men- women- and managers - are stereotypes finally changing

32
PERSONNEL PSYCHOLOGY 2006, 59, 815–846 MEN, WOMEN, AND MANAGERS: ARE STEREOTYPES FINALLY CHANGING? EMILY E. DUEHR and JOYCE E. BONO University of Minnesota, Twin Cities As the number of women in management roles increases and organi- zations place a greater emphasis on diversity, a subsequent change in perceptions of women as leader-like is expected. To test this notion, we examined gender and management stereotypes of male and female managers and students. Results reveal considerable change in male man- agers’ views of women over the past 30 years, as evidenced by greater congruence between their perceptions of women and successful man- agers and stronger endorsement of agentic and task-oriented leadership characteristics for women. Stereotypes held by male students changed less, remaining strikingly similar to stereotypes held by male managers 15 years ago. Across samples, there was general agreement in the char- acteristics of managers but less agreement about the characteristics of women. We also found men somewhat less likely than women to at- tribute successful manager characteristics to women. Respondents with positive past experiences with female managers tended to rate women higher on management characteristics. In the United States, the number of women in the managerial and pro- fessional ranks has steadily increased. According to Catalyst, a research and advisory organization committed to advancing women in business, women now hold 51% of managerial and professional specialty positions (Welle, 2004). Women also hold 51% of bachelor’s degrees and 45% of all advanced degrees (Eagly & Karau, 2002). Although these numbers are larger today than ever before, the progression of women into executive positions continues to be slow. For example, among the Fortune 500 com- panies, only 16% of corporate officers, 14% of board directors, 5% of top earners, and just over 1% of CEOs are women (Welle, 2004). Much research has focused on explaining the slow managerial ad- vancement of women (e.g., Cleveland, Vescio, & Barnes-Farrell, 2005; Stroh, Brett, & Reilly, 1992), ruling out reasons such as lesser skills, edu- cation, and time out of the workforce. One plausible explanation that has not been ruled out is that women face subtle barriers in the corporate climb. In a recent survey of 120 CEOs and 705 female executives drawn from the Correspondence and requests for reprints should be addressed to Emily E. Duehr, 75 East River Road, Department of Psychology, University of Minnesota, Minneapolis, MN 55455; [email protected]. COPYRIGHT C 2006 BLACKWELL PUBLISHING, INC. 815

Upload: rashidbhutta

Post on 18-Dec-2015

19 views

Category:

Documents


1 download

DESCRIPTION

Men- Women- And Managers - Are Stereotypes Finally Changing

TRANSCRIPT

  • PERSONNEL PSYCHOLOGY2006, 59, 815846

    MEN, WOMEN, AND MANAGERS: ARESTEREOTYPES FINALLY CHANGING?

    EMILY E. DUEHR and JOYCE E. BONOUniversity of Minnesota, Twin Cities

    As the number of women in management roles increases and organi-zations place a greater emphasis on diversity, a subsequent change inperceptions of women as leader-like is expected. To test this notion,we examined gender and management stereotypes of male and femalemanagers and students. Results reveal considerable change in male man-agers views of women over the past 30 years, as evidenced by greatercongruence between their perceptions of women and successful man-agers and stronger endorsement of agentic and task-oriented leadershipcharacteristics for women. Stereotypes held by male students changedless, remaining strikingly similar to stereotypes held by male managers15 years ago. Across samples, there was general agreement in the char-acteristics of managers but less agreement about the characteristics ofwomen. We also found men somewhat less likely than women to at-tribute successful manager characteristics to women. Respondents withpositive past experiences with female managers tended to rate womenhigher on management characteristics.

    In the United States, the number of women in the managerial and pro-fessional ranks has steadily increased. According to Catalyst, a researchand advisory organization committed to advancing women in business,women now hold 51% of managerial and professional specialty positions(Welle, 2004). Women also hold 51% of bachelors degrees and 45% ofall advanced degrees (Eagly & Karau, 2002). Although these numbers arelarger today than ever before, the progression of women into executivepositions continues to be slow. For example, among the Fortune 500 com-panies, only 16% of corporate officers, 14% of board directors, 5% of topearners, and just over 1% of CEOs are women (Welle, 2004).

    Much research has focused on explaining the slow managerial ad-vancement of women (e.g., Cleveland, Vescio, & Barnes-Farrell, 2005;Stroh, Brett, & Reilly, 1992), ruling out reasons such as lesser skills, edu-cation, and time out of the workforce. One plausible explanation that hasnot been ruled out is that women face subtle barriers in the corporate climb.In a recent survey of 120 CEOs and 705 female executives drawn from the

    Correspondence and requests for reprints should be addressed to Emily E. Duehr, 75East River Road, Department of Psychology, University of Minnesota, Minneapolis, MN55455; [email protected] C 2006 BLACKWELL PUBLISHING, INC.

    815

  • 816 PERSONNEL PSYCHOLOGY

    Fortune 1,000 companies (Wellington, Kropf, & Gerkovich, 2003), 72% ofCEOs and 51% of female executives perceived stereotypes about womensroles and abilities to be an important barrier to their advancement.

    Clearly, gender stereotypes are salient in organizations as a potentialbarrier to advancement; however, the degree to which stereotypes persistin the 21st century is unclear. Thirty years have passed since issues ofgender inequality in management and leadership reached the public eye(e.g., Kanter, 1977), and in that time women have become more commonin the boardroom. As the gender balance in management changes, parallelchanges in hiring practices, mentor availability, and eventually gender rolestereotypes should follow (Kanter, 1977).

    Over the course of the past several decades, there have also beenchanges on the management front. Contemporary books and articles onmanagement describe management work in qualities traditionally definedas feminine (Fondas, 1997, p. 257), such as helping and developing oth-ers, and building networks of relationships. In the academic literature, anew genre of leadership (i.e., charismatic and transformational leadership;Bass, 1985, 1998) has dominated recent research (Judge & Piccolo, 2004).There is also ongoing debate about whether women now hold a leadershipadvantage (Eagly & Carli, 2003; Vecchio, 2002, 2003).

    Given the changes in both womens work roles and models of leader-ship effectiveness, the aim of our research is to assess current perceptionsof men, women, and successful managers. Specifically, the purpose of ourstudy is to assess management and gender stereotypes today, comparingthem with those that existed in the 1970s and 1980s (Brenner, Tomkiewicz,& Schein, 1989; Heilman, Block, Martell, & Simon, 1989; Schein, 1973,1975).

    Gender Stereotypes: Time for Change?

    Gender stereotypes are categorical beliefs regarding the traits and be-havioral characteristics ascribed to individuals on the basis of their gender.They serve as expectations about the attributes and behaviors of individualgroup members (Cleveland, Stockdale, & Murphy, 2000) and are consid-ered one of the direct antecedents of discrimination at work (Dovidio &Hebl, 2005). Typically, women are stereotyped as more communal andmen as more agentic. Communal characteristics are primarily concernedwith the welfare of other people, including attributes such as compas-sionate, kind, sentimental, helpful, and generous. Agentic characteristicsdescribe a more assertive, dominant, and confident tendency, includingattributes such as aggressive, ambitious, independent, and self-confident.Agentic characteristics have traditionally been aligned with leadershiproles (Eagly, 1987; Eagly & Karau, 2002).

  • EMILY E. DUEHR AND JOYCE E. BONO 817

    Gender stereotypes have been documented for decades. Although someresearch suggests that stereotypes are not quick to change, even in the wakeof changing social influences (e.g., Lueptow, Garovich-Szabo, & Lueptow,2001), it is clear that the social environment with respect to women hasbeen changing. The past several decades have included changes in thelegal environment (e.g., Civil Rights Act of 1964, affirmative action) andassociated changes in the extent to which organizations focus on equalopportunity employment practices, both as a function of legal guidelinesand as a movement toward fostering diversity as a business goal (Rynes &Rosen, 1995). Changes in attitudes toward women have also been docu-mented (Twenge, 1997a). These environmental changes suggest two pos-sible reasons why gender stereotypes may be changing, especially forwomen in management. The first possibility is that a gradual change ingender stereotypes may be occurring due to changing social roles (e.g.,more women at work and in management and executive positions). Thesecond possibility is change due to organizational interventions, such asdiversity training aimed at decreasing gender stereotypes and other preju-diced attitudes. Both possibilities are discussed in the next section.

    According to social role theory (Eagly, 1987), agentic and communalcharacteristics are differentially attributed to men and women because un-equal distribution into occupations and families fosters such expectations(Diekman & Eagly, 2000). As the distribution of men and women intosocial roles shifts, perceptions of the characteristics of men and women(i.e., stereotypes) should also change; however, change cannot be expectedto occur quickly (Lueptow et al., 2001). Recently, Diekman and Eagly(2000) found evidence of changing conceptions of women, reporting thatstereotypes of women have shifted toward more masculine or agentic char-acteristics. In a meta-analysis, Twenge (1997b) reported that womensself-reported masculinity scores were rising over time and proposed thatthis increase resulted from the changing social climate for women. Con-trary to Twenge (1997b), Lueptow et al.s (2001) reviewwhich exam-ines gender stereotypes based largely on self-report personality and directcomparisons of men and womensuggested that gender stereotypes haveremained stable over time with a possible increase in the perceived femi-ninity of females. Although neither of these streams of research focuses ongender and management, they do suggest that whether or not stereotypesare changing is an unsettled issue.

    A decidedly different reason to expect changing gender stereotypes isdue to the increased focus on diversity in organizations, including spe-cific interventions (e.g., diversity training) designed to foster this goal. Itis now estimated that organizations spend $8 billion annually on diver-sity training, and in a recent survey of Fortune 1,000 companies, 88%reported providing diversity training on gender (Jayne & Dipboye, 2004).

  • 818 PERSONNEL PSYCHOLOGY

    Diversity training can be aimed at increasing awareness and appreciationof differences between individuals or decreasing stereotypes held by or-ganizational members. When focused on gender, the aim is typically toidentify stereotypes and promote inclusion, rather than highlighting dif-ferences between men and women (Jayne & Dipboye, 2004). Researchon schema change shows that once schemas are established, they are veryresistant to change, even in the face of disconfirming evidence (Epitropaki& Martin, 2004; Labianca, Gray, & Brass, 2000). Epitropaki and Martin(2004) stated that unless specific interventions and conscious efforts bymanagement for schema change happen in an organization, organizationalmembers schemas are likely to remain stable (p. 295). We suggest thatdiversity training is precisely the type of intervention and conscious effortneeded to promote changes in gender stereotypes.

    A recent study by Rudman, Ashmore, and Gary (2001) examined theimpact of diversity education on stereotypes and prejudices, and foundthat training can reduce these biases at multiple levels. Not only did diver-sity education lead to a decrease in directly reported, explicit stereotypes,but such education also reduced implicit stereotypes, which occur on anautomatic, unconscious level (Greenwald & Banaji, 1995). Implicit stereo-types are thought to be more stable and enduring associations because theyhave been learned through years of environmental influences (Kawakami& Dovidio, 2001). If such associations can be unlearned through diversitytraining, and diversity training is common in todays work organizations,then it is possible that gender stereotypes may be changing as a resultof direct interventions. If diversity training is a key factor influencingstereotype change, we would expect to see greater change among thoseindividuals with more time and experience in the workplace, due to theirincreased participation in such training.

    Research Using the Schein Descriptive Index

    A crucial consideration when examining gender stereotypes in workorganizations is the extent to which these stereotypes affect perceptionsof managers. Research within the Schein paradigm focuses on the rela-tionship between gender and management stereotypes, reflecting the ex-tent to which men and women are viewed as leader-like. In 1973, Scheindeveloped the Descriptive Index to assess the extent to which men andwomen were perceived to have the requisite personal characteristics ex-pected for management positions. Using a broad list of adjectives, Scheinfound that the characteristics of successful middle managers were muchmore similar to the characteristics commonly ascribed to men in gen-eral and not at all like the characteristics attributed to women in general.Schein (1975) replicated these results with a sample of female managers,

  • EMILY E. DUEHR AND JOYCE E. BONO 819

    demonstrating that it was not only males who held gender stereotypes in theworkplace.

    In a key extension of the Schein paradigm approximately 15 years afterthe original research, Heilman et al. (1989) examined the extent to whichgender stereotypes persisted in organizations. Heilman et al. (1989) repli-cated Scheins (1973) original work and found stereotypical views aboutthe characteristics of men in general, women in general, and success-ful managers at a level that closely paralleled Scheins (1973) findings,suggesting little change in the stereotypes of male managers over time.Heilman also extended Scheins research by comparing successful man-agers to male and female managers and to successful male and femalemanagers, finding considerably weaker gender stereotypes when more in-formation was provided about the managerial success of women (e.g.,female managers or successful female managers).

    Concurrent research with female managers yielded slightly differentresults. Brenner et al. (1989) replicated the original Schein studies usingboth male and female management samples. They found no evidence ofchanging stereotypes among male managers; however, female managersrated both men and women as similar to successful managers. This dispar-ity between male and female respondents was due largely to differencesin their view of women, not in their view of successful managers.

    Since 1989, researchers have continued to use the Schein paradigm toidentify gender stereotypes, but nearly all of this research has used studentsamples. Although some researchers have argued that college studentswould be less likely to report gender stereotypes due to a more egalitariansocial context (Lueptow et al., 2001), research using the Schein paradigmhas repeatedly shown that college students hold strong gender stereotypes,especially the male students (Schein & Mueller, 1992). Similar resultshave been reported among students in Germany, Great Britain, Japan,and China (Schein, Mueller, & Jacobson, 1989; Schein, Mueller, Lituchy,& Liu, 1996). This pattern of findings led Schein (2001) to conclude,In the United States many people believed that as women moved intomanagement, managerial sex typing would diminish. And it did, amongwomen. But men have continued to see women in ways that are not com-plimentary vis-a`-vis succeeding in positions of authority and influence(p. 684).

    This discouraging statement on gender and management stereotypesmay not apply uniformly to all men. Results derived from student sam-ples may not generalize to employees in work organizations, especiallymanagers, who experience both increased exposure to women leadersand direct interventions such as diversity training. Key replications ofthe Schein paradigm with managers took place in the late 1980s, at atime when women were fast increasing their presence in organizations

  • 820 PERSONNEL PSYCHOLOGY

    (Heilman et al., 1989), but since that time the number of women in lead-ership positions has continued to grow (Welle, 2004). Therefore, afteranother 15-year period, the time is ripe to examine whether gender stereo-types held by managers have changed. It is not sufficient to rely solelyon student samples to address this question. Therefore, we included bothmanagers and students in our research to provide a more thorough portraitof current gender and management stereotypes.

    Since the first Schein (1973) study of gender and management stereo-types, there have been many advances in the literature with respect to theconceptualization and measurement of stereotypes, and adjective check-lists, such as the Schein Descriptive Index, have been criticized. Devineand Elliot (1995) distinguished between ratings of stereotypes and rat-ings of personal beliefs. According to their distinction, the Schein Indexfocuses on personal beliefs, which may or may not be congruent witheither knowledge or endorsement of stereotypes (see Kunda & Spencer,2003). However, by aggregating the personal beliefs of male and femalemanagers and students, as we do in this research, we can examine thegender and management stereotypes held by groups of individuals (e.g.,male managers).

    Recent stereotype research has also demonstrated differences betweenexplicit and implicit stereotypes (e.g., Rudman et al., 2001; Ziegert &Hanges, 2005). Implicit stereotypes are the introspectively unidentifiedtraces of past experience that mediate attributions of qualities to membersof a social category (Greenwald & Banaji, 1995, p. 15). Additional re-search has focused on the difference between descriptive (i.e., consensualexpectations about what men and women actually do) and prescriptive (i.e.,consensual expectations about what men and women should do) stereo-types (e.g., Eagly & Karau, 2002; Heilman, Wallen, Fuchs, & Tamkins,2004). We recognize both the complexity of stereotypes and their mea-surement and the possible role that implicit and prescriptive stereotypesmay play in the advancement of women in management. However, as thepurpose of our study was to compare views of men, women, and man-agers over time, it was necessary for us to use an explicit measure, whichdue to its reliance on adjective descriptors of men, women, and managers,assesses descriptive gender stereotypes.

    A second concern raised by Devine and Elliot (1995) was the use of out-dated adjectives, which may provide a limited description of men, women,and managers. Given the changing leadership paradigms over the past30 years, we felt it was crucial to add adjectives reflecting a broader range ofleadership styles. In particular, adjectives describing relationship-orientedand transformational leadership were absent from the original Descrip-tive Index, whereas task-oriented leadership characteristics were well-represented. Task-oriented leadership behaviors emphasize group output;

  • EMILY E. DUEHR AND JOYCE E. BONO 821

    such as establishing objectives and goals, structuring tasks, and evaluat-ing work quality. In contrast, relationship-oriented behaviors emphasizesupportive personal relationships, a willingness to develop employees anddemonstrations of respect and warmth (Bales, 1954; Bowers & Seashore,1966; House & Aditya, 1997). Although task and relationship-orientedleadership have a long history in the leadership research, recent researchhas focused more on transformational leadership behaviors (Judge & Pic-colo, 2004). Transformational leaders inspire and motivate followers withoptimism and commitment to a compelling vision. They link work goalsto worker values, challenge established practices, and attend to the indi-vidual growth needs of followers (Bass, 1985, 1998). Given that recentmeta-analyses have highlighted the positive effects of both relationship1(Judge, Piccolo, & Ilies, 2004) and transformational (Judge & Piccolo,2004) leadership behaviors on employee attitudes and motivation, groupperformance, and leader effectiveness, we added adjectives describingthese behaviors to the index.

    Most past research using the Schein Index examined each of the 92 ad-jectives individually. In order to make comparisons over time, we deemedit important to use the original adjectives, but we also combined the adjec-tives to form several scales. In order to assess broad gender stereotypes,agentic and communal scales were formed. In addition, we combined ad-jectives to form scales for task-oriented leadership, relationship-orientedleadership, and transformational leadership to better link this researchto current models of effective leadership (Judge & Piccolo, 2004; Judgeet al., 2004).

    Individual Differences in Beliefs About Men, Women, and Managers

    Although the primary purpose of our research is to examine stereotypesheld by groups of individuals (e.g., male managers, female students), verylittle is known about the characteristics of individuals who predict theirbeliefs about men, women, and managers. Most existing research usingthe Schein Descriptive Index has used either student or managerial sam-ples, preventing direct comparison among these groups. Moreover, whencomparing the stereotypes of managers and students, it is not clear whetherdifferences in stereotypes between these groups are due to the effects ofage, years of work experience, experience with female managers, or hold-ing a managerial role. Therefore, an additional aim of our research is to

    1The Judge et al. (2004) meta-analysis uses the label consideration instead ofrelationship-oriented leadership. These categories refer to comparable and concurrent pro-grams of research. The labels have frequently been used interchangeably in research (e.g.,Eagly & Johnson, 1990; Sczesny et al., 2004).

  • 822 PERSONNEL PSYCHOLOGY

    tease apart the source of differences between individuals in their beliefsabout men, women, and managers by examining personal characteristics(i.e., age, gender, education) and organizational experiences (i.e., havinga female supervisor, being satisfied with a female supervisor).

    Summary of Research Questions

    Given the continued movement of women into management posi-tions and changes in dominant leadership paradigms over the past severaldecades, the time is ripe to examine whether anything has changed withrespect to management and gender stereotypes. To that end, we addressfive specific research questions.

    (1) Research Question 1. Have management and gender stereotypesheld by male and female managers changed relative to 15 and 30years ago?

    (2) Research Question 2. Have management and gender stereotypes heldby male and female students changed, and how do they compare tothe stereotypes of male and female managers?

    (3) Research Question 3. If gender stereotypes have changed, what isdriving that change? Have views of managers changed, have viewsof men and women changed, or have both changed?

    (4) Research Question 4. Do the broad gender stereotypic (agentic, com-munal) and leadership-specific (task-oriented, relationship-oriented,transformational) characteristics attributed to men, women, andmanagers differ by sample?

    (5) Research Question 5. Do individual differences in education, age,management experience, and experiences with female supervisorspredict beliefs about men, women, and managers?

    Method

    Participants and Procedures

    We used four distinct samples in this research: male managers, femalemanagers, male students, and female students. Managers (n = 620) whoparticipated in this research were drawn from a variety of public and privatesector organizations and came from a variety of job types (e.g., account-ing, human resources, law enforcement, public works, etc.). All managerswere enrolled in voluntary leadership development programs. Data werealso collected from undergraduate students (n = 688) at a large publicuniversity. Students were enrolled in a variety of psychology courses andreceived credit for their participation. They represented a broad array of

  • EMILY E. DUEHR AND JOYCE E. BONO 823

    TABLE 1Sample Characteristics for This Study

    Sample N Sample characteristics Summary statisticsMale managers 333 Age M = 48 years, SD = 8.8

    Education 81% BA or higher# Direct reports x = 9, SD = 13Race

    Caucasian 85.2%African American/Black 3.1%Hispanic 1.7%Asian 1.7%

    Female managers 287 Age M = 46 years, SD = 9.2Education 78% BA or higher# Direct reports x = 9, SD = 10Race

    Caucasian 88.7%African American/Black 4.0%Hispanic 0%Asian 1.3%

    Male students 221 Age M = 21 years, SD = 3.8Managerial experience 25% had been managersRace

    Caucasian 79.6%African American/Black 5.0%Hispanic 1.4%Asian 9.5%

    Female students 467 Age M = 20 years, SD = 3.6Managerial experience 15% had been managersRace

    Caucasian 78.6%African American/Black 2.4%Hispanic 1.3%Asian 13.3%

    academic majors, including economics, journalism, business, and psy-chology. Demographic information regarding participants age, race, edu-cation, and number of direct reports (for managers) is provided in Table 1.The age of the managers in our samples is comparable to Schein (1973,1975) and Heilman et al. (1989).

    Surveys were administered to managers as an optional component ofa survey used in leadership development programs. Surveys were dis-tributed during orientation and completed prior to the start of any formalprogram activity. The research portion of the survey was clearly identifiedas distinct from the leadership assessment, which was for developmentalpurposes only and not provided to the managers organization. Therefore,

  • 824 PERSONNEL PSYCHOLOGY

    managers were encouraged to provide honest responses throughout thesurvey as the results would be used for research purposes only. The re-search portion of the survey was confidential and anonymous and wasreturned in postage-paid envelopes to the authors. Among the managerswho completed a leadership development survey, 82% also completed aresearch survey. Surveys were distributed to student samples in severalsmall group sessions. Student surveys were anonymous.

    There were seven versions of our survey, each one representing a dif-ferent target condition. The seven conditions were (a) successful middlemanagers, (b) women in general, (c) men in general, (d) women managers,(e) men managers, (f) successful women managers, and (g) successful menmanagers. The first condition (successful middle managers) is the controlcondition and Conditions 2 through 7 are gendered conditions that rangein level of specificity (e.g., from women in general, to women managers,to successful women managers). Participants were randomly assigned toone of seven conditions. All seven surveys included the same list of de-scriptive adjectives and instructions but differed with respect to the targetgroup. For example, some respondents were asked to report the extent towhich each adjective was reflective of women in general, whereas oth-ers were asked to report the extent to which each adjective was reflectiveof successful men managers. Therefore, each participant responded toonly one target condition. The number of participants responding to eachtarget condition varies by sample and is reported in Table 2.

    Measures

    Gender stereotypes. A revised version of the Descriptive Index(Schein, 1973), including the original 92 items plus 26 additional newitems (described below), was used to measure gender stereotypes andcharacteristics of successful middle managers. Despite widespread use ofthe Descriptive Index, there is very little information published regardingits psychometric properties. Based on the suggestion of anonymous re-viewers, we collected some post hoc data to address this concern. Amonga student sample (n = 30), we found the 2-week testretest reliability tobe .90, suggesting relatively stable ratings for a given target condition. Wealso examined whether ratings would differ if the control condition waslabeled successful manager rather than successful middle manager.Among a student sample (n = 97), we found these ratings to be highlycorrelated (r = .98), indicating similar perceptions of the characteristicsof managers (more generally) and middle managers.

    We added 26 new items to the Descriptive Index to address concernsabout outdated adjectives (Devine & Elliot, 1995) and to better repre-sent current styles of leadership. We added 13 new adjectives to describe

  • EMILY E. DUEHR AND JOYCE E. BONO 825

    TABLE 2Sample Breakdown by Condition of the Descriptive Index

    Sample Condition NMale managers 1. Successful middle managers 57

    2. Women in general 503. Men in general 404. Women managers 395. Men managers 516. Successful women managers 387. Successful men managers 58

    Female managers 1. Successful middle managers 362. Women in general 353. Men in general 364. Women managers 355. Men managers 486. Successful women managers 507. Successful men managers 47

    Male students 1. Successful middle managers 322. Women in general 283. Men in general 364. Women managers 315. Men managers 336. Successful women managers 367. Successful men managers 25

    Female students 1. Successful middle managers 652. Women in general 723. Men in general 704. Women managers 775. Men managers 586. Successful women managers 607. Successful men managers 65

    transformational leaders. The new items were based on the most widelyused measure of transformational leadership, the Multifactor LeadershipQuestionnaire (MLQ; Avolio, Bass, & Jung, 1995). Each author separatelyreviewed the MLQ and developed a list of adjectives that were reflectiveof transformational leadership, resulting in 13 items to be added to theinventory (see Appendix A). Additional items were added to reflect man-agement characteristics that are relationship oriented, as such adjectiveswere largely unrepresented among the original 92 items of the Descrip-tive Index. Participants responded to all 118 items using a 5-point ratingscale ranging from 1 not characteristic to 5 characteristic. Survey instruc-tions were modeled after Schein (1975) and asked participants to rate eachadjective according to what they think the target group is like.

  • 826 PERSONNEL PSYCHOLOGY

    TABLE 3Intercorrelations Among Gender and Leadership Scales in This Study

    Scale 1 2 3 4 51. Agentic .782. Communal .33 .733. Task .69 .06 .804. Relationship .03 .68 .39 .875. Transformational .16 .60 .50 .89 .94

    Note. Combined sample = 1,363. Scale alphas are presented on the diagonal.p < .05. p < .01.

    We also used the original and new Descriptive Index adjectives toform scales. Scales were chosen a priori and a judgmental sort wasundertaken by the authors. This method was preferable to an empiricalsort (i.e., factor analysis) because specific scales were selected based ontheir theoretical relevance to gender and management stereotypes (Eagly,Johannesen-Schmidt, & van Engen, 2003; Eagly & Johnson, 1990; Eagly& Karau, 2002). The selection of adjectives to form each scale was guidedby previous research. More specifically, with respect to broad genderstereotypes, we formed agentic and communal scales using adjectivesor direct synonyms described by Eagly and Karau (2002). Furthermore,an effort was made to include agentic and communal adjectives with bothpositive and negative connotations, as in Diekman and Eagly (2000). Thisprocess identified 14 adjectives from Scheins original 92 to reflect agenticand communal characteristics. With respect to leadership-specific scales,we formed task and relationship-oriented scales based on recent researchby Sczesny (2003) and Sczesny, Bosak, Neff, and Schyns (2004), whoclassified attributes into highly reliable task and person-oriented scales.Whenever possible, we matched exact adjectives or synonyms from theDescriptive Index to form these scales; however, several new adjectiveswere included in the relationship-oriented leadership scale as these itemswere underrepresented among the original items of the Descriptive Index.We used the 13 new transformational leadership items to form a transfor-mational leadership scale.

    All gender and leadership scales and associated items are listed in Ap-pendix A. Scale alphas and intercorrelations are presented in Table 3. Therelatively high intercorrelations reported between some scales were notsurprising. Notably, the correlation between the agentic and task-orientedscales (r = .69) is in line with the stereotypic notion of task-orientedleadership as more masculine, and the correlation between the commu-nal and relationship-oriented scales (r = .68) reflects the more feminineassociations with this style of leadership (Cann & Siegfried, 1990; Eagly& Johnson, 1990). A high correlation between relationship-oriented and

  • EMILY E. DUEHR AND JOYCE E. BONO 827

    transformational leadership was also anticipated, although the associationbetween our transformational and relationship-oriented scales (r = .89) issomewhat higher than previous meta-analytic estimates of this relationship(Miliffe, Piccolo, & Judge, 2005). Given the strong, positive correlationsbetween transformational and relationship-oriented leadership behaviors,Judge et al. (2004) called for research aimed at assessing the extent towhich they represent distinct types of leadership behavior, but to date lit-tle empirical work has fully addressed this issue (see Seltzer & Bass, 1990for an exception). Therefore, consistent with existing leadership literature,we treat relationship-oriented and transformational leadership behaviorsas distinct constructs.

    Individual differences. To address our goal of understanding better theindividual differences in characteristics that predict individuals beliefsabout men, women, and managers, we asked a limited number of back-ground items at the end of the survey. Participants were asked to reporttheir age, gender, and level of education on a 5-point scale correspondingto: high school, 1 = associates degree 2 = BA/BS 3 = MA/MS 4 = orPhD 5 =. Managers were also asked to indicate the number of persons whoreported directly to them. Students were asked if they had ever been a man-ager. A small portion (15%) of the students reported having been a managerfor an average of 1.9 years. Due to the large discrepancy in experience andage between students and the manager sample, we retained students withmanagement experience in the student sample. In addition to these items,we assessed participants experiences with female supervisors, via twoquestions: Have you ever had a female supervisor (yes or no)? and Ifyes, on average how positively would you rate the experience (from 1 =poor to 5 = excellent)? Participants were instructed to provide an averageif they have had multiple female supervisors.

    Results

    In reviewing our results, it is important to keep in mind that we have foursamples and seven conditions. Thus, in some cases we will be presentingresults of 28 different comparisons for a single research question. Table 2,which describes the four samples and seven conditions in our study, maybe a helpful guide in following our results.

    Original Items of the Descriptive Index

    To determine the degree of correspondence between ratings of suc-cessful middle managers and men and women, intraclass correlation coef-ficients (ICCs) were used. ICCs were preferable to Pearsons correlationsfor these analyses because ICCs consider both the relative correspondenceand the absolute agreement between ratings. As in past research, ICCs

  • 828 PERSONNEL PSYCHOLOGY

    were computed from randomized-groups analyses of variance where thegroups, or classes, were the original 92 adjectives. The scores within eachclass were the mean item ratings for each adjective, provided separatelyfor each target condition and sample. ICCs were computed between thecontrol condition (successful middle managers) and each of the genderedconditions, which are treated as independent samples.2 This process wasrepeated separately for each of the four samples used in this study. Theresulting ICCs report the similarity of participants ratings of successfulmiddle managers to each of the six gendered conditions. A high correla-tion reflects similar ratings for a particular set of comparison conditions(e.g., successful middle managers and women managers). The size of thecorrelation between any two conditions reflects the degree to which thecomparison group (e.g., women in general or successful male managers)is perceived to have characteristics similar to those of successful mid-dle managers. If the difference in the correlations between two sets ofconditions (e.g., successful managers and men in general as compared tosuccessful managers and women in general) exceeds .29, the difference isstatistically significant (p < .05).3

    Male and female managers. In Table 4, we present correlations for oursamples, along with correlations from past research using managementsamples for comparison. The row label indicates which of the genderedconditions is being compared to successful middle managers (i.e., Row1 compares women in general and successful middle managers) and thecolumn label indicates which sample the data is drawn from. Most ofthe correlations were significant, indicating more than a chance level ofsimilarity between the six gendered conditions and successful managers;however, the magnitude of these relationships varies widely by conditionand sample.

    The most notable change in results over time is in the comparisonbetween perceptions of successful middle managers and women in general.

    2The current use of ICCs as a measure of correspondence is comparable to a two-wayrandom effects model/absolute agreement in reliability analyses where two raters rated92 objects. In our use of the ICCs, the raters are analogous to the control and genderedconditions while the objects are adjectives. We thank an anonymous reviewer for providingthis illustration to aid in understanding our analyses.

    3In our analyses using the Descriptive Index, the sample size is the number of adjec-tives (92), not the number of respondents in each condition. Therefore, if the difference inthe correlations between two sets of conditions exceeds .29, the difference between thesecorrelations reaches statistical significance (p < .05). We note that tests of statistical sig-nificance are heavily influenced by sample size, and correlations should only be comparedif variances are equal across samples (Cudeck, 1989). Because we do not have variabilitydata for the Heilman et al. (1989) data and variances in our data vary somewhat by sampleand target condition, the .29 difference marking significance (p < .05) between correlationsshould be used with some caution.

  • EMILY E. DUEHR AND JOYCE E. BONO 829

    TAB

    LE4

    Intra

    clas

    sCor

    rela

    tion

    Coeffi

    cients

    Acro

    ssVa

    riou

    sSam

    ples

    an

    dCo

    nditi

    onsf

    orth

    eO

    rigin

    al92

    Adjec

    tives

    ofth

    eSc

    hein

    Des

    crip

    tive

    Inde

    x

    Sam

    ple

    Bre

    nner

    Hei

    lman

    Bre

    nner

    Sche

    inSc

    hein

    Sche

    inet

    al.

    etal

    .Sc

    hein

    etal

    .et

    al.

    etal

    .(19

    73)

    (1989

    )(19

    89)

    (1975

    )(19

    89)

    (1989

    )(19

    89)

    Gro

    upsb

    eing

    mal

    em

    ale

    mal

    eM

    ale

    fem

    ale

    fem

    ale

    Fem

    ale

    mal

    eM

    ale

    fem

    ale

    Fem

    ale

    com

    pare

    dm

    anag

    ers

    man

    ager

    sm

    anag

    ers

    man

    ager

    sm

    anag

    ers

    man

    ager

    sm

    anag

    ers

    stud

    ents

    stud

    ents

    stud

    ents

    stud

    ents

    Wo

    men

    and

    man

    ager

    s.06

    .01

    .24

    .63

    .30

    .52

    .70

    .11

    .10

    .43

    .35

    Men

    and

    man

    ager

    s.62

    .72

    .54

    .61

    .54

    .59

    .49

    .70

    .40

    .51

    .45

    Wo

    men

    man

    ager

    san

    dm

    anag

    ers

    .58

    .81

    .96

    .69

    .91

    Men

    man

    ager

    san

    dm

    anag

    ers

    .86

    .74

    .61

    .68

    .78

    Succ

    essf

    ulw

    om

    en

    man

    ager

    san

    dm

    anag

    ers

    .93

    .97

    .98

    .93

    .98

    Succ

    essf

    ulm

    en

    man

    ager

    san

    dm

    anag

    ers

    .98

    .97

    .95

    .95

    .95

    Note

    .D

    ata

    from

    this

    stud

    yar

    epr

    esen

    ted

    inbo

    ld,i

    nda

    taCo

    lum

    ns4,

    7,9,

    and

    11.

    p