|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200409Ka 4500
controlfield tag 001 002317246
007 cr bnu|||uuuuu
008 100831s2009 flua ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003004
Poteat, Laura F.
Mentorship racial composition and the judgments made by individuals external to the relationship
h [electronic resource] /
by Laura F. Poteat.
[Tampa, Fla.] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 94 pages.
Thesis (M.A.)--University of South Florida, 2009.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
ABSTRACT: The purpose of this study was to examine how the racial composition of a mentoring relationship influences three types of judgments made by individuals external to the relationship: (1) causal attributions formed to explain successful protg performance; (2) evaluations of protg career advancement potential; and (3) reward recommendations for the mentor and protg. Additionally, the associations among causal attributions, evaluations of potential, and reward recommendations were investigated. A 2 (protg race: white vs. black) x 2 (mentor race: white vs. black) factorial between-subjects design was used. Mentor and protg races were manipulated within a written vignette. After reading the vignette, participants responded to items measuring their judgments about the mentor and protg depicted in the vignette. The final sample consisted of 194 white, employed individuals. Overall, results did not support the hypothesized racial effects on the three types of judgments. However, support was found for the predicted associations among the different judgment types. Implications of these findings, as well as directions for future research, are discussed.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Advisor: Tammy D. Allen, Ph.D.
Career advancement potential
t USF Electronic Theses and Dissertations.
Mentorship Racial Composition and the Judgmen ts Made by Individuals External to the Relationship by Laura F. Poteat A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Psychology College of Arts and Sciences University of South Florida Major Professor: Tammy D. Allen, Ph.D. Walter Borman, Ph.D. Joseph Vandello, Ph.D. Date of Approval: April 27, 2009 Keywords: diversified mentoring, mentor, pr otg, race, ethnicity, attribution, reward, career advancement potential Copyright 2009, Laura F. Poteat
Dedication I dedicate this thesis to my parents, Victor and Mary Poteat, for their constant love, prayers, and support; to my brother, Paul Poteat, for his advice and encouragement; to my family and friends, for their loya lty; and to my Lord, who has shown Himself faithful to me in all things.
Acknowledgments I would like to thank my committee members, Drs. Walter Borman and Joseph Vandello, for sharing their expertise and insi ghts with me in the completion of this project. I would also like to thank my major professor, Dr. Tammy Allen, for her guidance, commitment, and encouragement thr oughout my graduate school experience. I am grateful to call her my mentor. Finall y, I would like to thank the individuals and organizations who participated in this study.
i Table of Contents List of Tables iii List of Figures v Abstract vi Chapter One Introduction 1 Attributions for Protg Performance 3 Attribution Theory: A Brief Introduction 4 Racial Differences in Performance Attributions 5 Mentor and Protg Race a nd Attributions for Protg Performance 12 Attributions to Protg Ability and Effort 12 Attributions to Mentor Help 13 Attributions to External Factors 15 Attributions, Evaluations of Potential, and Reward Allocations 17 Evaluations of Protg Potential 17 Allocation of Rewards to the Protg 18 Allocation of Rewards to the Mentor 20 Mentor and Protg Race, Evaluati ons of Potential, and Reward Allocations 21 Protg Race, Evaluations of Po tential, and Reward Allocations 22 Mentor Race and Reward Allocations 23 Chapter Two Method 24 Participants and Design 24 Procedure 25 Materials 29 Measures 33 Causal Attributions 33 Career Advancement Potential 34 Reward Recommendations 34 Manipulation Check 35 Participant Demographics 35 Chapter Three Results 36 Preliminary Analyses 36 Manipulation Check 36
ii Factor Analysis of Caus al Attributions Measure 38 Checking MANOVA Assumptions 40 Hypothesis Testing 42 Supplemental Analyses 52 Comparison of Means Across Target Names 52 Post Hoc Power Analysis 53 Mentor Race Moderator Analysis 53 Chapter Four Discussion 58 Racial Differences in Judgments: Hypotheses 1 through 7 and 12 through 14 58 Associations Among Judgmen ts: Hypotheses 8 through 11 62 Effectiveness of the Racial Manipulation 67 Study Limitations and Directi ons for Future Research 73 Conclusions 75 References 77 Appendices 84 Appendix A: Vignette 85 Appendix B: Causal Attributions Items 89 Appendix C: Career Advancement Potential Items 90 Appendix D: Reward Recommendations Items 91 Appendix E: Manipulation Check Items 92 Appendix F: Participant Demographics Items 93
iii List of Tables Table 1 Participant Demographi c Characteristics by Source 26 Table 2 Participant Demographic Characteristics by Experimental Condition 27 Table 3 Number of Participants by Source and Experimental Condition 28 Table 4 Number of Participants Meeting Criteria for Inclusion in Analyses by Source 30 Table 5 Percentage of Participants Identifying the Intended Race by Target Name 37 Table 6 Percentage of Participants Identifying the Intended Races of Both Targets by Experimental Condition 39 Table 7 Causal Attributions Factors, Items, and Pattern Coefficients 41 Table 8 Means, Standard Deviations Alphas, and Intercorrelations Among Study Dependent Variables 43 Table 9 Attributions, Evaluations of Potential, and Reward Recommendations as a F unction of Mentor Race 44 Table 10 Attributions, Evaluations of Potential, and Reward Recommendations as a Function of Protg Race 45 Table 11 Attributions, Evaluations of Potential, and Reward Recommendations as a Function of Mentor and Protg Race 46 Table 12 Intercorrelations Among St udy Dependent Variables by Mentor Race 47 Table 13 Intercorrelations Among St udy Dependent Variables by Protg Race 48 Table 14 Intercorrelations Among St udy Dependent Variables by Mentor and Protg Race 49
iv Table 15 Intercorrelations Between Attributions and Protg Rewards 51 Table 16 Hierarchical Multiple Regression Results for Mentor Reward Recommendations 56
v List of Figures Figure 1 Interaction of Mentor Race a nd Mentor Help Attributions on Mentor Reward Recommendations 56
vi Mentorship Racial Composition and the Judgmen ts Made by Individuals External to the Relationship Laura F. Poteat ABSTRACT The purpose of this study was to examine how the racial compos ition of a mentoring relationship influences three types of judgments made by in dividuals external to the relationship: (1) causal attri butions formed to explain successful protg performance; (2) evaluations of protg career advancement potential; and (3) reward recommendations for the mentor and protg . Additionally, the associations among causal attributions, evalua tions of potential, and reward recommendations were investigated. A 2 (protg race: white vs black) x 2 (mentor race: white vs. black) factorial between-subjects design was used. Me ntor and protg races were manipulated within a written vignette. After reading th e vignette, participants responded to items measuring their judgments about the mentor and protg depi cted in the vignette. The final sample consisted of 194 white, employe d individuals. Overall, results did not support the hypothesized racial effects on the three types of judgments. However, support was found for the predicted associati ons among the different judgment types. Implications of these findings, as well as dire ctions for future research, are discussed.
1 Chapter One Introduction As the workforce becomes more racially diverse, the prevalence of diversified mentoring relationships is e xpected to increase (Ragins, 2007). According to Ragins (1997), diversified mentoring re lationships involve mentors and proteges who differ in one or more group memberships that are associ ated with power in organizations, such as race, ethnicity, or gender. With respect to race, a diversified mentoring relationship is composed of a racial majority member and a racial minority member, while a homogeneous mentoring relationship is composed of two racial majo rity members or two racial minority members. As explaine d by Ragins (1997), the terms minority and majority in this context refer to a gro ups possession of power rather than a groups numerical status. In the Unite d States, there are few people of color in top organizational positions (Powell & Butterfield, 2002). Because these top positions are associated with power, people of color are considered the raci al minority, while whites are considered the racial majority. In response to the projected increase in diversified ment oring relationships, researchers have called for more research examining racial diversity and workplace mentoring (e.g., Ragins, 1997; Wanberg, Wels h, & Hezlett, 2003). One question of particular interest is whether racial minor ities experience the same mentoring outcomes as do whites. Previous research using prim arily Caucasian samples has found that both
2 mentors and proteges benefit from their me ntoring relationships. For example, in a recent meta-analysis examining career outcomes associated with being mentored, Allen, Eby, Poteet, Lentz, and Lima (2004) found that proteges reported greater career advantages than did nonproteges, includi ng greater career satisfaction, higher compensation, and more promotions. Although re search from the mentors perspective is less developed, Allen, Poteet, and Burroughs (1997) identified four cate gories of benefits received by mentors: builds a support networ k, self-satisfaction, job -related rewards that focus on the self, and job-related rewards th at focus on others. Although research has demonstrated that both mentors and proteges receive benefits from mentoring, the answer to whether racial minorities receive fewer outcomes than whites is unclear. One perspective that may be informativ e to the research examining race and mentoring outcomes was discussed by Ragi ns (1997) in her theoretical work on diversified and homogeneous mentoring rela tionships. According to Ragins (1997), individuals external to a ment oring relationship, such as othe r managers, supervisors, and peers, can influence the development and outco mes of the relationship. Therefore, it is important to examine how these individuals pe rceive and evaluate members of mentoring relationships. This is partic ularly relevant to research on race and mentoring because the way that individuals perceive and evaluate members of a mentoring relationship may differ depending on the racial composition of th e dyad. Such differences may then have important consequences for the developmen t and outcomes of the relationship. For example, if group stereotypes cause individuals to perceive a minority mentor as having less power than he/she really has, then indi viduals may attribute the protgs successful performance to factors other than the me ntors grooming (Ragins, 1997). Such an
3 attribution may then decrease the benefits r eaped by the mentor, such as the amount of organizational recognition received by the mentor. The purpose of the current study is to c ontribute to the limited amount of research on race and mentoring by examining how th e racial composition of a mentoring relationship influences the per ceptions and evaluations formed by individuals external to the relationship. More specifically, this study investigates the causal attributions that individuals external to the mentoring relationship form to explain the performance of a successful protg. In addition, the study ex amines how individuals external to the mentoring relationship evaluate the potential of, and allocate rewards to, the members of the relationship. While there are many possi ble racial combinations that could be examined, the current study focuses on percep tions and evaluations of mentoring dyads composed of black and white individuals. By examining how the judgments made by individuals external to a me ntoring relationship about the mentor and protg may vary with dyad racial composition, th is study aims to make a valuable contribution to the research on racial diversity and workplace mentoring. Attributions for Pr otg Performance When observing the performance of a pr otg, individuals external to the mentoring relationship may form causal attri butions to explain the performance. For example, individuals may attribute the performa nce to the protgs ability or effort, the mentors help, or to external factors. Research examining racial differences in performance attributions suggests that the racial composition of the mentoring dyad may influence the attributions formed to explain protg performance. Differences in these attributions may have importa nt consequences for the ment ors and proteges involved.
4 The following section will provide a brief intr oduction to attribution theory, review the literature on racial differences in performan ce attributions, and discuss the implications for mentoring dyads with different racial co mpositions. This discussion will lead to the presentation of several hypotheses concerning how attributions are expected to differ by mentor and protg race. Attribution Theory: A Brief Introduction Attribution theory deals with the pr ocess by which individuals form causal explanations for behavioral events (Kelley, 1967). According to Kelley, people examine how behavior covaries with possible causes and rely on three type s of information to make causal attributions distinctiveness, consistency, and consensus. Much of the research investigating the types of attributi ons that people make to explain the outcomes of achievement-related events has followed the classification system proposed by Weiner et al. (1972). According to this system, individuals use four ca usal attributions to explain achievement-related outcomes: ability, effort task difficulty, and luck. These four attributions can be classified along the two dimensions of lo cus of control (internal vs. external) and degree of stabil ity (stable vs. unstable). W ithin the locus of control dimension, ability and effort are considered in ternal causes, while task difficulty and luck are considered external causes. Within the stability dimension, ability and task difficulty are considered stable causes, while effort and luck are considered unstable causes. The study of the attribution process in the context of the workplace is important because research has shown that the attri butions formed by supervisors to explain subordinate performance are linked to impor tant outcomes (e.g., Green & Mitchell, 1979; Greenhaus & Parasuraman, 1993; Heilman & Guzzo, 1978; Martinko & Gardner, 1987;
5 Pazy, 1986). For example, Heilman and Guzzo found that organizational rewards (pay raises, promotions) were judged as more appropriate when employee success was attributed to ability and effort than when succe ss was attributed to luck or task difficulty. Taking a broader perspective, researchers have suggested that supervisors causal explanations for subordinates performan ce influence their behavior toward the subordinates in terms of ev aluations, rewards, and punishments, as well as their expectations about subordinate s future performance (Martinko & Gardner, 1987; Green & Mitchell, 1979). Thus, it seems important to investigate the attribut ional process in the context of the workplace. Racial Differences in Performance Attributions While Kelley (1967) presented a rational information processing model to explain how individuals form causal attributions, there are a number of factors that can affect the ideal attributional process. For example, res earchers have proposed that factors such as psychological closeness between the actor and observer and personal characteristics of the actor and observer (e.g., gender and race ) can influence attributions (Green & Mitchell, 1979; Martinko, Douglas, & Harvey, 2006). Research examining the effect of race on the attributional process has found racial differences in the attributions formed to explain black and white performance. In general, white observers tend to form less favorable attributions for black performers than for white performers (e.g., Greenhaus & Parasuraman, 1993; Jackson, Sullivan, & Hodge, 1993; Yarkin, Town, & Wallston, 1982). On the other hand, results concerning the attributions formed by black observers appear to be less consistent with some studies finding black-favoring responses (e.g., Chatman & von Hippel, 2001; Stephan, 1977, as reinterpreted by Hewstone, 1990), and
6 others finding no effect (e.g., Banks, McQuar ter, & Pryor, 1977, as cited in Pettigrew, 1979; Whitehead, Smith, & Eichhorn, 1982). In the current study, the focus is on the perceptions and evaluations of white observe rs, and the discussion and hypotheses that follow are written from this perspective. The most common explanation for racial di fferences in performance attributions is based on the role of stereotypes in the attr ibutional process. St ereotypes are cognitive structures that contain an individuals knowledge, beliefs, and expectancies about the characteristics and behaviors of a particular group of people (Hamilt on & Trolier, 1986). These cognitive categories influence how indi viduals perceive and evaluate members of the stereotyped group. In studies of racial stereotypes, it has been found th at stereotypes of blacks often include assumptions of incomp etence, and whites expectations for black performance are often low (Pettigrew & Martin 1987). Such findings stand in agreement with existing theory linking group membership to perceptions of competence. According to status characteristics theory, people infer an individuals characteristics and abilities and form performance expectations based on st atus characteristics such as age, sex, and race (Berger, Cohen, & Zelditch, 1972). Those high in status characteristics (e.g., males, whites) are assumed to possess greater task competence than those low in status characteristics (e.g., females, blacks; Nemet h, 1988). The expectations formed based on the possession of status characteristics in fluence perceptions a nd evaluations of the individuals actual performance. In order to explain the racial differences in performance attr ibutions, attribution theorists have proposed that stereotypes influence causal attributions by shaping individuals expectations about performance (e.g., Greenhaus & Parasuraman, 1993;
7 Heilman, 1983; Jackson et al., 1993; Yarkin et al., 1982). For example, stereotypes characterizing blacks as incompetent may lead to expectations for poor performance. The type of attribution made then follows from a comparison of actual performance with the stereotype-based expectations for perf ormance (Jackson et al., 1993). If an individuals actual performance is consistent with expecta tions, there is a tendency to attribute the performance to internal causes such as ability (Heilman, 1983; Jackson et al., 1993). On the other hand, if performance is inconsistent with expe ctations, there is a tendency to attribute it to exte rnal causes (e.g., task difficulty and luck), or to internal, unstable causes (e.g., effort). Such attributi ons allow the observer to maintain his/her stereotypes and expectations about the performer (Heilman, 1983; Hewstone, 1990). Applying these propositions to the performance of black versus white employees, one would expect the successful performance of black employees to be more likely to be attributed to high effort, ease of the task, or good luck than the su ccessful performance of white employees (Ilgen & Youtz, 1986). A dditionally, one would expect the successful performance of white employees to be more likely to be attributed to the employees ability than the successful perf ormance of black employees. In general, research has provided support for the above propositions. For example, Yarkin et al. (1982) manipulated the sex and race (black or white) of a target bank employee and asked white college student s to make causal attributions for the employees successful performance. Results showed that the success of white stimulus persons was more likely to be attributed to ability and less likely to be attributed to motivation than was the success of black stimul us persons. However, the black and white stimulus persons did not differ significantly in attributions to task difficulty or luck.
8 Upon examining the interactions of stimulus person race and sex, Yarkin et al. found that participants attributed greater ability, less effort, and less lu ck to the performance of the white male compared to the performance of the white female, black male, and black female. Overall, these findings provide so me support for the propositions regarding the influence of racial stereot ypes on performance attributions. In another study, Jackson et al. (1993) asked white colleg e undergraduates to rate the importance of several factor s in explaining the weak or strong performance of a black or white college applicant. Contrary to expectations, they found that participants were not more likely to attribute strong white pe rformance to ability than strong black performance. However, they did find that strong black performance was more likely to be attributed to the external causes of task ch aracteristics and luck, as well as the internal, unstable cause of effort, than was strong white performance. According to Jackson et al., the finding that participants rated ability as important in explaining the strong performance of both white and black applicants may have been a function of the student sample, as the students may have favored attr ibuting academic performance to internal causes. Overall, Jackson et al. conclude d that their findings provided support for a preliminary model of the effects of stereoty pes on attributions, in which stereotypeconsistent performance is more likely to be at tributed to internal causes, and stereotypeinconsistent performance is more likely to be at tributed to external ca uses or to internal, unstable causes. Although much of the research on racial differences in performance attributions has relied on the use of hypothetical s timulus persons, Greenhaus and Parasuraman (1993) designed a study using ac tual employees to examine the effects of race and gender
9 on the performance attribution process. Th eir sample consisted of black and white managers and their supervisors. The managers were matched on a number of background characteristics, including age, organizatio nal tenure, job function, and organizational level. After rating the managers job performa nce, the supervisors were asked to rate the importance of five factors in explaining the managers performance. In addition to the four attributions included in Weiner et al.s (1972) framework (ability, effort, task difficulty, and luck), Greenhaus and Parasura man included the fifth attribution of help from others, because they believed that stereotypes may lead people to attribute successful minority performance to the efforts of others. Their results showed that black managers performance was less likely to be attributed to ability and effort, and more likely to be attributed to help from others, than was white managers performance. Thus, these findings provide additional evidence for the role of racial stereotypes in the attributional process and demons trate that racial differences in performance attributions can be found in a field setting. Also relevant to the discussion of racial differences in performance attributions is research on what Pettigrew (1979) termed the ultimate attribution error. According to Pettigrew, when individuals explain the beha vior of ingroup and outgroup members, they tend to make attributions that allow them to maintain their negative stereotypes about the outgroup. Thus, when an outgroup member perf orms a negative act, his/her behavior is likely to be attributed to internal, dispos itional causes. On the other hand, when an outgroup member performs a stereotype-inc onsistent positive act, an observer may attempt to explain away the positive behavior by attributing it to external, situational factors. Pettigrew proposed f our possible attributions that may be used to explain the
10 positive behavior of an outgroup member, and Pettigrew and Martin (1987) described more specifically how these four attributions may be used by whites to explain successful black performance. According to Pettigrew and Martin, the first possible attribution occurs when white perceivers distinguish the successful black performer from other black individuals as an exceptional case. In this instance, the white perceiver attempts to differentiate the successful black from other black individuals, and may even exaggerate the black individuals positive qualities or use the successful black as proof against claims of organizational prejudice or disc rimination. The second possibility is that whites may attribute black success to luck or to unfair special advantage. Thus, in this case, the success is seen as due to fact ors that are temporary or beyond the black individuals control, ra ther than as due to the black indivi duals own skills and abilities. The third way that whites may attempt to explain black success is by attributing it to extremely high motivation and effort. Such hi gh effort may be seen by whites as unstable and as a way for the black individual to compensa te for lack of talent or ability. Finally, the fourth possibility occurs when whites attr ibute black success to external situational factors, such as the availabil ity of good equipment or the rece ipt of plentiful assistance. Again, such an attribution overlooks the possibi lity that the black individuals success is due to his/her own skills and abil ities. Overall, each of th ese four possible explanations for black success allows white perceivers to ma intain their original negative stereotypes of blacks. In a review of the liter ature on intergroup causal attribution, Hewstone (1990) evaluated the level of support for Pettigrews (1979) ultima te attribution error. In general, Hewstone found that attributions tend to favor ingroup over outgroup members.
11 More specifically, some studies have found that stronger inte rnal attributions are made when members of the ingroup engage in posi tive behavior than when members of the outgroup engage in positive behavior. In addi tion, some studies have found that outgroup success is more likely than ingroup success to be explained away by attributing it to good luck, high effort, or task ease. Overall, He wstone concluded that there was some support for Pettigrews ultimate attribution error and th e associated predictions. However, given that the evidence was not overwhelming, he al so suggested using the more modest label of intergroup attributional bias to refer to this set of hypotheses. In his review of the intergroup attrib ution literature, Hewstone (1990) also discussed the possible roles of cognitive and motivationa l factors in the intergroup attributional bias. With respect to cognitive factors, individuals may make attributions that allow them to maintain their stereotypes about members of certain groups. Thus, to explain expectancy-confirming behavior, individuals may rely on dispositional attributions, even to the excl usion of possible situ ational factors. On the other hand, to explain expectancy-disconf irming behavior, individuals may use more thorough attributional processing and may be more likely to attribute the behavior to situational factors. In his discussion of the potential motivational f actors underlying the intergroup attributional bias, Hewstone re ferred to social identity th eory (Tajfel & Turner, 1985). According to this theory, individuals de fine themselves in part by their group memberships and seek to maintain a positive social identity. Hewstone proposed that group members may use intergroup attributions to achieve, enhance, or protect a positive social identity, contributi ng to the observed bias in intergroup attributions.
12 To summarize, existing research reveals r acial differences in the attributions made to explain black and white performance and provides support for the role of racial stereotypes in explaining these differences. In general, research has shown that the successful performance of whites is more likely to be attributed to internal causes, such as ability, than the successf ul performance of blacks. Additionally, compared to the successful performance of whites, the successful performance of blacks is more likely to be attributed to external causes, such as luc k, task difficulty, and help from others, and to internal, unstable causes, such as effort. Such findings have important implications in the organizational setting, as researchers have al so found evidence that the attributions formed by supervisors to explain their subord inates performance ar e linked to valuable employee outcomes, such as pay raises and promotions. Mentor and Protg Race and Attrib utions for Protg Performance Results from research examining racial differences in performance attributions suggest that when individuals external to a mentoring re lationship observe a protgs performance, the causal attributions they form to explain this performance may vary as a function of the mentors and protgs race. In general, there are th ree basic elements to which an observer may attribute the protgs performance: the protg, the mentor, or external factors. The following sections pr esent hypotheses regarding how attributions to each of these elements are expected to vary by mentor and protg race. Attributions to protg ability and effort Regarding the first possibility of attributing the protgs performance to th e protg, an observer may attribute the performance to either the protgs ability or to the protgs effort. As discussed earlier, previous research has shown th at white success is more likely to be attributed to ability
13 than is black success, while black success is more likely to be attributed to effort than is white success (e.g., Greenhaus & Parasuraman, 1993; Jackson et al., 1993; Yarkin et al., 1982). Thus, the following hypotheses are proposed: Hypothesis 1 The success of white protgs is more likely to be attributed to protg ability than is the success of black protgs. Hypothesis 2 The success of black protgs is more likely to be attributed to protg effort than is th e success of white protgs. Attributions to mentor help The second possibility is that an observer may attribute the protgs performance to the help the protg received fr om the mentor. In their study of black and white managers, Gr eenhaus and Parasuraman (1993) found that the performance of black managers was more likely to be attr ibuted to help from others than was the performance of white managers. When applied to the current context of mentoring, these findings suggest that an obser ver may be more like ly to attribute the successful performance of a black protg to the help received from the mentor than the successful performance of a white protg. Furthermore, as explai ned by Ragins (1997), the mentor may be perceived as providing reme dial attention to the black protg. This leads to the following hypothesis: Hypothesis 3 The success of black protgs is more likely to be attributed to the mentors help than is the success of white protgs. In addition to the protgs race influenci ng attributions to help from the mentor, the mentors race may also influence the degr ee that an observer attributes the protgs performance to the mentors help. As discussed by Ragins (1 997), group membership, and the associated group stereotypes, influence perceptions of power and competence,
14 such that minority group members may be perceived as having less power and competence than they really possess. If bl ack mentors are percei ved as possessing less power or competence, they may be perceived as less able to meet their proteges career needs (Ragins, 1997). As a result, black mentor s may be seen as less responsible for and receive less credit for their proteges succe ssful performance. The following hypothesis reflects these ideas: Hypothesis 4 White mentors receive more cred it for the success of their protgs than do black mentors. In other words, the success of white mentors protgs is more likely to be attributed to the mentors help th an is the success of bl ack mentors protgs. When considering attributions to help from the mentor, not only may mentor and protg race exert the main effects described above, but there may also be an interaction between mentor and protg race. Specificall y, the effect of prot g race may depend on the race of the mentor such that the effect of protg race is stronger when the mentor is white than when the mentor is black. As disc ussed previously, indivi duals external to the mentoring relationship may hold st ereotypes that lead them to believe that a black mentor lacks the power or competence to meet a prot gs needs, regardless of the protgs race. In this case, the race of the protg ma kes less of a difference, as the black mentor will receive little credit fo r the success of either a black or a white protg. On the other hand, the race of the protg may make more of a difference when the mentor is white and perceived as having the power and compet ence to meet the protgs needs. Under these circumstances, protg race may have a stronger effect on the degree to which individuals external to the relationship attribute the prot gs success to the mentors
15 help, such that their attribu tions are stronger when the protg is black than when the protg is white. These ideas are su mmarized in the following hypothesis: Hypothesis 5. The effect of protg race on attr ibutions of protg success to the mentors help depends on the race of the mentor, such that the effect is stronger when the mentor is white than when the mentor is black. Attributions to external factors When those observing a mentoring relationship attribute a protgs performance to elements other than the mentor or protg, they are making attributions to external factors. Acco rding to attribution th eorists, observers may attempt to explain an individua ls stereotype-inconsistent perfo rmance by attributing it to external, situational fact ors (e.g., Heilman, 1983; Hewstone, 1990; Pettigrew, 1979). Research on racial differences in performa nce attributions supports this proposition, finding that the stereotype-incons istent performance of successf ul blacks is more likely to be attributed to external causes, such as lu ck and task difficulty, th an is the stereotypeconsistent performance of successful whites (e.g., Hewstone, 1990; Jackson et al., 1993). Applying these findings to the context of a mentoring relationship leads to the following hypothesis concerning attributions of pr otg success to ex ternal factors: Hypothesis 6 The success of black protgs is more likely to be attributed to external factors (e.g., luck or task difficu lty) than is the succe ss of white protgs. When examining attributions of protg su ccess to external fact ors, it is important to consider the race of both the mentor and the protg. In addition to the main effect of protg race hypothesized above, there may be an interaction between mentor and protg race. Specifically, the effect of mentor race on attrib utions of protg success to external factors may depend on the race of the prot g, such that the effect of mentor race
16 is stronger when the protg is black than when the protg is white. When the protg is white, individuals observing the mentoring rela tionship may be unlikely to attribute the protgs success to external factors, regardle ss of the race of the mentor. Thus, mentor race would have little effect on the degree that the white protg s success is attributed to external factors. On the ot her hand, when the protg is black, the race of the mentor may have more of an effect on attributions to external factor s. Specifically, when a black protg is paired with a black mentor, i ndividuals may be more likely to attribute successful protg performance to external f actors than when a black protg is paired with a white mentor. This proposition is based on the id ea that individuals holding negative stereotypes of black competency woul d be unlikely to attribute a black protgs success to either the protg or a black mentor, and would thus attribute the success to factors external to both memb ers of the relationship. In co ntrast, if a black protg is paired with a white mentor, the protgs suc cess may be attributed to a lesser extent to external factors because the white mentor ma y be seen as making a greater contribution to the protgs success. Thus attributions of protg suc cess to external factors would be greater when a black protg is paired wi th a black mentor than when a black protg is paired with a white mentor. Taken toge ther, these ideas agree with Ragins (1997) suggestion that if a mentoring dyad is com posed of a minority me ntor and a minority protg, the protgs successful performance may be attributed to extraneous factors. The following hypothesis summarizes these ideas: Hypothesis 7 The effect of mentor race on at tributions of protg success to external factors depends on the race of the protg, such that the effect is stronger when the protg is black than when the protg is white.
17 Attributions, Evaluations of Potential, and Reward Allocations Within the context of the workplace, th e importance of the causal attributions formed to explain the performance of others is seen in their influence on subsequent judgments. For example, past research has demonstrated associations between attributions and evaluations of managerial potential a nd promotability, as well as allocations of rewards (Allen, Russell, & Rush, 1994; Greenhaus & Parasuraman, 1993; Heilman & Guzzo, 1978; Pazy, 1986). With re spect to the current study, this suggests that the attributions formed by individuals ex ternal to a mentoring relationship to explain the successful performance of a protg may influence important judgments made by these individuals about the prot g and mentor. If the causal attributions vary with the racial composition of the mentoring relations hip as hypothesized, judgments of potential and reward allocations may also vary with dyad racial composition, potentially resulting in differences in the benefits received by protgs and mentors for participating in a mentoring relationship. The di scussion that follows reviews past research that has examined the link between attributions and evaluations of potential and reward allocations, and applies this information to the current study to fo rm specific hypotheses regarding the associat ion between attribu tions and judgments about protgs and mentors. Next, the section immediately follo wing this discussion draws from these ideas to present hypotheses about how evaluations of potential and reward allocations are expected to differ between white ve rsus black mentors and protgs. Evaluations of Protg Potential In the literature examining the conseque nces of causal attributions in the workplace, researchers have reported an important association between the causal
18 explanations provided for an employees perf ormance and evaluations of that employees managerial potential and promotability. Sp ecifically, research ers have found that attributions to employee ability play an importa nt role in these type s of evaluations. For example, Heilman and Guzzo (1978) found that, when an employees success was attributed to ability, raters evaluated th e employee as having higher top management potential than employees whose success was attr ibuted to effort, luck, or task difficulty. In a more recent field study, Greenhaus a nd Parasuraman (1993) found that ability attributions were positivel y related to assessments of promotability. A possible explanation for these findings is that the attributions raters form to explain an employees performance influence the raters expectations for the employees future performance (Green & Mitchell, 1979; Weiner et al., 1972). If employee perf ormance is attributed to stable causes, such as ability, it is perceived as likely to co ntinue in the future (Green & Mitchell, 1979). Thus, an employee whose successful performance is attributed to his/her ability may be expected to perfor m at a high level in the future. Such expectations are likely to influence the raters evaluation of the employees career advancement potential. Specifically, attribut ions to ability are positively related to ratings of potential (Greenhaus & Parasu raman, 1993). Based on this explanation and findings from previous research, the following hypothesis is proposed: Hypothesis 8. There is a positive relationship between attributions to protg ability and ratings of protg career advancement potential. Allocation of Rewards to the Protg In addition to examining the associat ion between causal attributions and evaluations of employee potential researchers have also cons idered how attributions are
19 related to the allocation of various organizational rewards. In general, performance attributed to internal causes has been asso ciated with greater reward allocations. For example, Heilman and Guzzo (1978) found th at raters judged both a pay raise and a promotion as more appropriate rewards when an employees successful performance was attributed to ability or effort rather than to luck or task difficulty. Furthermore, while raters judged a pay raise as equally appropria te for success due to ability or effort, they judged a promotion as more appropriate for suc cess due to ability than to effort. Thus, attributions to ability appear to be associ ated with greater and more desirable rewards than do attributions to other causal factors. Pazy (1986) al so found that raters perceived promotions to be more appropriate for ability-based success than for effort-based success. In another study examining the associat ion between attributions and reward recommendations, Allen et al. (1994) found that attributions to ability were positively related to all six of their reward measures, while attributions to effort were positively related to only two of the six rewards. Thus, while both types of internal attributions are associated with greater reward recommendations ability attributions appear to be more so. The explanation provided by researchers fo r this effect is similar to the explanation presented earlier for the finding th at ability attributions are as sociated with evaluations of employee potential: When performance is attribut ed to stable causes, such as ability, it is perceived as likely to continue in a similar ma nner in the future. On the other hand, when performance is attributed to uns table causes, such as effort, it is more difficult to predict future performance. Because the allocati on of rewards such as promotions involves judgments about an employees ability to ma intain high performance, decision makers may be more confident allocating such rewards to employees whose successful
20 performance is attributed to ability rather than effort The following hypotheses are based on the results of previous research discussed above: Hypothesis 9. There is a positive relationship between attributions to protg ability and effort and prot g reward recommendations. Hypothesis 10. Attributions to protg ability are associated with greater reward recommendations than are attributions to protg effort. Allocation of Rewards to the Mentor Individuals external to a mentoring relationship may also deem it appropriate to allocate organizational rewards to the mentor for his/her contributions to the performance of the protg. Research examining the be nefits of mentoring for the mentor provides support for this proposition. Quantitative studi es investigating the actual career outcomes associated with being a mentor have found mentoring others to be related to both objective (salary, promotion rates) and subjec tive measures of caree r success (e.g., Allen, Lentz, & Day, 2006; Bozionelos, 2004). Althou gh additional research is needed to test the processes by which mentoring others may re late to mentor career success, one of the processes proposed by resear chers suggests that ment ors may be rewarded by organizational decision makers who recognize their contributions to the organization. Case study and qualitative research lends suppor t to this proposed process. For example, through in-depth interv iews with mentors, Allen et al (1997) found that mentoring was associated with increased organizational visibility and recognition for the mentors. Such visibility and recognition may enhance the mentors prospects for receiving organizational rewards. Ramaswami and Dreh er (2007) describe this process in more detail. According to these re searchers, a mentors visibil ity, reputation, and credibility
21 are enhanced by a protgs successful perf ormance. As others in the organization become aware of the mentors ability to identify and prom ote talent, the mentor earns respect, admiration, and recognition for his/her contributions to the organization. Senior management may respond by showing a greater willingness to sponsor the mentors other activities and by assigning additiona l protgs to the mentor. As a whole, these activities may result in career and salary attainment for the mentor. This research on the benefits of mentor ing for the mentor suggests that, when individuals external to a mentoring relationship observe a protgs successful performance, they recognize and admire th e mentors contributions. However, as discussed earlier, the amount of credit give n to mentors for their protgs success is likely to vary across relationships (Ragins, 1997). Thus, the rewards allocated to mentors for their efforts may also vary. The associ ation between the assignm ent of credit and the allocation of organizational rewards is suppor ted by research. For instance, Crant and Bateman (1993) found that, when an actor wa s assigned a high level of credit for a successful performance, he rece ived a greater allocation of re wards. When applied to the current study, these findings suggest that mentors who receive more credit for the successful performance of their protgs will receive greater reward allocations. This leads to the following hypothesis: Hypothesis 11. There is a positive relationship between attributions to mentor help and mentor reward recommendations. Mentor and Protg Race, Evaluations of Potential, and Reward Allocations The previous two sections have argued th at there are racial differences in the causal attributions formed to explain protg performance and that these attributions are
22 linked to evaluations of potential and reward allocations. When these two arguments are combined, it follows that there are racial diffe rences in judgments of potential and reward allocations. Thus, the purpose of this section is to presen t specific hypotheses regarding how these judgments vary by mentor and protg race. Protg Race, Evaluations of Potential, and Reward Allocations According to Hypothesis 1, the success of white protgs is more likely to be attributed to their ability than is the success of black protg s. In contrast, Hypotheses 2, 3, and 6 predict that black protg success is mo re likely than white protg success to be attributed to protg e ffort, mentor help, and external fact ors, respectively. As discussed earlier, attributions of success to ability appear to be the most favorable, in that they are associated with higher ratings of potential and greater reward allocations (Allen et al., 1994; Greenhaus & Parasuraman, 1993; Heilman & Guzzo, 1978). Thus, it seems reasonable to predict that individuals extern al to a mentoring relationship will provide higher ratings of potential and greater reward allocations to white protgs than to black protgs. Additional support for this predic tion comes from field research examining racial differences in these kinds of judgments. For ex ample, researchers have found whites to receive higher ratings of promotab ility than blacks (Greenhaus, Parasuraman, & Wormley, 1990; Landau, 1995) and have found white s to be more likely to be promoted than blacks (Elvira & Zatzick, 2002; Maum e, 1999; Powell & Butterfield, 1997, 2002). Furthermore, some researchers have suggested th at the possibility for racial bias to enter into evaluations of potential may be elevated due to the complex, subjective nature of such evaluations (Landau, 1995; Ruble, Cohe n, & Ruble, 1984). Ta ken together, these ideas and findings lend suppor t to the following hypotheses:
23 Hypothesis 12 White protgs receive highe r ratings of career advancement potential than black protgs. Hypothesis 13 White protgs receive grea ter reward recommendations than black protgs. Mentor Race and Reward Allocations According to Hypothesis 4, white mentors ar e expected to receive more credit for the successful performance of their proteges than are black mentors. Combining this proposition with past research that has f ound a positive association between assignments of credit and allocations of organizational re wards (Crant & Batema n, 1993) leads to the following hypothesis: Hypothesis 14 White mentors receive greater reward recommendations than black mentors.
24 Chapter 2 Method Participants and Design The final sample consisted of 194 white, employed individuals, who worked at least 20 hours a week and had been employed in their current job for at least 6 months. Participants were recruited fr om five sources: (1) a nationa l engineering consulting firm, (2) a pool of undergraduate psychology students, (3) two graduate-level business classes, (4) two medical industry call centers, and (5 ) a local government agency. Of the 194 participants that comprised the final sample, 101 were employees of a national engineering consulting firm, lo cated in offices throughout the United States. Seventy one of the 194 participants were undergraduate students taking psychology courses at a large southeastern university. These students received course credit for their participation in this study. The final sample also included 11 students from two graduate-level business classes at the same university. A total of 6 employees from two medical industry call centers located in the Midwest and southeas tern regions of the United States were included in the final sample. Lastly, the final sample included 5 employees from a government office located in the southeastern United States. Overall, the final sample consisted of 115 females and 79 males. The average age was 34.48 ( SD = 13.27) and the median level of education reached was a four-year college degree. Mean job tenure was 5.75 years ( SD = 5.95) and the mean number of
25 hours worked per week was 39.48 ( SD = 10.62). A variety of job titles and industries were represented in the sample. The major ity of participants ( 68.0%) reported having experience in a formal and/or informal workplace mentoring relationship. More specifically, 17.0% reported havi ng experience as a protg only, 8.2% reported having experience as a mentor only, and 42.8% reporte d having experience as both a mentor and protg. Furthermore, most participants ( 77.3%) reported having supe rvisory experience. The mean number of years of supervis ory experience for this group was 8.07 ( SD = 8.52). Thus, these data suggest that most of the participants had persona l experience with the topics addressed in this st udy (i.e., mentoring, performance ev aluations). Table 1 shows demographics by sample source. The experiment was a 2 (protg race: wh ite vs. black) x 2 (men tor race: white vs. black) factorial between-subjects design. Th e races of the mentor and protg were manipulated by using a written vignette, and participants were randomly assigned to one of the four experimental condi tions. The final sample incl uded 51 participants in the white mentor-white protg condition, 47 partic ipants in the black mentor-black protg condition, 47 participants in the white mentor-black protg condition, and 49 participants in the black mentor-white protg condition. Table 2 presents the demographic characteristics of the particip ants in each condition, and Table 3 shows the number of participants from each samp le source assigned to each condition. Procedure Across the five sample sources, a tota l of 1,261 individuals were recruited to participate in the study. All of these indivi duals were sent an em ail inviting them to participate in the study and providing them with a link to the on line survey, with the
26 Table 1 Participant Demographic Characteristics by Source Demographic characteristic Source Engineering firm Undergraduate pool Graduate classes Call centers Government office N 101 71 11 6 5 No. of males, females 58 M, 43 F 13 M, 58 F 6 M, 5 F 0 M, 6 F 2 M, 3 F Mean age ( SD )a 42.39 ( 11.32 ) 23.51 ( 7.32 ) 30.36 ( 5.87 ) 31.20 ( 8.70 ) 43.00 ( 16.69 ) Median education 4-year degree 2-year degree Graduate work Some college Graduate work Mean job tenure ( SD )a 7.84 ( 6.60 ) 2.87 ( 3.51 ) 4.06 ( 4.17 ) 4.46 ( 3.25 ) 9.83 ( 7.58 ) Mean hours worked per week ( SD ) 45.87 ( 6.21 ) 30.13 ( 9.20 ) 40.00 ( 10.00 ) 43.67 ( 4.97 ) 37.00 ( 9.75 ) Percent with mentoring experience 78.2% 62.0% 27.3% 50.0% 60.0% Percent with supervisory experience 87.1% 62.0% 90.9% 83.3% 60.0% Note Values enclosed in parentheses represent st andard deviations. M = males. F = females. aAge and job tenure were coded in years.
27 Table 2 Participant Demographic Characteristics by Experimental Condition Demographic characteristic White mentor Black mentor White protg Black protg White protg Black protg N 51 47 49 47 No. of males, females 25 M, 26 F 19 M, 28 F 18 M, 31 F 17 M, 30 F Mean age ( SD )a 35.06 ( 15.39 ) 35.83 ( 13.92 ) 33.19 ( 12.04 ) 33.83 ( 11.44 ) Median education 4-year degree 4-year degree 4-year degree 4-year degree Mean job tenure ( SD )a 4.85 ( 5.38 ) 6.35 ( 6.61 ) 4.94 ( 4.69 ) 6.98 ( 6.85 ) Mean hours worked per week ( SD ) 38.53 ( 11.37 ) 38.40 ( 9.87 ) 41.06 ( 9.60 ) 39.94 ( 11.56 ) Percent with mentoring experience 66.7% 70.2% 73.5% 61.7% Percent with supervisory experience 80.4% 76.6% 77.6% 74.5% Note Values enclosed in parentheses represent st andard deviations. M = males. F = females. aAge and job tenure were coded in years.
28 Table 3 Number of Participants by Sour ce and Experimental Condition Source White mentor Black mentor White protg Black protg White protg Black protg Engineering firm 24 24 30 23 Undergraduate pool 19 19 15 18 Graduate classes 1 3 2 5 Call centers 4 0 2 0 Government office 3 1 0 1
29 exception of those individuals recruited fr om the medical industry call centers. Individuals from the call centers were pr ovided with a paper version of the study materials. Each recruited individual received the form of the survey that corresponded to the experimental condition to which they were randomly assigned. After reading the informed consent and survey instructions, pa rticipants were asked to read a written vignette and complete the survey items. All responses were submitted to the researcher and were both anonymous and confidential. A total of 498 surveys were returned to the researcher. Of these, 436 contained complete data, resulting in a response rate of 34.6%. Of the 436 completed surveys, 308 were submitted by participants meeting the studys inclusion criteria (i.e., white participants who worked at least 20 hours per week and had been employed in their current job for at least 6 months). Of th ese 308 participants, 194 correctly responded to the experimental manipulation and were included in the su bsequent analyses. Table 4 shows the number of participants meeting each of these hurdles, broken down by source. Materials Participants were asked to complete a survey packet consisting of the informed consent, survey instructions, a written vi gnette, and measures assessing the studys dependent variables. In or der to disguise the true pur pose of the study, the survey instructions stated that the purpose of the study was to examine formal workplace mentoring relationships. After reading the survey instructions, participants proceeded to the written vignette, which was created to serve as the s timulus for the present study. A copy of the
30 Table 4 Number of Participants Meeting Criteria for Inclusion in Analyses by Source Criterion met Source Engineering firm Undergraduate pool Graduate classes Call centers Government office 1. Recruited to participate in study 841 222 77 103 18 2. Returned survey to researcher 232 156 31 69 10 3. Completed required survey items 206 148 25 47 10 4. Met study inclusion criteria 174 100 16 11 7 5. Responded correctly to manipulation 101 71 11 6 5
31 vignette is provided in Appendix A. The vignette described the formal mentoring program developed by a fictitious financial in stitution (XYZ Bank). As explained in the vignette, this program was designed to ma tch new bank branch managers with more senior-level branch managers. There were three qualities associated with the occupation of bank branch management that made it an appropriate choice as the context of the studys vignette. First, creating a vignette in which bank branch managers were involved in a formal mentoring program was realisti c because such programs were actually found in the banking industry (Best practices, 2006). Second, statistics reported by the U.S. Department of Labor (U.S. Department of Labor & U.S. Bureau of Labor Statistics, 2005) showed that women represented about half of those employed as financial managers, suggesting that the occupation of bank branch management was gender neutral. Third, although black individuals were underrepresented in managerial occupations, their presence in such careers wa s realistic. This assertion was supported by labor statistics, as well as research showing that individuals consid ered banking to be an occupation appropriate for both black and wh ite individuals (e.g., Ya rkin et al., 1982). After a brief description of the banks formal mentoring program, the vignette presented evaluations of the mentoring pr ogram purportedly provided by one mentorprotg pair. These evaluations consisted of separate narratives from the mentor and protg describing their experi ences in the program. The manipulation of mentor and protg race was embedded in these fictiti ous mentoring program evaluations by way of the mentors and protgs names. Prev ious research has used this method of manipulating a target persons race through hi s/her name (e.g., Bertrand & Mullainathan, 2004; Cuddy, Rock, & Norton, 2007). For the cu rrent study, black and white names
32 were chosen from lists provided by Bertra nd and Mullainathan and by Levitt and Dubner (2005). These lists identified names that were distinctively white and distinctively black based on frequency data from birth certif icates. Bertrand and Mullainathan also conducted a survey to confirm that individua ls perceived the names on their lists as distinctively white or black. To test whether individuals would assi gn the intended races to the mentors and protgs depicted in the cu rrent studys vignette, a pilot study was conducted. The pilot sample consisted of 68 white undergraduate st udents taking psychology classes at a large southeastern university who recei ved extra credit for their pa rticipation. White and black names were chosen from the lists of Be rtrand and Mullainathan (2004) and Levitt and Dubner (2005), and different forms of th e survey packets representing the four experimental conditions were cr eated. The packets were randomly distributed to the pilot participants. Participants were asked to complete the entire packet, which included items asking them to indicate the race of the mentor and protg depicted in the vignette. Analyses showed that the percentage of partic ipants that correctly identified the races of each of the chosen names ranged from 87.9% to 95.8%, which was deemed an acceptable level of correct identification. Following the mentor and protg eval uations of the mentoring program, the survey packet included an assessment of the protgs job performance purportedly provided by the protgs supervisor. The vi gnette explained that this assessment was a part of the mentoring program evaluation. The assessment was written in narrative form and was designed to portray the protg as an above average performer. An above average level of protg job performance was chosen based on previous research
33 suggesting that the level of performance must be sufficiently high in order to observe stereotypes operating (e.g., Greenhaus & Pa rasuraman, 1993). To test whether the assessment depicted an above average leve l of performance, a small pilot study was conducted. Five psychology gr aduate students were asked to read the fictitious assessment and rate the level of performan ce on a 7-point Likert-type scale ranging from 1 (significantly below average) to 7 (signifi cantly above average). The mean of their responses was 6.2, with four of the five partic ipants selecting the response option labeled above average. These results supported the assertion that the assessment depicted the desired level of performance. Measures Causal Attributions A scale was developed to measure partic ipants attributions for the performance of the protg portrayed in the vignette. Pa rticipants were asked to indicate the extent that each of the following fact ors contributed to the performa nce of the protg: protg ability, protg effort, mentor help, protg luck, and task di fficulty. The scale consisted of three items per causal factor, and responses were made on a 5-point Likert-type scale ranging from 1 (not at all) to 5 (great exte nt). A sample item from the protg ability dimension is The mentees high ability. A sample item from the protg effort dimension is The mentees high effort. A sa mple item from the mentor help dimension is The help provided by the mentor. A sa mple item from the protg luck dimension is The mentees good luck. A sample item from the task difficulty dimension is The mentees job is easy. Scores on each causa l factor were calculated by averaging item
34 responses. The coefficient alphas for each of the causal factor scales ranged from .76 to .95. All scales used in the present stu dy are provided in Appendices B through F. Career Advancement Potential A 3-item scale was developed to measure participants ratings of the protgs career advancement potential. Responses were made on 5-point Likert-type scales with anchors specific to the items. A sample it em includes How woul d you rate the potential of the mentee for advancing to positions of gr eater responsibility in the company? Scale scores were calculated by aver aging item responses, with high er scores indicating higher ratings of protg career adva ncement potential. The coefficient alpha for this scale was .83. Reward Recommendations The extent that participants would recommend the protg and the mentor for various organizational rewards was measured using a modified version of the scale developed by Allen and Rush (1998). Using a 5-point Likert-type scale ranging from 1 (would definitely not recommend) to 5 (woul d recommend with confidence and without reservation), participants indicated the extent that they would recommend the protg for the following rewards: salary increase, promotion, high profile project, public recognition (e.g., company award), and fast-t rack developmental program. The scale measuring mentor reward recommendations wa s identical to the pr otg version, except the reward fast-track developmental progr am was replaced with opportunities for executive development. Item responses were averaged to obtain se parate overall reward recommendation scores for the protg and the mentor. Allen and Rush reported a
35 reliability of = .90 for their scale. In the curr ent study, the coefficient alphas for the protg and mentor reward recommendati ons scales were .78 and .83, respectively. Manipulation Check The effectiveness of the racial manipul ation was assessed by asking participants to indicate to which racial groups the me ntor and protg belonged. The response options were White/Caucasian and Black/African American. In order to disguise the true purpose of the study, these two items were locate d toward the end of the survey and were embedded in other items asking the particip ants to indicate to which gender and age groups the mentor and protg belonged. Participant Demographics To determine whether participants met th e inclusion criteria, they were asked to provide their race, their curre nt employment status, the num ber of hours they worked per week, and the length of time they had been employed in their current job. Other demographic data was also collected, incl uding information concerning participants gender, age, education, job title, work industry, experience in workplace mentoring relationships, and supervisory experience. When responding to questions regarding their experience in workplace mentoring relationshi ps, participants were provided with the definition of formal vs. informal ment oring used by Ragins and Cotton (1999).
36 Chapter Three Results Preliminary Analyses Manipulation Check Of the 308 participants that met th e studys inclusion criteria (i.e., white individuals who worked at least 20 hours per week and had been employed in their current job for at least 6 months), a total of 281 indicated the correct gender (male) for both the mentor and protg depicted in the vignette. Data from th ese 281 participants were used to examine the effectiveness of the racial manipulation by calculating the percentage of participants w ho correctly identified the intend ed races of the mentor and protg in the vignette. Table 5 shows these percentages by target name. For purposes of comparison, Table 5 also shows results obtained from the pilot study. As data from the primary study came in, it became apparent that the percentage of participants correctly identifyi ng the race of Darnell was much lower than the percentage obtained during the pilot study (62.5% vs. 87.9%, respectively). Therefore, the decision was made to replace the name Darnell with the name DeAndre for the remainder of the data collection. However, this change only re sulted in a minor incr ease in the percentage of participants correctly identifying th e race of the target, from 62.5% to 66.7%. In order to be included in subsequent analyses, participants had to correctly identify the intended races of both the mentor and the protg depicted in the vignette.
37 Table 5 Percentage of Participants Identifyi ng the Intended Race by Target Name Name Intended race Role N Percent correct Primary study Greg white mentor 128 85.9 Brad white protg 140 90.7 Darnell black mentor 24 62.5 DeAndre black mentor 129 66.7 DeShawn black protg 141 80.9 Pilot study Greg white mentor 35 94.3 Brad white protg 20 90.0 Darnell black mentor 33 87.9 DeAndrea black mentor DeShawn black protg 24 95.8 Note Percent correct represents the percentage of participants that correctly identified the race of the target. N = total number of participants who responded to the item asking them to identify the race of the target as either White /Caucasian or Black/African American. aThe name DeAndre was not te sted during the pilot study.
38 Table 6 presents the percentage of participants who correctly identified the intended races of both targets, broken down by experimental condition. Results from the pilot study are also included for purposes of comparison. Thes e data indicated that the percentage of participants who correctly identified th e race of both targets was lower in the experimental conditions where the mentor was black. The data also revealed differences between the percentages obtai ned during the pilot study a nd those obtained during the primary study. Possible reasons for these differe nces will be offered in the Discussion. In total, 194 participants correc tly identified the races of both targets, and their data were included in subsequent analyses. Factor Analysis of Caus al Attributions Measure A principal axis factor analysis with oblique rotation was conducted to determine the number of dimensions underlying particip ant responses to the causal attributions items. Prior to performing the analysis, da ta screening procedures were conducted. Inspecting the correlations among the items a nd plotting a sample of item pairs revealed the presence of linear relationships among th e items, thus supporting the use of factor analysis. Examining the univariate normality of each item revealed a lack of normality. Specifically, items designed to assess attrib utions to protg abil ity and mentor help exhibited slight negative skew, with values ranging from -.14 to -.48; items designed to assess attributions to protg e ffort exhibited a slightly larg er negative skew, with values ranging from -.71 to -.85; and items designed to assess attributions to protg luck and task difficulty exhibited relatively large pos itive skew, with values ranging from .74 to 1.69. Although screening procedures revealed the presence of multivariate outliers, the decision was made to include these outliers in the analyses, as all item responses fell
39 Table 6 Percentage of Participants Identifying the Intended Races of Both Targets by Experimental Condition Condition N Percent correct Primary study White mentor-white protg 61 83.6 Black mentor-black protg 74 63.5 White mentor-black protg 67 70.1 Black mentor-white protg 79 62.0 Pilot study White mentor-white protg 12 75.0 Black mentor-black protg 12 83.3 White mentor-black protg 12 91.7 Black mentor-white protg 8 100.0 Note Percent correct represents the percentage of participants that correctly identified the races of both targets. N = total number of pa rticipants who responded to the items asking them to identify the race of the targets as either White/Caucasian or Black/African American.
40 within the possible range of values. Because f actor analysis is a descriptive rather than inferential procedure, it is forgiving toward non-normality, and the decision was made to proceed with the analysis. Based on the Kaiser rule, results of the f actor analysis suggested that three factors may be worth interpreting. Howe ver, visual inspection of the scree plot and results of a parallel analysis indicated a five-factor solution. Thus, the five-factor solution was interpreted. Item assignment to factors was based on items having pattern coefficients greater than or equal to .30. Table 7 presents the five factors, 15 items, and rotated pattern coefficients, as well as item means a nd standard deviations. Results supported the five a priori dimensions of th e causal attributions measure, with items contributing to the expected factors. Final communality estimates ranged from .46 to .91, indicating that the individual items were repres ented from a moderate to hi gh extent by the five-factor solution. Correlations among the five factors ranged in absolute value from .05 to .60, with the largest correlations between the Task Difficulty and Protg Luck factors ( r = .60) and between the Protg Effort and Protg Ability factors ( r = .57). Checking MANOVA Assumptions Prior to testing the hypotheses using MANOVA, the data were screened for violations of the assumptions of indepe ndence of observations, multivariate normality, and homogeneity of covariance matrices. The study was designed such that the assumption of independence was met by following the studys procedures. The data were examined for univariate and multivariate no rmality by examining plots, skewness and kurtosis values, and potential outliers by group. These procedures revealed a lack of univariate normality. Specificall y, the distributions for attri butions to protg luck and
41 Table 7 Causal Attributions Factors, Items, and Pattern Coefficients Factor/item Pattern coefficients M SD 1 2 3 4 5 Factor 1: Task Difficulty 15. The mentees job duties are not very difficult. .93 .02 .01 .00 -.02 1.57 .88 14. The tasks the mentee is required to perform are easy. .92 .04 -.04 -.01 .01 1.58 .88 13. The mentees job is easy. .81 .13 .01 -.01 .01 1.51 .85 Factor 2: Protg Luck 10. The mentees good luck. .00 .92 -.01 .01 .01 1.80 .98 11. The mentees good fortune. .05 .90 -.01 -.04 -.02 1.80 .96 12. The mentee was in the right place at the right time. .14 .72 .06 .03 .01 1.98 1.00 Factor 3: Mentor Help 9. The mentors valuable guidance. -.02 -.07 .86 .00 .01 3.80 .79 7. The help provided by the mentor. -.03 .01 .85 .02 -.02 3.72 .81 8. The mentors support. -.03 .09 .81 .00 .01 3.76 .75 Factor 4: Protg Effort 5. The mentees hard work. .01 .06 -.04 .87 .02 4.01 .77 4. The mentees high effort. -.06 -.04 -.02 .85 -.01 4.01 .74 6. The mentee s high motivation. .02 -.03 .12 .69 .10 4.00 .76 Factor 5: Protg Ability 2. The mentee has the skills needed. -.10 -.01 -.02 -.07 .74 3.69 .70 1. The mentees high ability. .04 .10 -.08 .14 .65 3.64 .71 3. The mentees high level of competence. .07 -.08 .10 .07 .61 3.78 .65 Eigenvalue 4.65 3.26 1.51 .82 .54 Percent variance 46.2 32.4 15.0 8.1 5.4 Note Results are based on N = 194. Item numbers are indicated to the left of each item.
42 attributions to task difficulty were positively skewed, with skewness values ranging from .48 to 1.78. Screening procedures also reve aled the presence of multivariate outliers; however, these outliers were included in subs equent analyses because all item responses fell within the possible range of values. Given the robustness of MANOVA against violations of normality, the decision was made to proceed with the analysis. A Boxs M test was conducted to examine the assumption of homogeneity of covariance matrices. The test was significant ( 2 = 148.32, p = .006), suggesting that the homogeneity assumption may have been violated. Howeve r, given that the groups were relatively close in size, such a violati on would have only minimal effects on the error rate. Thus, it seemed reasonable to proceed with the analysis. Hypothesis Testing Table 8 presents the means, standard deviations, coefficient alphas, and intercorrelations among the study dependent variables for the overall sample ( N = 194). Tables 9 through 11 present the means and st andard deviations by group (i.e., mentor race and protg race) and by subgroup (i.e., me ntor x protg race). Tables 12 through 14 present the intercorrelations among the de pendent variables by group (i.e., mentor race and protg race) and by subgroup (i.e., mentor x protg race). A 2 x 2 factorial MANOVA was conducted to test hypotheses 1 through 7 and 12 through 14, which predicted main effects and in teractions for mentor and protg race on the studys dependent variables. Results of this multivariate test were not statistically significant for mentor race (Wilks = .96, F (8, 183) = .87, p = .55), protg race (Wilks = .94, F (8, 183) = 1.44, p = .18), or the interaction of mentor and protg race (Wilks = .96, F (8, 183) = .91, p = .51). Thus, these hypothes es were not supported, as
43 Table 8 Means, Standard Deviations, Alphas, and Inte rcorrelations Among Study Dependent Variables Variable 1 2 3 4 5 6 7 8 1. Protg ability attribution (.76) 2. Protg effort attribution .51** (.89) 3. Mentor help attribution .20** .38** (.89) 4. Protg luck attribution .04 -.11 -.03 (.92) 5. Task difficulty attribution -.11 -.21** -.14* .63** (.95) 6. Protg potential .41** .42** .29** -.06 -.21** (.83) 7. Protg rewards .32** .25** .23** -.04 -.20** .60** (.78) 8. Mentor rewards .24** .24** .36** -.10 -.21** .34** .57** (.83) M 3.70 4.01 3.76 1.86 1.56 3.81 3.91 3.82 SD .57 .68 .71 .91 .83 .60 .62 .68 Note Numbers in parentheses represent coefficient alphas. N = 194. p < .05. ** p < .01.
44 Table 9 Attributions, Evaluations of Potential, and Reward Recommendations as a Function of Mentor Race Variable White mentor Black mentor M SD M SD Protg ability attribution 3.69 .55 3.71 .59 Protg effort attribution 3.94 .70 4.07 .67 Mentor help attribution 3.74 .77 3.78 .66 Protg luck attribution 1.93 .90 1.79 .92 Task difficulty attribution 1.61 .87 1.50 .80 Protg potential 3.83 .61 3.80 .59 Protg rewards 3.85 .66 3.96 .58 Mentor rewards 3.80 .71 3.84 .66 Note N = 98 for white mentor group. N = 96 for black mentor group.
45 Table 10 Attributions, Evaluations of Potential, and Reward Recommendations as a Function of Protg Race Variable White protg Black protg M SD M SD Protg ability attribution 3.74 .61 3.67 .52 Protg effort attribution 4.01 .75 4.00 .61 Mentor help attribution 3.74 .74 3.78 .68 Protg luck attribution 1.75 .87 1.98 .94 Task difficulty attribution 1.42 .68 1.70 .95 Protg potential 3.90 .56 3.72 .63 Protg rewards 3.97 .59 3.83 .66 Mentor rewards 3.87 .64 3.77 .73 Note N = 100 for white protg group. N = 94 for black protg group.
46 Table 11 Attributions, Evaluations of Potential, and Reward Reco mmendations as a Function of Mentor and Protg Race Variable White mentor Black mentor White protg Black protg White protg Black protg M SD M SD M SD M SD Protg ability attribution 3.71 .61 3.67 .48 3.77 .61 3.66 .57 Protg effort attribution 3.93 .77 3.96 .62 4.10 .73 4.04 .60 Mentor help attribution 3.68 .79 3.80 .74 3.80 .69 3.77 .63 Protg luck attribution 1.72 .83 2.16 .92 1.78 .91 1.80 .94 Task difficulty attribution 1.46 .75 1.76 .97 1.37 .61 1.65 .94 Protg potential 3.87 .53 3.78 .69 3.94 .58 3.66 .57 Protg rewards 3.85 .63 3.86 .70 4.11 .52 3.81 .61 Mentor rewards 3.83 .61 3.76 .81 3.90 .67 3.78 .65 Note N = 51 for white mentor-white protg group. N = 47 for white mentor-black protg group. N = 49 for black mentor-white protg group. N = 47 for black mentor-black protg group.
47 Table 12 Intercorrelations Among Study Dependent Variables by Mentor Race Variable 1 2 3 4 5 6 7 8 1. Protg ability attribution .40** -.05 .05 .00 .39** .31** .07 2. Protg effort attribution .63** .33** -.15 -.15 .50** .24* .15 3. Mentor help attribution .43** .42** -.01 -.18 .33** .26** .18 4. Protg luck attribution .02 -.06 -.04 .61** -.11 -.06 -.18 5. Task difficulty attribution -.21* -.26** -.11 .64** -.20 -.20* -.30** 6. Protg potential .43** .36** .25* -.01 -.22* .54** .19 7. Protg rewards .34** .25* .21* -.01 -.19 .65** .54** 8. Mentor rewards .41** .32** .50** -.01 -.13 .46** .60** Note Correlations for the white mentor group app ear below the diagonal. Correlations for the black mentor group appear above the diagonal. N = 98 for the white mentor group. N = 96 for the black mentor group. p < .05. ** p < .01.
48 Table 13 Intercorrelations Among Study Depende nt Variables by Protg Race Variable 1 2 3 4 5 6 7 8 1. Protg ability attribution .57** .22* .07 -.09 .42** .29** .13 2. Protg effort attribution .47** .32** .01 -.07 .45** .23* .17 3. Mentor help attribution .19 .42** -.00 -.06 .45** .36** .45** 4. Protg luck attribution .02 -.22* -.06 .66** .01 -.04 -.14 5. Task difficulty attribution -.11 -.38** -.26** .58** -.19 -.19 -.20 6. Protg potential .40** .42** .15 -.09 -.18 .68** .37** 7. Protg rewards .35** .28** .13 -.01 -.18 .48** .57** 8. Mentor rewards .34** .31** .28** -.03 -.20 .28** .58** Note Correlations for the white protg group app ear below the diagonal. Correlations for the black protg group appear above the diagonal. N = 100 for the white protg group. N = 94 for the black protg group. p < .05. ** p < .01.
49 Table 14 Intercorrelations Among Study Dependent Variables by Mentor and Protg Race Variable 1 2 3 4 5 6 7 8 White mentor 1. Protg ability attribution .66** .25 .05 -.18 .50** .37* .39** 2. Protg effort attribution .61** .29* .16 -.04 .46** .25 .36* 3. Mentor help attribution .55** .50** -.01 .01 .40** .34* .61** 4. Protg luck attribution .02 -.26 -.11 .56** .03 .00 .03 5. Task difficulty attribution -.25 -.50** -.28 .72** -.18 -.15 -.06 6. Protg potential .40** .29* .11 -.01 -.26 .67** .45** 7. Protg rewards .33* .26 .09 -.03 -.24 .63** .60** 8. Mentor rewards .45** .30* .39** -.04 -.21 .48** .61** Black mentor 1. Protg ability attribution .51** .20 .09 -.02 .35* .22 -.13 2. Protg effort attribution .31* .37* -.11 -.10 .46** .21 -.08 3. Mentor help attribution -.25 .30* -.01 -.15 .52** .38** .22 4. Protg luck attribution .03 -.19 -.02 .76** -.06 -.10 -.33* 5. Task difficulty attribution .08 -.21 -.24 .44** -.23 -.23 -.38** 6. Protg potential .40** .55** .18 -.17 -.09 .69** .27 7. Protg rewards .38** .28 .15 -.02 -.06 .32* .52** 8. Mentor rewards .23 .32* .15 -.04 -.18 .10 .57** Note Correlations for white protg groups appear below the diagonal. Correlati ons for black protg groups appear above the diagonal. N = 51 for white mentor-white protg group. N = 47 for white mentor-black protg group. N = 49 for black mentor-white protg group. N = 47 for black mentor-black protg group. p < .05. ** p < .01.
50 there was no support for racial differences in the causal attributions, ratings of potential, or reward recommendations made by participants. Hypotheses 8 through 11 predicted re lationships among cau sal attributions, ratings of protg potential, and mentor and protg reward recommendations. These hypotheses were tested by exam ining zero-order correlations. In support of Hypothesis 8, there was a positive association between attrib utions to protg ability and ratings of protg career advancement potential ( r = .41, p < .01; see Table 8). Hypothesis 9 was supported, in that attributions to prot g ability were positively related to protg rewa rd recommendations ( r = .32, p < .01), and attributions to protg effort were positively related to protg reward recommendations ( r = .25, p < .01). Hypothesis 10 predicted that at tributions to protg ability would be associated with greater protg reward recommendations than w ould attributions to protg effort. To test this hypothesis, correlat ions between protg ability an d effort attributions and the individual rewards making up the protg rewa rd recommendations scale were examined (see Table 15). Whereas protg ability attri butions had significant positive relationships with four of the five rewards, protg effort attributions had significant positive relationships with all five of the rewards. However, an examination of the magnitude of the correlations revealed that protg abil ity attributions had hi gher correlations with three of the rewards (i.e., salary increase, promotion, and high profile project) than did protg effort attributions. To determine whether these differences were significant, the Hotelling-Williams test was conducted. Resu lts showed that the correlation between protg ability attributions and promotion was significantly larger than the correlation between protg effort attr ibutions and promotion ( t (191) = 2.66, p = .009). Results were
51 Table 15 Intercorrelations Be tween Attributions and Protg Rewards Variable 1 2 3 4 5 6 7 1. Protg ability attribution 2. Protg effort attribution .51** Protg reward variables 3. Salary increase .30** .17* 4. Promotion .33** .15* .58** 5. High profile project .30** .24** .40** .41** 6. Public recognition .17* .20** .31** .40** .47** 7. Fast-track developmental program .11 .17* .29** .33** .48** .50** M 3.70 4.01 4.23 3.89 3.91 3.72 3.79 SD .57 .68 .77 .79 .89 .96 .86 Note. N = 194. p < .05. ** p < .01.
52 not significant for salary incr ease or high profile project. Thus, Hypothesis 10 received mixed support, in that protg effort attributions were si gnificantly associated with a greater number of rewards than were protg ab ility attributions, but the magnitude of the correlation between protg ability attributi ons and promotion was larger than that between protg effort at tributions and promotion. Hypothesis 11 predicted a posit ive association between attr ibutions to mentor help and mentor reward recommendations. As s hown in Table 8, results supported this hypothesis, yielding a significant positive correlation ( r = .36, p < .01). However, results differed according to the race of the mentor (see Table 12), such that the association between attributions to mentor help and mentor reward recommendations was significant when the mentor was white ( r = .50, p < .01), but was not significant when the mentor was black ( r = .18, p > .05). A test of the equality of these two correlations confirmed that the correlation between mentor he lp attributions and mentor reward recommendations was different for th e white and black mentor groups ( z = 2.52, p = .006). Additional analyses were conducted exam ining mentor race as a moderator of the association between mentor help attributi ons and mentor reward recommendations and are presented in the Supplemental Analyses section of this paper. Supplemental Analyses Comparison of Means Across Target Names As explained previously, the decision wa s made during data co llection to replace the name Darnell with the name DeAndre in an effort to increase the percentage of participants correctly identif ying the race of the target. Thus, data collected using Darnell were combined with data collected using DeAndre in the final data set. A
53 MANOVA was conducted to determine whethe r the means on the set of dependent variables differed across these two names. Results were not significant (Wilks = .91, F (8, 87) = 1.12, p = .36), suggesting that the means did not differ. Post Hoc Power Analysis A post hoc estimation of power was conduc ted to explore the possible reasons for the nonsignificant results of the MANOVA pe rformed to test hypotheses 1 through 7 and 12 through 14. Results of the power anal ysis revealed poor power for detecting differences on the dependent variables, with power estimates of .40 and .64 for the main effects of mentor and protg race, resp ectively, and a power estimate of .42 for the interaction of mentor and protg race. On e reason for this poor power may have been inadequate sample size. An a priori pow er analysis had suggested a need for 84 participants per group, for a total sample size of 336, to achieve a power of .80. However, after excluding participants who di d not meet the studys inclusion criteria and participants who did not res pond correctly to the manipula tion, the current studys final sample size was only 194. Another reason for the observed poor power may have been small effect sizes. Multivariate 2 values were small, ranging from .037 to .059, which suggest that mentor and protg race accounted for little of the va riance in the set of dependent variables. Furthermore, when e ffect sizes were calculated for each of the hypothesized main effects of mentor and protg race on th e study dependent variables, the obtained values were small in size (Cohens d = .02 to .34). Mentor Race Moderator Analysis In order to examine mentor race as a m oderator of the associ ation between mentor help attributions and mentor reward recomm endations, hierarchical multiple regression
54 was used. Mentor race was c oded as a dummy variable (0 = white mentor, 1 = black mentor). The mentor help attributions vari able was standardized in order to aid in interpretation and reduce potential problems associated with multic ollinearity (Frazier, Tix, & Barron, 2004). Mentor help attributions and mentor race were entered in the first step of the equation. The inte raction term (the product of me ntor help attributions and mentor race) was entered in the second step. The dependent variable was mentor reward recommendations. Prior to interpreting the results of th e analysis, the data were examined to determine whether the assumptions of the multiple regression model had been met (Aguinis, 2004). Based on the correlation coefficient and a plot of the variables, the relationship between mentor help attribut ions and mentor reward recommendations appeared linear. Plots of the residuals revealed that the a ssumptions of homoscedasticity and normality had been met. Correlations among the predictors revealed less than complete multicollinearity. Results of Bartletts M test indicated that the homogeneity of error variance assumption had been met ( M = .28, p = .59). Furthermore, the error variance ratio was 1:1.11, which meets the rule of thumb derived by DeShon and Alexander (1996), providing a dditional evidence that hom ogeneity was met. Taken together, these results indicated that the assumptions of the multiple regression model were met. Results of the hierarchical multiple re gression are presented in Table 16. The addition of the interaction te rm resulted in a significant R2 change of .02, F (1, 190) = 4.47, p = .04, supporting the presence of the modera ting effect of mentor race. A graph of the interaction was created to show the re lationship between mentor help attributions
55 and mentor reward recommendations for white and black mentors (see Figure 1). To create this graph, low and high values of mentor help attri butions and mentor race were substituted into the final re gression equation, and the resu lting predicted values for mentor reward recommendations were pl otted. The low value for mentor help attributions was one standa rd deviation below the mean, and the high value was one standard deviation above the mean. The low value for mentor race was 0 and the high value was 1, based on the dummy coding used to represent white and black mentors, respectively. As shown in the graph, resu lts indicated that th e relationship between mentor help attributions a nd mentor reward recommendations was stronger for white mentors than for black mentors. An additional analysis was conducted to te st the significance of the slopes of the simple regression lines for each group. As explained by Frazier et al. (2004), when regression equations contain inte raction terms, the regression coefficient for the predictor represents the relation between the predictor and outcome when the moderator has a value of 0 (p. 125). Thus, in the original regression analysis, when the white mentor group was coded as 0, the coefficient for me ntor help attributions represented the relationship between mentor help attributi ons and mentor reward recommendations for the white mentor group. Results from this an alysis indicated a signi ficant positive slope for the white mentor group ( B = .33, p < .001). To determine whether the slope of the regression line for the black mentor group was significant, an additional regression analysis was conducted, in which the black mentor group was coded as 0 and the white mentor group was coded as 1. Results indicated that the slope for the black mentor group did not significantly differ from zero ( B = .13, p = .07). Thus, there was a significant
56 Table 16 Hierarchical Multiple Regression Resu lts for Mentor Reward Recommendations Step and variable B SE B 95% CI R2 Step 1 Mentor help attribution .33 .06 .21, .44 .48** Mentor race .03 .09 -.15, .21 .02 .13** Step 2 Interaction term -.20 .09 -.38, -.01 -.19* .02* Note White mentor coded 0, black mentor coded 1. B = unstandardized regression weights for the final equation. = standardized regression we ights for the final equation. CI = confidence interval. N = 194. p < .05. ** p < .001. 3 3.2 3.4 3.6 3.8 4 4.2 LowHigh Mentor Help AttributionsMentor Reward Recommenda t White mentor Black mentor Figure 1 Interaction of mentor race and mentor help attributions on mentor reward recommendations.
57 relationship between mentor help attributi ons and mentor reward recommendations when the mentor was white, but not when the mentor was black.
58 Chapter Four Discussion The purpose of the current study was to examine how the racial composition of a mentoring relationship influences three types of judgments made by individuals external to the relationship: (1) causal attribu tions formed to explain successful protg performance; (2) evaluations of protg car eer advancement potential; and (3) reward recommendations for the mentor and protg . Overall, resu lts do not support the hypothesized racial differences in these judgm ents. However, results do provide support for hypotheses concerning the associations among these judgments. Furthermore, findings suggest that mentor race may modera te one of these hypothe sized associations. Results are discussed in more deta il in the sections that follow. Racial Differences in Judgments: Hypotheses 1 through 7 and 12 through 14 Hypotheses 1 through 7 predicted racial di fferences in the causal attributions formed to explain successful protg performance, and Hypotheses 12 through 14 predicted racial differences in ratings of potential and reward recommendations. However, results do not support these hypothe ses, failing to find significant racial differences in the judgments under investigatio n. These results conflict with previous research, which has shown racial differe nces when using both experimental and nonexperimental designs (e.g., Greenhaus & Parasuraman, 1993; Jackson et al., 1993; Yarkin et al., 1982).
59 One possible explanation for the current studys findings is poor statistical power for detecting racial differences. A post hoc power analysis revealed poor power, and although one reason for this poor power may have been inadequate sample size, another reason may have been small effect sizes. In fact, the effect sizes found in the current study are smaller than those f ound in previous experimental research (Jackson et al., 1993; Yarkin et al., 1982). There are a few possible explanations for why the racial effects found in the current study are smaller than those obtained in previous research. First, the racial manipulation used in the current study may have been less effective than the racial manipulations used in previous studies. Fo r example, whereas 100% of the participants in Yarkin et al.s (1982) study and 98% of the partic ipants in Jackson et al.s (1993) study correctly identified the race of the target, only 69% of the participants in the current study correctly identified the race of both targets. As a result, a la rger percentage of participants in the current study had to be excluded from the an alyses. It is possible that those excluded differed from those included in important ways, which may have affected the results and led to the smaller effect si zes found in the current study. Additionally, the larger percentage of participants that incorr ectly identified the race of the targets in the current study suggests that th e racial manipulation used ma y have been more ambiguous than those used in other studies. For exam ple, whereas the current study used target name to manipulate race, Yarkin et al. manipulated target race in a resume via the targets undergraduate institution (Howard University vs. American University) and community activities (NAACP vs. Chamber of Commerce). Jackson et al. provided participants with a college application that explicitly indicated that the target was either
60 White/Caucasian or Black/African American , and included either representative to the student union or representative to the bl ack student union in a list of the targets activities and interest s. If the racial manipulatio n in the current study was more ambiguous, it is possible that some of the part icipants included in th e analyses may have been uncertain of the race of the targets, but guessed the correct response to the manipulation check items. If this occurred, then it may have contributed to the smaller effect sizes. Another possibility is that the racial manipulation used in the current study may have been less salient than those used in pr evious studies. If th is was the case, the manipulation may not have been strong enough to activate racial stereotypes, resulting in the smaller effect sizes. Alternatively, it may be that the different racial manipulations carry with them additional information that in fluences the operation of individuals racial stereotypes. For example, Yark in et al. (1982) conveyed to pa rticipants that a target was black by indicating that the target had atte nded a historically black university (i.e., Howard University) and was involved in the NAACP. It is possible that information such as this may activate more negative bl ack stereotypes in some white individuals. Additionally, this information may convey to par ticipants that the black target identifies closely with the black community. Thus, part icipants may be less likely to attempt to distinguish the target from ot her black individuals. Accord ing to Pettigrew and Martin (1987), one way that white observers may e xplain successful black performance is by distinguishing the successful black from other black individuals as an exceptional case. When this occurs, the white observers may exaggerate the black individuals positive qualities. Therefore, if white participants are less likely to differentiate the black target as
61 an exceptional case due to the targets appa rent ties to the black community, they may provide less favorable judgment s about the target. This may then result in greater differences between the judgments made about wh ite versus black targets. In contrast to the manipulation used by Yarkin et al., mani pulating target race solely by way of the targets name may not convey as much additional information to participants. As a result, participants may be more likely to di stinguish a successful black from other black individuals as an exceptional case and exaggerate the successful blacks positive qualities. This may result in smaller diffe rences between partic ipants judgments of white versus black targets. Taken togeth er, this argument provides another potential explanation for the smaller racial e ffects obtained in the current study. Aside from the racial manipulation, anot her possible reason that the effect sizes for the current study are smaller than those f ound in previous research may be that the performance level of the protg was not high enough to effectively activate racial stereotypes. Previous resear ch has suggested that, in orde r to observe the operation of stereotypes, the level of performance must be sufficiently high such that it deviates from expectations for performance (e.g., Greenha us & Parasuraman, 1993). Although a small pilot study indicated that the perceived level of protg performance in the current study was above average, it is possible that this level of performance is not high enough to effectively activate stereotypes. If this is the case, it may contri bute to the explanation for why the racial effects observed in the current study are smaller than those found in previous research. Another possible explanati on for the absence of signifi cant racial differences in judgments is that participants may have elevat ed their judgments of th e black targets, thus
62 minimizing the difference between the white a nd black targets. There are two reasons that participants may have done this. Th e first reason stems from the exceptional case attribution proposed by Pettigrew and Martin (1987), which was referred to earlier. According to Pettigrew and Martin, white observers may view a successful black performer as an exceptional case and attempt to differentiate the successful black from other black individuals by exaggerating the su ccessful blacks positive qualities. If this happened in the current study, participants ma y have provided more favorable ratings to the black mentors and protgs, thereby d ecreasing the gap between white and black targets. The second reason that participants may have elevated their judgments of the black targets is that they were attempting to appear as though they did not hold negative racial stereotypes. In other words, their responses may have been influenced by social desirability. However, given that the true purpose of the study was disguised, and that participants were assured anonymity, the possibi lity of this occurri ng should have been minimized. A final explanation for the lack of signi ficant racial differences in the current study is that participants do not hold racial stereotypes, and any mean differences are due to chance. However, given the current study s findings concerning the moderating effect of mentor race, this explana tion does not seem likely, as j udgments do not seem to have been totally unaffected by the race of the target. Associations Among Judgments : Hypotheses 8 through 11 The remaining hypotheses predicted associ ations among attributions for protg performance, evaluations of protg pote ntial, and mentor and protg reward
63 recommendations. As predicted in Hypot hesis 8, there was a positive relationship between attributions to pr otg ability and ratings of protg career advancement potential. This finding agrees with previ ous research that has demonstrated an association between ability attributions a nd ratings of employee career potential (e.g., Greenhaus & Parasuraman, 1993; Heilman & Guzzo, 1978). Although not hypothesized, attributions to protg effort were also positively associated with protg career advancement potential in the current study, and the magnitude of the correlation was similar in size to the correlation between ability attributi ons and potential ( r = .42 for effort attributions; r = .41 for ability attributions). These results are similar to those obtained by Greenhaus and Parasuraman, who found career advancement prospects to be positively correlated with both ability attributions ( r = .20, p < .01) and effort attributions ( r = .24, p < .01). In support of Hypothesis 9, both attributions to protg ability and attributions to protg effort were positivel y related to protg reward recommendations. These results fall in line with previous research showi ng associations between ability and effort attributions and organizational rewards (A llen et al., 1994; Heilman & Guzzo, 1978). However, results in the current study were mixed with respect to Hypothesis 10, which predicted that attributions to protg ability would be associated with greater protg reward recommendations than would attributi ons to protg effort. In a study conducted by Allen et al., attributions to ability were significantly related to six organizational rewards, whereas attributions to effort were significantly related to only two of the six rewards. By contrast, in the current study, attributions to abilit y were significantly related to four of five organizational rewa rds, whereas attributions to effort were
64 significantly related to all five of the rewards. Thus results from the current study may seem to suggest that effort a ttributions are associ ated with greater organizational rewards than are ability attributions. However, wh en the magnitudes of the associations are compared, statistical analyses show that the association between ability attributions and promotion is significantly greater than the a ssociation between effort attributions and promotion. This result is cons istent with previous research that has found that raters judge promotions as more appropriate for ability-based success than for effort-based success (Heilman & Guzzo, 1978; Pazy, 1986). Therefore, although effort attributions may be associated with a greater number of the rewards included in the current study, ability attributions are mo re strongly associated with promotion, which may be considered one of the highest organizational rewards. It is interesting to note how career advancement potential and promotion recommendations differed in their relationships with attributions to ability and effort. Specifically, whereas ratings of career advan cement potential were similarly related to both ability and effort attributions, prom otion recommendations were more strongly related to ability attributions than to effort attributions Given that both of these constructs involve assessing whether an individual should move into a higher-level position, it seems reasonable to expect that th ey would show similar associations with ability and effort attributions. Furthe rmore, the theoretical reasoning behind both constructs associations with attributions is si milar. As discussed ear lier, the attributions formed to explain an employees performance can influence an obser vers expectations for the employees future performance (Green & Mitchell, 1979; Weiner et al., 1972). When employee performance is attributed to st able causes (e.g., ability), it is perceived as
65 likely to continue in a similar manner in the future. When performance is attributed to unstable causes (e.g., effort), it is more difficult to predict future performance. Thus, an observer may be more confident providing high er ratings of career advancement potential and promotion recommendations to an em ployee whose successful performance is attributed to ability rather than effort. As a result, it seems reasona ble to conclude that both ratings of advancement potential and promotion recommendations would be more strongly associated with ability attributions than with effort attributions. However, in the current study, this only held true for promotion recommendations. One possible explanation for these findings is the di ssimilarity of the constructs operational definitions. Specifically, the wording used in the promotion recommendation item gave respondents a more active role in the d ecision making process and implied a more immediate change in the employees position, comp ared to the wording used in the career advancement potential items. Perhaps in situ ations such as these, individuals make greater distinctions amongst the attributions they use to make decisions. Further research is needed to explor e these possibilities. Regarding Hypothesis 11, results supported the prediction that there is a positive association between attributi ons of protg success to the mentors help and mentor reward recommendations. This finding is consis tent with research and theory suggesting that organizational decision makers may recognize and reward mentors for their contributions to the organization (Allen et al., 1997; Ramaswami & Dreher, 2007). However, further examination of the data revealed that, although there was a positive association between attributions to mentor help and mentor reward recommendations when the mentor was white, there was no such association when the mentor was black.
66 Thus, whereas white mentors seem to have b een rewarded in part according to their contributions to their protgs success, black mentors do not seem to have been rewarded according to their contributions. Interestingly, there were no differences between black and white mentors in the mean ra tings of attributions to mentor help or mentor reward recommendations. However, given the differences in the correlations, it appears that the cognitive processes used by respondents to make these ratings may have differed. More specifica lly, although respondents may have based their reward recommendations for the white mentor in pa rt on the mentors contributions to the protgs success, they may have based th eir reward recommendations for the black mentor on some other factor(s). Although there are many factors that ma y have influenced respondents reward recommendations for the black mentor, one po ssibility is based on the exceptional case phenomenon described by Pettigrew and Martin (1987). If some of the respondents regarded the black mentor as an excepti onal case, they may have exaggerated the mentors positive qualities in order to differentiate him from other black individuals. This in turn may have influenced the re spondents reward recommendations for the mentor. It is also possible that some re spondents may have believed it to be unusual to see a black individual in such a position of power. They may have reasoned that the black mentor must have excelled in his pe rformance to overcome obstacles and achieve his current position. Such assumptions may then have influenced their reward recommendations for the mentor. It is in teresting to note that the percentage of participants that incorrectly identified the ra ce of a target was greatest in the case of the black mentor. Although there are several r easons why this may have occurred, one
67 explanation may be that part icipants found it unusual for a me ntor to be black, and thus failed to identify him as such. Perhaps those who correctly identified the mentor as black also believed it to be unusual, but rather than assuming the mentor must be white, they made other assumptions about the mentor, such as those described above Clearly, this is only one possible explanation and additional re search is needed to understand the current studys findings concerning the black mentor and the moderating role of mentor race. Effectiveness of the Racial Manipulation In the current study, the race of the ment or and of the protg depicted in the vignette was manipulated by way of their names. The effec tiveness of this manipulation was assessed by calculating the pe rcentage of participants that correctly identified the intended races of the targets. Results of a pilot test indicated that an acceptable percentage of participants correctly identifie d the races of the targets. However, in almost all cases, the percent correct obtained during the pr imary study data collection was lower than that obtained during the pilot test. One possible reason for this difference between the pilot study and primary study may be the differences between the samples used in each study. The pilot study sample consisted of white undergraduate students taking a psychology course at a large southeastern university. The primary study sample, on the other hand, was composed of white, employed individuals fr om five different sources, with the majority being employees at a national engineering consul ting firm. However, the second largest contributor to the primary sample was quite similar to the pilot sample, consisting of undergraduate students taking ps ychology courses at a large s outheastern university. The students in the primary sample were very simila r to those in the pilot sample in terms of
68 their demographic characteristics, includi ng gender, age, education, and work-related experience. In contrast, the employees from the e ngineering consulting firm differed in many respects from the students in the pilot and primary samples. For example, the engineering firm employees were majority male, were older, had higher levels of education, and had more work-related experi ence. Whereas 36.5% of the engineering firm employees who participated in the primary study were excluded from the final sample because they incorrectly identified the intended race of the targets, only 21.1% of the undergraduate students who participated in the primary study were excluded. Thus, the percentage of students in the primary samp le that correctly identified the race of the targets was closer to that obtained during the pilot study, although still not as high. Given the similarities between the student s in the pilot and primary samples, it may not be very surprising that the percen tages correctly identif ying the races of the targets were closer in value for these two groups in comparison to the engineering firm employees. However, the question remains of why a greater per centage of students correctly identified the targets races compar ed to the engineering firm employees. In addition to the demographic characteristic s already mentioned, there were other differences between the student s and the engineering firm em ployees that may shed some light on this question. For example, student s in the primary and pilot samples were probably more familiar with psychological expe riments compared to the engineering firm employees. Furthermore, some of these st udents had probably participated in other psychological experiments prior to the current study. Perhaps this familiarity influenced
69 the students responses to the manipulation ch eck items, especially if they had previous experience in research examining racial issues. It is also possible that the students and engineering fi rm employees had different degrees of exposure to black individuals, wh ich then affected th eir responses to the manipulation check items. According to the U. S. Bureau of Labor Statistics (2008), only 5.1% of individuals employed in architecture and engineering occupa tions are black or African American. By contrast, 11.7% of the st udents enrolled at th e large southeastern university are black (Office of Decision Support, 2008). These statistics sugge st that the students who participated in the current study may have had more exposure to black individuals than the employees of the engi neering consulting firm. This may be one reason why a larger perc entage of the engineering firm employees incorrectly identified the race of the targets, particularly in the case of the black mentor. If the engineering firm employees had less exposure to black in dividuals, they may be less familiar with names that, according to birth frequency data, are distinctively black. Furthermore, the employees may also see fewer examples of bl ack individuals in pos itions of power, and thus find it more unusual for a mentor to be bl ack. Either of these scenarios may explain why a greater percentage of the engineering firm employees incorrectly responded to the manipulation check items. In addition to the differences in the samp les used in the pilot and primary studies, there were also differences in the data coll ection procedures. Pilo t data were collected from students in a classroom setting using pape r versions of the study survey materials. Primary study data were collected primarily th rough an online survey. It is possible that
70 these different procedures may have produced different demand characteristics, which then influenced participants respon ses to the manipulation check items. Given the findings regarding the effectiven ess of the racial manipulation used in the current study, it is important to consider the implications of these findings and how the racial manipulation may have influenced the results of the study. One concern may be the loss of data that resulted from ex cluding a higher-than-des ired percentage of participants from the final sample due to their incorrect responses to the racial manipulation check items. It is possible that those excluded from the final sample differed in some way from those included, re sulting in an unrepresentative sample. Although a comparison of the demographic char acteristics of those included and those excluded did not reveal any c onsistent differences, it is st ill possible that the groups differed on other important variables, such as their exposure to black individuals, their familiarity with distinctively black names, or their expectations for black performance. Thus, the results obtained in the current study may not generalize to the excluded individuals or others having characteristics similar to those excluded. Additionally, the results of the study should be interpreted with this in mind, as excluding these individuals may have influenced the results. For example, the finding that the racial effects on the set of dependent variables were smaller than t hose obtained in previous research may be partially a result of excluding participants who differed on important variables. In the future, researchers should use less ambiguous ra cial manipulations in order to reduce the percentage of participants that have to be excluded. Another concern regarding the racial manipulation may be that the ambiguity of the manipulation resulted in participants responding randomly to the manipulation check
71 items. However, there is some evidence that suggests that participants were not simply responding at random to these items. First, the percentage of pa rticipants correctly identifying the targets race was above 80% for all of the names, except the names used for the black mentor. Second, when the correl ation between mentor he lp attributions and mentor reward recommendations was comput ed using responses of participants who incorrectly identified the black mentor as white, the correlation was significant ( r = .42, p < .01). As reported in the Results section, this correlation was significant for participants who correctly identified the white ment or as white, but was not significant for participants who correctly identified the bl ack mentor as black. Thus, the results for participants who incorrectly identified the black mentor as white were similar to the results for participants who correctly iden tified the white mentor as white. These findings are consistent with what might be expected if participants who incorrectly identified the black mentor as white really thought the mentor was white, rather than if they were responding at random to the manipulation check item. A third concern is that not only do the results of the manipulation check suggest that the racial manipulation was too ambi guous, but they may also suggest that the manipulation was too weak. As discussed earlier, a weak manipulation may be one explanation for why the racial effects observed in the curr ent study were not significant and were smaller than those obs erved in previous research. Again, future research should incorporate a less ambiguous and more salien t racial manipulation to increase the likelihood that participants will correctly identify the race of the targets. A fourth concern regarding the racial manipulation used in the current study stems not from the percentage of par ticipants that correctly identifi ed the race of the targets, but
72 rather from the design of the study. Specifical ly, for practical reasons and ease of survey administration, only one name was used for each target (e.g., the name Brad was used for the white protg target, the name Greg wa s used for the white mentor target, etc.). The only exception was in the case of the black mentor, where the name Darnell was replaced with the name DeAndre during da ta collection, resulting in two names being used for the black mentor target. Alt hough using this design was more practical, a potential problem is that it may have in troduced a confound, in which participants responses may have been influenced not only by the intended race of the target, but also by the specific name chosen for the target. However, there is some evidence suggesting that participants were respondi ng to the race of the targets rather than to their names. First, when examining the means on the se t of dependent variables for Darnell and DeAndre, there appeared to be no differences suggesting that par ticipant responses did not vary according to the name of the black mentor. Second, the finding that the correlation between mentor help attributions and mentor reward recommendations was significant for participants who identified the black mentor as white, but was not significant for participants w ho identified the black mentor as black suggests that participants may have been responding to the pe rceived race of the target rather than to the targets name. If participants had b een responding to the targets name, it seems reasonable to expect that these correlationa l results would not have differed. Despite these arguments, the possibility of confounding target race with target name cannot be totally dismissed in the current study, and future research sh ould consider balancing the target names or using multiple names for each target. For example, the name Brad could be used for the white protg target in some instances, and for the white mentor
73 target in other instances. This would incr ease confidence that any racial differences observed are due to target race rather than target name. Taken together, results of the manipula tion check analysis suggest that future research should use a less ambiguous racial manipulation. For example, researchers could use a picture of the targets depicted in the vignette or provide additional racerelated information about the targets, as ha s been done in previous research. Using a more salient and less ambiguous racial mani pulation will allow for more confidence in the results of the study. On the other hand, so me of the results of the current studys racial manipulation are interes ting and lead to questions that may be worth pursuing. For instance, why was the percentage of participan ts who correctly identified the race of the black mentor lower than for the other targets? Was this finding simply a function of the names chosen for the black mentor, or does it reflect a stereotype held by some people about black individuals being in positions of power? It is important to address such questions, as the answers may have important consequences for black individuals in the workplace. Study Limitations and Directi ons for Future Research In addition to the study limitations associat ed with the effectiveness of the racial manipulation, there are other limita tions to the current study that should be noted. First, results of a post hoc power analysis revealed poor power for detecting differences on the dependent variables. This may be one reas on for the lack of support for the hypothesized racial differences. A second limitation is that the design of the study did not allow for testing the causal direction of some of the relationships under investig ation. Specifica lly, although it
74 is implied that individuals causal attribu tions influence their judgments of career advancement potential and reward recommendations this was not actually tested in the current study. However, theory and previ ous experimental research have provided support for this argument (e.g., Heilman & Guzzo, 1978). A third possible limitation to the study is that participants were asked to judge a mentor and protg depicted in a written vi gnette, as opposed to an actual mentor and protg in an organizational se tting. Although this type of design provides more control over extraneous variables, it may also limit th e generalizability of the findings. However, steps were taken to make the situation more realistic. For example, study materials were presented in the form of mentor and protg evaluations of a formal mentoring program. Additionally, all of the participants were currently employed and the majority of participants had supervisory experience, sugges ting that they were familiar with the kinds of judgments examined in this study. Future research should invest igate how individuals evaluate actual mentor-prot g dyads in the workplace. A fourth potential limitation to the curre nt study is that the vignette depicted a mentor and protg in one specific occ upation bank branch management. This occupation was chosen on the basis of several criteria, however, it is possible that the results of the study may vary depending on the occupation chosen. For example, participant responses may differ depending on th e extent to which black individuals are represented in the chosen occupation, or the extent to which indivi duals stereotypes of blacks fit with their conceptual ization of the occupa tion. Therefore, fu ture research may examine judgments of homogeneous and diversified mentorships in different occupations.
75 In addition to the suggestions for future research already prov ided, there are other directions that may be wort h pursuing. For example, alt hough the current study focused on the evaluations formed by white observers, it would be interesting to examine the evaluations formed by black individuals and members of other racial minority groups. Further, the current study could be expande d upon by examining not ju st perceptions of mentoring partners who are white or black, but also perceptions of mentoring partners who are members of other racial and ethnic groups. Aside from race, there are other characteristics of the mentoring dyad that may influence observers evaluations of the mentor and protg, such as gender or age. Researchers should consider examining the influence of these characteristics. Fina lly, although the current study investigated judgments when the protg was successful, it would be interesting to investigate observers judgments when the protg is unsuccessful. Conclusions The aim of the current study was to contri bute to the limited amount of research on racial diversity and workplace mentor ing by examining how mentorship racial composition influences observers evaluations of the mentor and protg. Because outside observers can influence the de velopment and outcomes of a mentoring relationship, this research is particularly relevant to the question of whether racial minorities receive the same benefits from mentoring as do whites. Although results did not support the hypothesized racial differen ces in judgments, an interesting finding emerged concerning the association between attr ibutions to the mentors help and reward recommendations for the mentor. Specifica lly, results suggested that, whereas white mentors may be rewarded in part according to their contributions as a mentor, black
76 mentors may not be so rewarded. However, given the methodological limitations of the current study, additional resear ch is needed to explore th is finding and to determine whether black individuals receive the same benefits as do white individuals for their service as mentors. Such research is impor tant as the workforce becomes more racially diverse and organizations strive to prom ote racial equality in the workplace.
77 References Aguinis, H. (2004). Regression analysis for categorical moderators New York: The Guilford Press. Allen, T. D., Eby, L. T., Poteet, M. L., Lent z, E., & Lima, L. (2004). Career benefits associated with mentoring for proteges: A meta-analysis. Journal of Applied Psychology, 89 127-136. Allen, T. D., Lentz, E., & Day, R. (2006). Ca reer success outcomes associated with mentoring others: A comparison of mentors and nonmentors. Journal of Career Development, 32, 272-285. Allen, T. D., Poteet, M. L., & Burroughs, S. M. (1997). The mentors perspective: A qualitative inquiry and future research agenda. Journal of Vocational Behavior, 51 70-89. Allen, T. D., & Rush, M. C. (1998). The effect s of organizational citizenship behavior on performance judgments: A field stud y and a laboratory experiment. Journal of Applied Psychology, 83, 247-260. Allen, T. D., Russell, J. E. A., & Rush, M. C. (1994). The effects of gender and leave of absence on attributions for high perf ormance, perceived organizational commitment, and allocation of organizational rewards. Sex Roles, 31 443-464. Berger, J., Cohen, B. P., & Zelditch, M., Jr. (1972). Status charac teristics and social interactions. American Sociological Review, 37 241-255.
78 Bertrand, M., & Mullainathan, S. (2004). Ar e Emily and Greg more employable than Lakisha and Jamal? A field experime nt on labor market discrimination. The American Economic Review, 94, 991-1013. Best practices in bank employee me ntoring programs. (2006, February). Community Banker. Retrieved June 1, 2007, from http://www.mentorscout.com/news/newsarchive.cfm?id=869 Bozionelos, N. (2004). Mentoring provided: Relation to mentors career success, personality, and mentoring received. Journal of Vocational Behavior, 64 24-46. Chatman, C. M., & von Hippel, W. (2001). A ttributional mediation of in-group bias. Journal of Experiment al Social Psychology, 37 267-272. Crant, J. M, & Bateman, T. S. (1993). Assi gnment of credit and blame for performance outcomes. Academy of Manage ment Journal, 36 7-27. Cuddy, A. J. C., Rock, M. S., & Norton, M. I. (2007). Aid in the aftermath of Hurricane Katrina: Inferences of secondary emotions and intergroup helping. Group Processes & Intergroup Relations, 10 107-118. DeShon, R. P., & Alexander, R. A. (1996). A lternative procedures for testing regression slope homogeneity when group error variances are unequal. Psychological Methods, 1 261-277. Elvira, M. M., & Zatzick, C. D. (2002). Whos displaced first? The role of race in layoff decisions. Industrial Relations, 41 329-361. Frazier, P. A., Tix, A. P., & Barron, K. E. ( 2004). Testing moderator and mediator effects in counseling psychology research. Journal of Counseling Psychology, 51 115134.
79 Green, S. G., & Mitchell, T. R. (1979). Attr ibutional processes of leaders in leadermember interactions. Organizational Behavior and Human Performance, 23 429458. Greenhaus, J. H., & Parasuraman, S. (1993). Job performance attributions and career advancement prospects: An examination of gender and race effects. Organizational Behavior and Huma n Decision Processes, 55, 273-297. Greenhaus, J. H., Parasuraman, S., & Wormley, W. M. (1990). Effects of race on organizational experiences, job perfor mance evaluations, and career outcomes. Academy of Manage ment Journal, 33 64-86. Hamilton, D. L., & Trolier, T. K. (1986). St ereotypes and stereotyping: An overview of the cognitive approach. In J. F. D ovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 127-158). Orlando, FL: Academic Press. Heilman, M. E. (1983). Sex bias in work set tings: The lack of fit model. In L. L. Cummings & B. M. Staw (Eds.), Research in organizational behavior (Vol. 5, pp. 269-298). Greenwich, CT: JAI Press. Heilman, M. E., & Guzzo, R. A. (1978). The perceived cause of work success as a mediator of sex discrimi nation in organizations. Organizational Behavior and Human Performance, 21 346-357. Hewstone, M. (1990). The ultimate attributi on error? A review of the literature on intergroup causal attribution. European Journal of Social Psychology, 20 311335.
80 Ilgen, D. R., & Youtz, M. A. (1986). Factor s affecting the evaluations and development of minorities in organizations. In K. M. Rowland & G. R. Ferris (Eds.), Research in personnel and human resources management (Vol. 4, pp. 307-337). Greenwich, CT: JAI Press. Jackson, L. A., Sullivan, L. A., & Hodge, C. N. (1993). Stereotype effects on attributions, predictions, and evaluations: No two social judgments are quite alike. Journal of Personality and Social Psychology, 65, 69-84. Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska symposium on motivation (Vol. 15, pp. 192-238). Lincoln, NE: University of Nebraska Press. Landau, J. (1995). The relationship of race and gender to managers ratings of promotion potential. Journal of Organizational Behavior, 16 391-400. Levitt, S. D., & Dubner, S. J. (2005). Freakonomics: A rogue economist explores the hidden side of everything. New York: HarperCollins Publishers Inc. Martinko, M. J., Douglas, S. C., & Harvey, P. (2006). Attribution theo ry in industrial and organizational psychology: A review. In G. P. Hodgkinson & J. K. Ford (Eds.), International review of industr ial and organizational psychology (Vol. 21, pp. 127-187). Hoboken, NJ: Wiley Publishing. Martinko, M. J., & Gardner, W. L. (1987). The leader/member at tribution process. Academy of Management Review, 12, 235-249. Maume, D. J., Jr. (1999). Glass ceilings a nd glass escalators: O ccupational segregation and race and sex differences in managerial promotions. Work and Occupations, 26 483-509.
81 Nemeth, C. J. (1988). Style without status expectations: The special contributions of minorities. In M. Webster, Jr., & M. Foschi (Eds.), Status generalizations: New theory and research (pp. 281-290). Stanford, CA: Stanford University Press. Office of Decision Support. (2008). InfoMart Report Retrieved March 28, 2009, from http://usfweb2.usf.edu/infomart/infomartapps/Major10.aspx Pazy, A. (1986). The persistence of pro-male bias despite identical information regarding causes of success. Organizational Behavior and Human Decision Processes, 38 366-377. Pettigrew, T. F. (1979). The ultimate attribut ion error: Extending Allports cognitive analysis of prejudice. Personality and Social Psychology Bulletin, 5, 461-476. Pettigrew, T. F., & Martin, J. (1987). Shap ing the organizational context for black American inclusion. Journal of Social Issues, 43 41-78. Powell, G. N., & Butterfield, D. A. (1997) Effect of race on promotions to top management in a federal department. Academy of Manage ment Journal, 40 112128. Powell, G. N., & Butterfield, D. A. (2002). E xploring the influence of decision makers race and gender on actual promotions to top management. Personnel Psychology, 55 397-428. Ragins, B. R. (1997). Diversified mentori ng relationships in or ganizations: A power perspective. Academy of Management Review, 22 482-521.
82 Ragins, B. R. (2007). Diversity and workpl ace mentoring relationships: A review and positive social capital approach. In T. D. Allen & L. T. Eby (Eds.), Blackwell handbook of mentoring: A multiple perspectives approach (pp. 281-300). London, U.K.: Blackwell Publishing. Ragins, B. R., & Cotton, J. L. (1999). Mentor functions and outcomes: A comparison of men and women in formal and in formal mentoring relationships. Journal of Applied Psychology, 84, 529-550. Ramaswami, A., & Dreher, G. F. (2007). The benefits associated with workplace mentoring relationships. In T. D. Allen & L. T. Eby (Eds.), Blackwell handbook of mentoring: A multiple perspectives approach (pp. 211-231). London, U.K.: Blackwell Publishing. Ruble, T. L., Cohen, R., & Ruble, D. N. ( 1984). Sex stereotypes: Occupational barriers for women. American Behavioral Scientist, 27 339-356. Tajfel, H., & Turner, J. C. (1985). The social id entity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7-24). Chicago: Nelson-Hall. U.S. Bureau of Labor Statistics. (2008). Current Population Survey Retrieved March 28, 2009, from http://www.bls.gov/cps/cpsaat11.pdf U.S. Department of Labor & U.S. Bu reau of Labor Statistics. (2005, May). Women in the labor force: A databook. Retrieved May 24, 2007, from http://www.bls.gov/cps/wlf-databook-2005.pdf Wanberg, C. R., Welsh, E. T., & Hezlett, S. A. (2003). Mentoring research: A review and dynamic process model. Research in Personnel and Human Resources Management, 22 39-124.
83 Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S., & Rosenbaum, R. M. (1972). Perceiving the causes of success and failure. In E. Jones, D. Kanouse, H. Kelley, R. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Percei ving the causes of behavior (pp. 95-120). Morristown, NJ : General Learning Press. Whitehead, G. I., III, Smith, S. H., & Eichhorn, J. A. (1982). The effect of subjects race and others race on judgments of causality for success and failure. Journal of Personality, 50 193-202. Yarkin, K. L., Town, J. P., & Wallston, B. S. (1982). Blacks and women must try harder: Stimulus persons race and sex attributions of causality. Personality and Social Psychology Bulletin, 8, 21-24.
85 Appendix A Vignette Instructions The purpose of this study is to examine fo rmal workplace mentoring relationships. PLEASE READ the information below CA REFULLY, as you will later be asked questions about this information. Background Information XYZ Bank is a nationwide financial instituti on, providing a broad rang e of services to individual consumers, businesses, and instit utional clients. The company operates over 1,500 retail branches throughout the United States. In 2005, XYZ Bank launched a formal mentoring program in which new bank branch managers are matched with more senior-level branch managers. The purpose of the program is three-fold: (1) to help new em ployees acclimate to their positions, (2) to provide new employees with a source of suppor t, and (3) to facilitate employee career development. Administrators of the mentor ing program match the mentors and mentees. Mentees may be matched with any higher-level branch manager except their direct supervisor. The administrators of the program formally monito r the mentoring relationships for one year. At the end of this one-year monitoring pe riod, the program administrators conduct an evaluation of the mentoring experience. As part of this evaluation, they collect information from the mentee, the mentor, and the mentees direct supe rvisor, as described below: Formal Mentoring Program Evaluation from the Mentee : Each mentee describes his/her experience in the mentoring program. Formal Mentoring Program Evaluation from the Mentor : Each mentor describes his/her experience in the mentoring program. Supervisor Evaluation of Mentee Performance: Each direct supervisor of each mentee evaluates the performance of the mentee over the past year. On the next few pages, you will find the eval uations associated with one mentor-mentee pair that participated in the mentoring pr ogram. Please read the information carefully, and then answer the questions that follow.
86 Appendix A (Continued) FORMAL MENTORING PR OGRAM EVALUATION FROM THE MENTEE General Information Mentee Name: _____
87 Appendix A (Continued) FORMAL MENTORING PR OGRAM EVALUATION FROM THE MENTOR General Information Mentor Name:
88 Appendix A (Continued) SUPERVISOR EVALUATION OF MENTEE PERFORMANCE In the space below, please provide an evaluation of _____
89 Appendix B Causal Attributions Items Please indicate the extent that each of the fo llowing factors contributed to the success of the MENTEE,
90 Appendix C Career Advancement Potential Items Using the scales below, please circle your response to the following items. 1. How would you rate the potential of th e MENTEE,
91 Appendix D Reward Recommendations Items MENTEE Reward Recommendations Please indicate the extent that you would recommend the MENTEE,
92 Appendix E Manipulation Check Items Please respond to the following questions. 1. To which of the following age groups does the MENTEE,
93 Appendix F Participant Demographics Items Please answer the following questions about yourself. 1. What is your gender? ___ Male ___ Female 2. What is your race? ___ Caucasian/White ___ African American/Black ___ Hispanic ___ Asian ___ Native American ___ Other (please specify) ________________________ 3. What is your age? ______________ 4. What is the highest level of e ducation that you have completed? ___ High school degree or less ___ Some college ___ Associate/2-year degree ___ Four year degree ___ Graduate work ___ Graduate degree 5. Are you currently employed? ___Yes ___No 6. For how long have you been employed in your current job? _______ Years _______ Months 7. How many hours do you typically spend on work per week (include work done outside of the office): _______________ 8. What is your cu rrent job title? ________________________ 9. In what industry are you currently employed? _____________________
94 Appendix F (Continued) 10. In order to assist individuals in their development and advancement, some organizations have established formal ment oring programs, where mentees and mentors are linked in some way. This may be acco mplished by assigning mentors or by just providing formal opportunities aimed at deve loping the relationship. Other types of mentoring relationships develop on their own without organizational intervention. To recap: Formal mentoring re lationships are developed with organizational assistance Informal mentoring relationships are developed spontaneously without organizational assistance. Which of the following best describes your personal involvement in a workplace mentoring relationship? ___ I have been involved in a FORMAL mentoring relationship ___ I have been involved in an INFORMAL mentoring relationship ___ I have been involved in BOTH types of relationships (informal and formal) ___ I have NEVER BEEN INVOLVED in a workplace mentoring relationship 11. If you have been involved in a wor kplace mentoring relationship, which of the following best describes your role in the relationship? ___I am/was the MENTEE. ___I am/was the MENTOR. ___I am/have been BOTH a mentee and a mentor. ___N/A 12. If you have been involved in a workplac e mentoring relationshi p, please answer the following questions regarding you and your mentoring partner(s): ___a. How many mentoring pa rtners have you had who ha ve been of the SAME RACE as you? ___b. How many mentoring partners ha ve you had who have been of a DIFFERENT RACE than you? ___c. How many mentoring pa rtners have you had who ha ve been of the SAME GENDER as you? ___d. How many mentoring partners ha ve you had who have been of a DIFFERENT GENDER than you? 13. Have you ever been in a position in wh ich you supervised th e work of others? ____ Yes ____ No 14. If yes, how many years of supe rvisory experience have you had? _____ Years 15. Have you previously part icipated in this study? ____ Yes ____ No