gender relationships were 9.8 and 9.1,respectively, but the means for women and men in cross-gender relationships were 8.2 and 8.1.

Post Hoc Analyses: Comparisons With NonmentoredRespondents

One reasonable question that comes to mind when view-ing these results is whether there were differences betweenthose not mentored and those who had formal or informalmentoring relationships. This question replicates Chao etal.'s (1992) study, but it is also an extension of her studybecause we investigated promotion rates. We used an anal-ysis of covariance (ANCOVA) to compare formal proteges,informal proteges, and nonmentored individuals on the ca-reer outcomes of compensation and promotion. Our resultsreplicated Chao et al.'s (1992) findings. When controllingfor occupation, tenure in position, and number of careerinterruptions, we found that employees with a history ofprimarily informal mentors received significantly morecompensation (M = $56,629) than those without mentors(M = $51,389), F(2, 924) = 6.48, p < .01, but there wereno significant differences in compensation between non-mentored individuals and those with a history of formalmentoring relationships. As discussed earlier, when addingthe additional control variables of length of mentoring re-


lationship and supervisory status of mentors to our analyses,we found support for our hypothesis that proteges with ahistory of informal mentors would receive greater compen-sation (M = $56,629) than proteges with a history of formalmentors (M = $48,107). These results also replicated Chaoet al.'s (1992) findings.

We also extended Chao's work by exploring the effect ofmentoring on promotion rate. Consistent with prior researchthat compared informally mentored and nonmentored in-dividuals (e.g., Dreher & Ash, 1990; Scandura, 1992),an ANCOVA revealed that nonmentored individuals(M = 1.78) had significantly fewer promotions than thosewith a history of informal mentors (M = 2.55), F(2,924) = 9.893, p < .0001. However, nonmentored individ-uals (M = 1.78) did not significantly differ from those witha history of formal mentors (M = 2.04), with respect to thenumber of promotions received. As discussed earlier, indi-viduals with a history of formal and informal mentors didnot significantly differ in the number of promotionsreceived.

In sum, those with a history of formal mentors did notreceive greater career outcomes related to compensation orpromotion than those lacking mentors. Those with a historyof informal mentors received more compensation and pro-motions than nonmentored individuals and more compen-sation than those with formal mentors.

Discussion

The Comparison of Formal and Informal MentoringRelationships

The first objective of our study was to provide a com-prehensive test of mentor role theory (Kram, 1985a) bycomparing the specific mentoring functions received byproteges in formal and informal mentoring relationships andthe career outcomes associated with these relationships. Wefound that proteges with informal mentors received greaterbenefits than proteges with formal mentors. Proteges withinformal mentors reported that their mentors provided morecareer development and psychosocial functions than pro-teges with formal mentors; significant differences favoringinformal mentors were found in 9 of the 11 mentor roles. Inline with these findings, proteges with informal mentorsreported greater overall satisfaction with their mentors thanproteges with formal mentors. Proteges with a history ofinformal mentors also earned significantly more than pro-teges with a history of formal mentors. Although protegeswith a history of informal mentors also reported morepromotions than proteges with formal mentors, the differ-ence was not statistically significant.

These findings support the theoretical prediction thatformal mentoring relationships provide less mentoring func-tions and are less effective than informal relationships (cf.

Kram, 1985a; Kram, 1986). Our findings were consistentwith the significant findings of other research (Chao et al.,1992; Fagenson-Eland et al., 1997) and extend these studiesby examining the full range of mentoring functions andcareer outcomes related to promotion and compensation.However, one question that comes to mind when viewingthese results is why proteges in formal relationships re-ceived fewer benefits than proteges in informal relation-ships. As discussed in the introduction, existing theorypoints to a host of differences in the initiation, structure, andprocesses in formal and informal mentoring relationships.For example, formal relationships are short term and mayhave less identification, comfort, and motivation than infor-mal relationships. Although these factors remain to be as-sessed directly in the future research, they can certainlyexplain why proteges with informal mentors in the presentstudy reported more psychosocial functions and greatersatisfaction with the relationship than proteges with formalmentors. However, the differences between formal and in-formal mentoring in compensation and career developmentroles may be due to selection; proteges who are selected orselect informal relationships may be better performers thanproteges who join formal mentoring programs. In otherwords, individuals who become informal proteges wouldhave achieved more than those who become formal protegeseven without their mentor's assistance. Individuals whobecome- informal proteges may also be more career-drivenand may seek and be more responsive to their mentor'scareer development functions than individuals who becomeformal proteges. Future research could take a longitudinalapproach and collect performance measures to assess howmuch of the variance in proteges' career outcomes is due tothe protege's ability, the mentor's ability, or some combi-nation of these two variables.

In support and extension of other research (Chao et al.,1992; Dreher & Ash, 1990; Scandura, 1992), we found thatthose with informal mentors received more compensationand promotions than those without mentors, but no signif-icant differences were found between those with formalmentors and nonmentored respondents. Although thesefindings indicate that the effects of formal mentoring oncareer outcomes may be quite limited, this does not auto-matically mean that formal mentors are without value; Chaoet al., (1992), for example, found that proteges with formalmentors reported higher levels of some forms of organiza-tional socialization than nonmentored individuals. Futureresearch needs to expand the scope of career variablesstudied to capture the potential benefits of formal mentoringrelationships. For example, formal mentoring relationshipsmay be quite useful for immediate performance measures,such as on-the-job training, or as an impetus for the devel-opment of early career and performance goals.


The Impact of Gender Composition of theRelationship

The second objective of our study was to provide a directtest of new theory on gender and mentoring (Ragins, 1997a)by assessing the relationship between the gender composi-tion of the mentoring relationship and mentoring functionsand outcomes. We also explored whether the gender com-position of the relationship interacted with the type ofrelationship in influencing mentoring functions.

Our study revealed that the gender composition of thementoring relationship affected reports of mentor functionsand compensation. Although there was no support for thegeneral hypothesis that same-gender relationships wouldreport more psychosocial functions than cross-gender rela-tionships, female proteges with female mentors were sig-nificantly more likely than female proteges with male men-tors to engage in after-work, social activities with theirmentors. This finding is congruent with other research in-dicating decreased reports of social interaction for womenin cross-gender mentoring relationships (Ragins & McFar-lin, 1990). One explanation for this finding is that femaleproteges in cross-gender relationships may be reluctant toengage in after-work, social activities with their male men-tors for fear that the interaction would be misconstrued assexual in nature (Clawson & Kram, 1984; Hurley &Fagenson-Eland, 1996).

Although the presence of a male mentor was not associ-ated with more career development functions, having ahistory of primarily male mentors was significantly relatedto compensation. Specifically, when controlling for positiontenure, number of career interruptions, occupation, length ofprior mentoring relationships, and whether those relation-ships involved formal or informal mentors and supervisoryor nonsupervisory mentors, both male and female protegeswith a history of male mentors reported more compensationthan proteges with a history of female mentors. Moreover,female proteges with a history of male mentors earnedsignificantly more than female proteges with a history offemale mentors, suggesting that these findings reflect morethan simple gender differences in protege salary. Thesefindings are congruent with recent research by Dreher andCox (1996), who found a significant relationship betweenthe presence of a White male mentor and compensation. Thefindings are also congruent with existing theory that positsa relationship between mentor power and outcomes of men-toring relationships (Ragins, 1997a, 1997b). According tothis theory, male mentors can provide more career outcomesthan female mentors because they have more power inorganizations (Ragins & Sundstrom, 1989).

Our study also found some support for emerging theorythat explores outcomes associated with specific gendercombinations of mentoring relationships. Diversified men-torship theory predicts that male proteges with male men-

tors- should receive the most benefits from their relation-ships, whereas male proteges with female mentors shouldreceive the least benefits (Ragins, 1997a). We found somesupport for both propositions. Male proteges with malementors did not report more mentoring functions than othergender combinations, but male proteges with a history ofmale mentors had greater compensation than any othergender combination. Although the relatively small samplesize requires cautious interpretation, male proteges withfemale mentors did report less satisfaction with their men-tors than any other gender combination and were less likelyto report that their mentor provided acceptance in theirprofessional development. Male proteges with female men-tors were also less likely to report that their mentor providedchallenging assignments and exposure in the organizationthan female proteges with male mentors. These findingssupport theoretical predictions related to gender differencesin power and the mentor's ability to provide exposure,challenging assignments, and compensation for their pro-teges (Ragins, 1997a). The finding that male proteges withfemale mentors reported less acceptance from their mentorand less satisfaction with the relationship may also reflectpersonality differences among men who chose female, asopposed to male mentors. In a study of male graduatestudents, Infante (1990) found that men who chose femalementors were higher in self-esteem and the need for supportand affiliation than men who chose male mentors. Thissuggests that even if female mentors give equivalent supportas their male counterparts, they may still be viewed asproviding insufficient support relative to the need of theirmale proteges. Future research could explore this possibilityand examine other personality characteristics that predictmale proteges choosing female mentors and vice versa.

An unexpected but striking finding of our study was thatalthough female proteges with a history of male mentorsreceived more promotions than their male counterparts, theyreceived less compensation. This finding may reflect genderdiscrimination in organizations and is congruent with Stroh,Brett, and Reilly's (1992) finding that women managerswere promoted as frequently as their male counterparts butwere less likely to receive the compensation associated withtheir promotions. One important implication of this findingis that although male mentors may help female protegesadvance in the organization, they may not be able to buffertheir female proteges from biased compensation decisions.

The gender composition of the mentoring relationship notonly influenced mentoring functions and outcomes but alsomoderated the relationship between the type of mentor andthe challenging assignment mentoring function. Proteges incross-gender relationships reported that their mentors pro-vided less challenging assignments with formal as com-pared with informal relationships. Those in same-genderrelationships, however, actually reported receiving morechallenging assignments from their mentors when the men-


tor was formally assigned. This suggests that same-gendermentoring relationships should be used in formal mentoringprograms that are aimed at developing the protege's jobperformance by giving them stretch assignments.

A surprising finding of our study was that the benefits offormal and informal mentoring varied by protege gender.Male proteges with formal mentors not only reported morecounseling than female proteges with formal mentors, butthey also reported more of this function than both male andfemale proteges with informal mentors. It is clear that forthe counseling function, male proteges stand to gain themost from having a formal mentor. Although the presenceof a formal mentor did not have an adverse effect on maleproteges, it reduced reports of coaching, role modeling,social, counseling, and friendship functions for female pro-teges. This suggests that female proteges may have the leastto gain from entering a formal mentoring relationship.

Limitations of This Study

Several limitations of this study should be mentioned.First, although retrospective measures of mentoring historyare more appropriate than static measures for assessing therelationship between mentoring and career outcomes, retro-spective measures are self-report measures that are suscep-tible to recall and other biases. A longitudinal approachwould certainly be preferable for studying mentoring his-tory efforts, and future research should use longitudinaldata.

A second limitation of this study was the relatively smallsample of male proteges with female mentors. We recog-nized the difficulty in obtaining this sample when weviewed prior research; the Ragins and McFarlin (1990)study, for example, had only 11 of these relationships out ofa sample of 510 employees. These researchers observed thatstudies of male-dominated organizations and occupationsare unlikely to produce adequate samples of female mentorswith male proteges. With this caveat in mind, we explicitlysought to obtain gender balance in our study by using anequally weighted sample of social workers, journalists, andengineers. It is telling that of the 1,162 respondents, we hadonly 24 male proteges with female mentors. Our tentativefinding of less mentoring functions and outcomes associatedwith these relationships may be due not just to the rarity offemale mentors but also to the dynamics of the relation-ship. Future research could explore in greater depth theinterpersonal dynamics in female mentor-male protegerelationships.

A third potential limitation of our study came about as aresult of our attempt to obtain balance in the gender com-position of the mentoring relationships in our sample. Spe-cifically, although a sample of social workers, journalists,and engineers increases the probability of different gendercombinations of mentoring relationships, at the same time it

raises the possibility that occupational differences in men-toring may account for some of the findings in our study. Asa point in fact, social workers were more likely to haveformal mentors than journalists, ^(1, N = 614) = 13.44,p < .01, and female mentors were less likely to be reportedamong engineers than the other two occupations, x*(l, N =615) = 37.56, p < .001. With this in mind, we controlledfor occupation by entering it in the first step of all of ourhierarchical regression analyses and by using means thatwere adjusted for occupation in our means tests. We alsoreran the analyses separately for each of the three occupa-tions and found generally equivalent results. Nevertheless,occupational differences may still account for some of thefindings of the present study.

Finally, although regression analyses allowed us to hier-archically control for mentoring and organizational vari-ables, and the type of mentor in separate steps, the use ofmany separate regression analyses results in restricted con-trol over experiment-wide Type I error. Accordingly, mar-ginal levels of significance were not reported. We alsoreanalyzed the data using a multivariate analysis of covari-ance (MANCOVA) and ANCOVA analyses, and the regres-sion results were replicated with one exception: The three-way interaction between mentor gender, protege gender,and type of mentor for the dependent variable of challeng-ing assignments did not maintain its significance in theMANCOVA analyses.

Implications for Organizations

One clear implication of this study is that formal men-toring programs should not be considered as a substitute forinformal mentoring relationships but should be offered inpartnership with informal relationships. Formal mentors areprobably most effective when they approximate informalmentors in as many ways as possible. Along those samelines, where possible, formal mentoring programs shouldmimic the development of informal relationships. One ex-ample of this is when the organization identifies pools ofpotential mentors and proteges, trains these individuals inthe skills necessary for effective mentoring relationships,but then allows them to select their own mentoring partners(cf. Forret, Turban, & Dougherty, 1996). Another sugges-tion is for organizations to use formal mentoring relation-ships as a springboard for the development of informalrelationships; proteges with formal mentors would be en-couraged to identify and select informal mentors while inthe last stage of their formal mentoring relationship. It isimportant for organizations to avoid sending the implicitmessage that once a protege has an assigned mentor, thismentor is sufficient and that they should not attempt to gainan informal mentor. One way to accomplish this is fortraining programs to include strategies and skills for devel-oping relationships with informal mentors.


A second key implication of this study is that althoughformal mentoring relationships are less effective than infor-mal relationships, they may be even less effective for femaleproteges. This is an important issue because formal men-toring programs are being developed as part of emergingdiversity initiatives, and many are aimed at helping womenbreak through gender-related barriers to advancement (Cat-alyst, 1993; Kerry, 1994; Kram & Hall, 1996; Scott, 1992).It would therefore be misleading for organizations to offerformal mentors as a substitute for informal relationships, orfor them to suggest, directly or indirectly, that the relation-ships are interchangeable. Instead, the results of this studysuggest that for optimal promotion rate, female protegesshould develop informal mentoring relationships with malementors. However, even these male mentors are no panacea,since promotion and compensation appear to be indepen-dent events for women in organizations (Stroh et al., 1992).

Future Research

This study examined the relationship between gender, thetype of mentoring relationship, and mentoring functions andoutcomes. Although we found direct effects between gen-der, the type of mentoring relationship and compensation, itwas beyond the scope of this study to examine whethermentoring functions mediate these relationships. A reviewof the bivariate correlations suggests that the relationshipbetween mentoring functions and career outcomes is rela-tively weak and varies by the type of function. Futureresearch needs to examine mediational effects and the rela-tionship between mentoring functions and career and orga-nizational outcomes. If the mentor's behaviors or functionsdo not mediate the relationship between the type of rela-tionship and career outcomes, perhaps other mediators thathave not been studied are operative, such as the protege'sbehaviors in the relationship or the interaction betweenmentor and protege behavior.

An interesting finding in our study was that supervisorymentors provided more career development functions thannonsupervisory mentors for four of the five career develop-ment functions (sponsorship, protection, challenging assign-ments, and exposure) but did not differ in provision ofpsychosocial functions. These findings, which replicate Ra-gins and McFarlin's (1990) study, suggest that supervisorymentors may have a better assessment of the career needs oftheir proteges than nonsupervisory mentors and may be in abetter position to provide career development functionsbecause of proximity, contact, and control over work as-signments. However, although proximity may allow super-visory mentors to provide more career development func-tions, it may not allow them to become closer to theirproteges. Supervisory mentors may hold back from provid-ing psychosocial functions because intimacy and friendshipin their mentoring relationships may conflict with their

supervisory roles. These relationships should be explored infuture research. The significance of the supervisory men-toring variable also illustrates the importance of includingmentoring control variables, particularly when investigatinggender effects in mentoring relationships (cf. Ragins, 1999).

Finally, future research should explore the effects ofdifferent historical combinations of mentoring relationshipson career outcomes and processes. In this study, we exam-ined separately the historical effects of gender compositionand the type of mentor on career outcomes. The results ofthis study could be used as a foundation for the developmentof a theoretical model that explains how the history ofvarious combinations of mentoring relationships combine toaffect career outcomes. This model could examine howgender composition interacts with type of mentor (formal,informal, supervisory, and nonsupervisory) in affecting ahost of career and organizational outcomes. Which types ofrelationships are critical in which career stages for theprotege and the mentor? What individual, interpersonal andorganizational factors affect the historical composition ofmentoring relationships? Which types of compositions aremost effective for which types of situations? One point thatis driven home by the present study is that the outcomes andprocesses in mentoring relationships are influenced by thegender of the members of the relationship. Future studies,therefore, need to use large enough samples that allow foran in-depth exploration of these important, but complex,gender interaction effects.

References

Allen, T. D., Poteet, M. L., & Burroughs, S. M. (1997). Thementor's perspective: A qualitative inquiry and future researchagenda. Journal of Vocational Behavior, 51, 70-89.

Arthur, M. B., & Rousseau, D. M. (Eds.) (1996). The boundarylesscareer: A new employment principle for a new organizationalera. New York: Oxford University Press.

Beggs, J. M., & Doolittle, D. C. (1993). Perceptions now and thenof occupational sex typing: A replication of Shinar's 1975 study.Journal of Applied Social Psychology, 23, 1435-1453.

Bragg, A. (1989). Is a mentor program in your future? Sales andMarketing Management, 141, 54-59.

Burke, R. J., & McKeen, C. A. (1989). Developing formal men-toring programs in organizations. Business Quarterly, 53, 76-99.

Burke, R. J., & McKeen, C. A. (1997). Benefits of mentoringrelationships among managerial and professional women: Acautionary tale. Journal of Vocational Behavior, 51, 43-57.

Byrne, D. (1971). The attraction paradigm. New York: AcademicPress.

Catalyst. (1993). Mentoring: A guide to corporate programs andpractices. New York: Author.

Chao, G. T. (1997). Mentoring phases and outcomes. Journal ofVocational Behavior, 51, 15-28.

Chao, G. T., Walz, P. M., & Gardner, P. D. (1992). Formal andinformal mentorships: A comparison on mentoring functions


and contrast with nonmentored counterparts. Personnel Psy-chology, 45, 619-636.

Clawson, J. G., & Kram, K. E. (1984). Managing cross-gendermentoring. Business Horizons, 27, 22-32.

Dillman, D. A. (1978). Mail and telephone surveys: The totaldesign method. New York: Wiley.

Douglas, C. A. (1997). Formal mentoring programs in organiza-tions: An annotated bibliography. Greensboro, NC: Center forCreative Leadership.

Dreher, G. P., & Ash, R. A. (1990). A comparative study ofmentoring among men and women in managerial, professional,and technical positions. Journal of Applied Psychology, 75,539-546.

Dreher, G. F., & Cox, T. H., Jr. (1996). Race, gender, and oppor-tunity: A study of compensation attainment and the establish-ment of mentoring relationships. Journal of Applied Psychol-ogy, 81, 297-308.

Ensher, E. A., & Murphy, S. E. (1997). Effects of race, gender,perceived similarity, and contact on mentor relationships. Jour-nal of Vocational Behavior, 50, 460-481.

Erikson, E. H. (1963). Childhood and society. New York: W. W.Norton.

Fagenson, E. A. (1989). The mentor advantage: Perceived career/job experiences of proteges versus non-proteges. Journal ofOrganizational Behavior, 10, 309-320.

Fagenson-Eland, E. A., Marks, M. A., & Amendola, K. L. (1997).Perceptions of mentoring relationships. Journal of VocationalBehavior, 51, 29-42.

Forret, M. L., Turban, D. B., & Dougherty, T. W. (1996). Issuesfacing organizations when implementing formal mentoring pro-grammes. Leadership & Organization Development Jour-nal, 17, 27-30.

Gaskill, L. R. (1993). A conceptual framework for the develop-ment, implementation, and evaluation of formal mentoring pro-grams. Journal of Career Development, 20, 147-160.

Geiger-DuMond, A. H., & Boyle, S. K. (1995). Mentoring: Apractitioner's guide. Training and Development, 49, 51-54.

Gray, W. A. (1988). Developing a planned mentoring program tofacilitate career development. Career Planning and Adult De-velopment Journal, 4, 9-16.

Hall, D. T., & Mirvis, P. H. (1996). The new protean career:Psychological success and the path with a heart. In D. T. Hall(Ed.), The career is dead—Long live the career: A relationalapproach to careers (pp. 15-45). San Francisco: Jossey-Bass.

Kerry, W. (1994). Corporate mentoring can break the glass ceiling.HRFocus, 71, 17.

Hunt, D. M., & Michael, C. (1983). Mentorship: A career trainingand development tool. Academy of Management Review, 8,475-485.

Hurley, A. E., & Fagenson-Eland, E. A. (1996). Challenges incross-gender mentoring relationships: Psychological intimacy,myths, rumours, innuendoes and sexual harassment. Leadership& Organization Development Journal, 17, 42-49.

Infante, R. A. (1990). Gender and the mentoring process: Com-parison of men who choose male or female mentors [CD-ROM].(Abstract from: ProQuest File: Dissertation Abstracts Item:9016901)

Kalbfleisch, P. J., & Davies, A. B. (1993). An interpersonal model

for participation in mentoring relationships. Western Journal ofCommunication, 57, 399—415.

Keele, R. L., Buckner, K., & Bushnell, S. J. (1987). Formalmentoring programs are no panacea. Management Review, 76,67-68.

Koberg, C. S., Boss, R. W., Chappell, D., & Ringer, R. C. (1994).Correlates and consequences of protege mentoring in a largehospital. Group and Organization Management, 19, 219-239.

Kram, K. E. (1983). Phases of the mentor relationship. Academy ofManagement Journal, 26, 608-625.

Kram, K. E. (1985a). Mentoring at work: Developmental relation-ships in organizational life. Glenview, IL: Scott Foresman.

Kram, K. E. (1985b). Improving the mentoring process. Trainingand Development Journal, 39, 40-43.

Kram, K. E. (1986). Mentoring in the workplace. In D. T. Hall(Ed.), Career development in organizations (pp. 160-201). SanFrancisco: Jossey-Bass.

Kram, K. E., & Bragar, M. C. (1992). Development throughmentoring: A strategic approach. In D. Montross & C. Shink-man (Eds.), Career development: Theory and practice (pp.221-254). Springfield, IL: Charles C Thomas.

Kram, K. E., & Hall, D. T. (1996). Mentoring in a context ofdiversity and turbulence. In E. E. Kossek & S. A. Lobel (Eds.),Managing diversity: Human resource strategy for transformingthe workplace (pp. 108-136). Cambridge, MA: Blackwell.

Levinson, D. J., Darrow, C. N., Klein, E. B., Levinson, M. H., &McKee, B. (1978). The seasons of a man's life. New York:Knopf.

Murray, M. (1991). Beyond the myths and magic of mentoring:How to facilitate an effective mentoring program. San Fran-cisco: Jossey-Bass.

Neter, J., & Wasserman, W. (1974). Applied linear statisticalmodels: Regression, analysis of variance, and experimentaldesign. Homewood, IL: Irwin.

Noe, R. A. (1988). An investigation of the determinants of suc-cessful assigned mentoring relationships. Personnel Psychol-ogy, 41, 457-479.

Olian, J. D., Carroll, S. J., & Giannantonio, C. M. (1993). Mentorreactions to proteges: An experiment with managers. Journal ofVocational Behavior, 43, 266-278.

Olian, J. D., Carroll, S. J., Giannantonio, C. M., & Feren, D. B.(1988). What do proteges look for in a mentor? Results of threeexperimental studies. Journal of Vocational Behavior, 33, 15-37.

Ostroff, C., & Kozlowski, S. W. (1993). The role of mentoring inthe information gathering processes of newcomers during earlyorganizational socialization. Journal of Vocational Behav-ior, 42, 170-183.

Paine, F. (1986). Students, media professionals are paired in men-tor program. Journalism Educator, 11, 24-25.

Phillips-Jones, L. (1983). Establishing a formalized mentoringprogram. Training and Development Journal, 37, 38-42.

Ragins, B. R. (1989). Barriers to mentoring: The female manager'sdilemma. Human Relations, 42, 1-22.

Ragins, B. R. (1997a). Diversified mentoring relationships inorganizations: A power perspective. Academy of ManagementReview, 22, 482-521.


Ragins, B. R. (1997b). Antecedents of diversified mentoring rela-tionships. Journal of Vocational Behavior, 51, 90-109.

Ragins, B. R. (1999). Where do we go from here, and how do weget there? Methodological issues in conducting research ondiversity and mentoring relationships. In A. J. Murrell, F. J.Crosby, & R. J. Ely (Eds.), Mentoring dilemmas: Developmen-tal relationships within multicultural organizations (pp. 227-247). Mahwah, NJ: Erlbaum.

Ragins, B. R., & Cotton, J. (1991). Easier said than done: Genderdifferences in perceived barriers to gaining a mentor. Academyof Management Journal, 34, 939-951.

Ragins, B. R., & McFarlin, D. (1990). Perception of mentor rolesin cross-gender mentoring relationships. Journal of VocationalBehavior, 37, 321-339.

Ragins, B. R., & Scandura, T. A. (in press). Burden or blessing?Expected costs and benefits of being a mentor. Journal ofOrganizational Behavior.

Ragins, B. R., & Sundstrom, E. (1989). Gender and power inorganizations: A longitudinal perspective. Psychological Bulle-tin, 105, 51-88.

Scandura, T. A. (1992). Mentorship and career mobility: An em-pirical investigation. Journal of Organizational Behavior, 13,169-174.

Scandura, T. A., & Katerberg, R. J. (1988, August). Much adoabout mentors and little ado about measurement: Developmentof an instrument. Paper presented at the meeting of the NationalAcademy of Management, Anaheim, CA.

Scott, M. E. (1992). Designing effective mentoring programs:Historical perspectives and current issues. Journal of Humanis-tic Education and Development, 30, 167-177.

Smith, J. M., Chase, P. N., & Byrd, J., Jr. (1986). A formalizedmentor system in an educational setting. Engineering Educa-tion, 76, 216-218.

Stroh, L. K., Brett, J. M., & Reilly, A. H. (1992). All the rightstuff: A comparison of female and male managers' career pro-gression. Journal of Applied Psychology, 77, 251-260.

Taibbi, R. (1983). Supervisors as mentors. Social Work, 28, 237-238.

Tajfel, H. (1978). The achievement of group differentiation. In H.Tajfel (Ed.), Differentiation between social groups: Studies inthe social psychology ofintergroup relations (pp. 77-98). Lon-don: Academic Press.

Tsui, A., Egan, T. D., & O'Reilly, C. A. (1992). Being different:Relational demography and organizational attachment. Admin-istrative Science Quarterly, 37, 549-579.

Tsui, A. S., & O'Reilly, C. A. (1989). Beyond simple demographiceffects: The importance of relational demography in superior-subordinate dyads. Academy of Management Journal, 32, 402-423.

U.S. Department of Commerce. (1996). Statistical Abstract of theUnited States 1996. Washington, DC: Author.

Viator, R. E., & Scandura, T. A. (1991). A study of mentor-protegerelationships in large public accounting firms. Accounting Ho-rizons, 5, 20-30.

Whitely, W., Dougherty, T. W., & Dreher, G. F. (1991). Relation-ship of career mentoring and socioeconomic origin to managers'and professionals' early career progress. Academy of Manage-ment Journal, 34, 331-351.

Zey, M. G. (1985). Mentor programs: Making the right moves.Personnel Journal, 64, 53-57.

(Appendix follows)


Appendix

Instruments Used in Study

Mentor Role Instrument (Ragins & McFarlin, 1990)

My Mentor:

(SPONSOR)helps me attain desirable positions.uses his/her influence to support my advancement in the organization.uses his/her influence in the organization for my benefit.

(COACH)helps me learn about other parts of the organization.gives me advice on how to attain recognition in the organization.suggests specific strategies for achieving career aspirations.

(PROTECT)protects me from those who may be out to get me."runs interference" for me in the organization.shields me from damaging contact with important people in the organi-

zation.(CHALLENGE)

gives me tasks that require me to leam new skills.provides me with challenging assignments.assigns me tasks that push me into developing new skills.

(EXPOSURE)helps me be more visible in the organization.creates opportunities for me to impress important people in the organi-

zation.brings my accomplishments to the attention of important people in the

organization.(FRIENDSHIP)

is someone I can confide in.provides support and encouragement.is someone I can trust.

(SOCIAL)and I frequently get together informally after work by ourselves.and I frequently socialize one-on-one outside the work setting.and I frequently have one-on-one, informal social interactions.

(PARENT)is like a father/mother to me.reminds me of one of my parents.treats me like a son/daughter.

(ROLE MODEL)serves as a role-model for me.is someone I identify with.represents who I want to be.

(COUNSELING)serves as a sounding board for me to develop and understand myself.guides my professional development.guides my personal development.

(ACCEPTANCE)accepts me as a competent professional.sees me as being competent.thinks highly of me.

Satisfaction With Mentor Scale

My Mentor:

is someone I am satisfied with,fails to meet my needs,disappoints me (reverse-scored)has been effective in his/her role.

Note. Both instruments were measured using a 7-point Likert scale withresponses ranging from strongly disagree (1) to strongly agree (7). Itemswere randomly ordered in the survey.

Received December 23, 1997Revision received September 21, 1998

Accepted September 28, 1998 »

mentor functions and outcomes: a comparison of men …€¦ · mentor functions and outcomes: a...

Documents