USF Libraries
USF Digital Collections

Perception of leadership qualities in higher education

MISSING IMAGE

Material Information

Title:
Perception of leadership qualities in higher education impact of professor gender, professor leader style, situation, and participant gender
Physical Description:
Book
Language:
English
Creator:
LaRocca, Michela A
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
Leadership -- Psychological aspects   ( lcsh )
Sex differences -- Social aspects   ( lcsh )
leadership dimension
task/personal
democratic/autocratic
theoretical implications
gender stereotypes
Dissertations, Academic -- Measurement and Evaluation -- Doctoral -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: This experimental study used eight written vignettes to analyze the effects of professor gender, professor leadership style (democratic/autocratic), and type of situation (task/personal) and participant gender on evaluations of professorsʹ competence, likeability and masculinity characteristics. Undergraduates from the College of Arts and Science (N=932; Males=464, Females=467), and the College of Education (N=722; Males=140, Females=582) were used. Results indicated that research participants rated democratic professors significantly more competent, likeable, and more feminine than autocratic professors. Contrary to expectations derived from gender spill-over and gender congruency theories, male participants did not rate female professors more negatively than their male counterparts when they acted autocratically in a personal situation (i.e., gender incongruent manner.) Exploratory results revealed trends that are discussed along with theoretical and practical implications.
Thesis:
Thesis (Ph.D.)--University of South Florida, 2003.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Michela A. LaRocca.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 184 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 52281590
notis - AJL4046
usfldc doi - E14-SFE0000103
usfldc handle - e14.103
System ID:
SFS0024799:00001


This item is only available as the following downloads:


Full Text

PAGE 1

PERCEPTION OF LEADERSHIP QUALITIES IN HIGHER EDUCATION: IMPACT OF PROFESSOR GENDER, PROFESSOR LEADER STYLE, SITUATION, AND PARTICIPANT GENDER by MICHELA A. LAROCCA A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Educational Measurement & Research College of Education University of South Florida Co Major Professor: Robert F. Dedrick Ph.D. Co Major Professo r: John M. Ferron Ph.D. Bruce W. Hall, Ed.D. Kathy Mcnelis, Ph.D. Daphne Thomas, Ph.D. Date of Approval: March 31, 2003 Keywords: leadership dimension, tas k/personal, democratic/autocratic, gender stereotypes, theoretical implications Copyright 2003 Michela A. Larocca

PAGE 2

Table of Contents List of Tables v List of Figures vii Abstract x Chapter I Introduction 1 Problem Statement and Purpose of the Study 3 Organization of the Study 5 Chapter II Review of the Literature 7 Section I Gender Role Models 7 Physical Characteristics as they Relate to Task/Person Dimension 9 Gender role spillover 10 Gender role congruence 11 Social role theory 12 Section II Observed Gender Differenc es in Leadership Styles 13 Section III Perception of Gender Differences in Leadership Positions and Leader Evaluation 17 Gender and Leader Competence 19 Gender and Leader Likeability 23 Interaction Betwee n Rater Gender and Leader Gender 29 Gender and Leadership Issues in Academia 30 Section IV Conceptual Issues 34 Leadership Dimensions 34 Summary and Purpose of Current Study 38 i

PAGE 3

Chapter III Method 41 Section I Overview 41 Hypotheses 41 Main Effects 42 Dimension of leader competence 42 Dimension of leader masculinity/femininity 42 Dimension of leader likeability 42 Interaction Effects 42 Interaction between participant gender, leader style, and situation on competence and likeability ratings 42 Section II Research Participants 45 Section III Instrument Develop ment 52 Experimental Conditions 52 Task Situation 53 Personal Situation 53 Professor Nonverbal Democratic Leadership Style for Task Situations 53 Professor Nonverbal Autocratic Leadership Style for Task Situations 54 Professor Nonverbal Democratic Leadership Style for Personal Situations 54 Professor Nonverbal Autocratic Leadership Style for Personal Situations 54 Professor Study Dyads 54 Development of Dependent Measures 55 Procedure 57 Chapter IV Results 59 Section I Psychometric Analysis of the Four Leadership Dimensions 59 Competence 59 ii

PAGE 4

Masculinity/Femininity 59 Likeability 60 Professional Traits 60 Principal axis factoring specifying four factors 61 Principal axis procedure with two factors 64 Comp arison of two factors results with originally proposed four factors 71 Overview of Confirmatory Factor Analysis Procedure (CFA) 72 Four Factors Revisited 74 Section II Hypotheses Testing 78 Hypothesis I 87 Hypothesis 2 87 Hypothesis 3 91 Hypothesis 4 93 Hypothesis 5 94 Additional Findings for the Competence Factor 94 Int eraction effects 94 Additional Findings for the Masculinity Factor 98 Additional Findings for the Likeability Factor 100 Chapter V Discussion 107 Section I Summary 108 Section II Psychometric Properties of Leadership Construct 111 Section III Hypotheses Testing 111 Hypothesis 1: Democratic professors will be rated as significantly more competent than autocratic professors 11 1 Hypothesis 2: Autocratic professor will be rated as significantly more masculine than democratic professors 114 Hypothesis 3: Democratic professors will be rated as significantly more likeable than autocratic professors 113 iii

PAGE 5

Hypothesis 4: There will be an interaction effect between participant gender, professor gender, professor leader style and type of situation on the ratings of leader competence 114 Hypothesis 5: There will be an interaction effect betw een participant gender, professor gender, professor leader style and type of situation on the ratings of leader likeability 117 Additional Findings 118 College of Arts and Science 118 College of Education 11 9 Section IV Future Research 120 Design 120 Implications 123 References 126 Appendices 134 Appendix A. Pilot I: Validation of Scenario Manipulation 135 Appendix B. Pilot I I: Manipulation Check on Dependent Measures 151 Appendix C. Task Vignette Booklets 157 Appendix D. Personal Vignette Booklets 164 Appendix E. Survey Introduction Letter 171 About the Author End Page iv

PAGE 7

List of Tables Table 1. Colleges Within the Participating University 45 Table 2. Student Race/Ethnicity Profile of Sample and University Population 46 Table 3. Source Information of Study Sample 47 Table 4. Reported Majors for Male and Female Arts and Science Students 48 Table 5. Reported Majors for Male and Female Education Students 49 Table 6. Summary Statistics for Age by College and Participant Gender 50 Table 7. Frequency Distribution of Scho ol Standing by College and Gender 51 Table 8. Racial/Ethnicity Frequency Distribution by College and Gender 52 Table 9. Initial Items for the Four Leadership Dimensions 57 Table 10. Summary Statistics for Proposed Competence Fact or 60 Table 11. Summary Statistics for Proposed Masculinity/Femininity Factor 60 Table 12. Summary Statistics for Proposed Likeability Factor 61 Table 13. Summary Statistics for Proposed Professional Traits Factor 61 Table 14. Eigenvalues for Principal Factor Extraction Method compared to Null Model 63 Table 15. Factor Pattern and Structure Matrices for Four Factor Solution 65 Table 16. Factor Pattern and Structure Matrices for the Two Fact or Solution 67 Table 17. Factor Names and Indicator Items 70 vi

PAGE 8

Table 18. Summary Statistics for Competence and Masculinity/Felinity Factors 70 Table 19. Summary Statistics of the Seven Largest Chisquare Change Statistic s 74 Table 20. Likeability Indicator Item Intercorrelations 75 Table 21. Summary Statistics for Leadership Factors 77 Table 22. Descriptive Statistics for Male Research Participants in the College of Arts and Scien ce for Leader Competence 80 Table 23. Descriptive Statistics for Female Research Participants in the College of Arts and Science for Leader Competence 80 Table 24. Descriptive Statistics for Male Research Participant s in the College of Education for Leader Competence 81 Table 25. Descriptive Statistics for Female Research Participants in the College of Education for Leader Competence 81 Table 26. Descriptive St atistics for Male Research Participants in the College of Arts and Science for Leader Masculinity 82 Table 27. Descriptive Statistics for Female Research Participants in the College of Arts and Science for Leade r Masculinity 82 Table 28. Descriptive Statistics for Male Research Participants in the College of Education for Leader Masculinity 83 Table 29. Descriptive Statistics for Female Research Participants in the Colle ge of Education for Leader Masculinity 83 Table 30. Descriptive Statistics for Male Research Participants in the College of Arts and Sciences for Leader Likeability 84 Table 31. Descriptive Stati stics for Female Research Participants in the College of Arts and Sciences for Leader Likeability 84 Table 32. Descriptive Statistics for Male Research Participants in the College of Education for Leader Likeab ility 85 vii Table 33. Descriptive Statistics for Female Research Participants in the College of Education for Leader Likeability 85 Table 34. Modified Levenes Test Summary Statistics for the College of Arts and Sciences and Education by Booklet Form and Participant Gender 86

PAGE 9

Table 35. ANOVA Source Table for the College of Arts and Science on Competence 88 Table 36. ANOVA Source Table for the College of Arts and Science on Masculinity/Femininity 89 Table 36. ANOVA Source Table for the College of Education on Masculinity/Femininity 90 Table 39. ANOVA Source Table for the College of Arts and Science on Likeability 92 Tab le 40. ANOVA Source Table for the College of Education on Likeability 93 viii

PAGE 10

List of Figures Figure 1. Competence Ratings By Male and Female Participants for Professors Acting Democratically in a Task Oriented Situation 43 Figure 2. Competence Ratings By Male and Female Participants for Professors Acting Democratically in a Personal Oriented Democratic Personal 44 Figure 3. Democratic Personal 44 Figure 4. Autocratic Personal 44 Figure 5. Two factor confirmatory factor analysis model for leadership measures 77 Figure 6. Competence Ratings for the College of Arts and Science: 3 Way Interaction Effects for Prof Leaders Style, Situation, and Rater Gender (Male Participants) 95 Figure 7. Competence Ratings for the College of Arts & Science: 3 Way Interaction Between Prof Leader Style, Situation, and Rater Gender (Females) 96 Figure 8. Competence Ratings for the C ollege of Education: 3 Way Interaction between Prof Gender, Prof Leader Style, and Situation (Ratings for Male Professors) 97 Figure 9. Competence Ratings for the College of Education: 3 Way Interaction between Prof Gender, Pro f Leader Style, and Situation (Ratings for Female Professors) 98 Figure 10. Masculinity Ratings for the College of Arts & Science: 2 way Interaction Effect between Prof Leader Style and Participant Gender 100 Figure 11. L ikeability Ratings for the College of Arts & Science: 2 way Interaction Effect between Type of Situation and Rater Gender 101 Figure 12. Likeability Ratings for the College of Arts & Science: 2 way Interaction between Prof Leader Style and Rater Gender 103 ix

PAGE 11

Figure 13. Likeability Ratings for the College of Arts & Science: 3 way Interaction Effect between Prof Leader Style, Situation, and Rater Gender (Male Raters) 104 Figure 14. Likeability Ratings for the College of Arts & Science: 3 way Interaction Effect between Prof Leader Style, Situation, and Rater Gender (Female Raters) 105 Figure 15. Likeability Ratings for the College of Education: 2 way Int eraction Effect between Professor Leader Style and Type of Situation 106 ix

PAGE 12

ABSTRACT This experimental study used eight written vignettes to analyze the effects of professor gender, professor leadership style (democr atic/autocratic), and type of situation (task/personal) and participant gender on evaluations of professors competence, likeability and masculinity characteristics. Undergraduates from the College of Arts and Science (N=932; Males=464, Females=467), and the College of Education (N=722; Males=140, Females=582) were used. Results indicated that research participants rated democratic professors significantly more competent, likeable, and more feminine than autocratic professors. Contrary to expectations deri ved from gender spill over and gender congruency theories, male participants did not rate female professors more negatively than their male counterparts when they acted autocratically in a personal situation (i.e., gender incongruent manner.) Exploratory r esults revealed trends that are discussed along with theoretical and practical implications. x

PAGE 13

1 Chapter I Introduction An extensive amount of empirical research has examined how men and women in leadership positions are evaluated. The majority of these studies have been conducted within an organizational setting and have evaluated leadership styles along two dimensions: (a) task accomplishment/interpersonal relationships, and (b) autocratic/ democratic style. Task accomplishment refers to a leaders hip style that concentrates on following rules, performing assigned tasks, and making leader and subordinate roles explicit; the interpersonal style is more concerned with employees morale and their relationships. Even though aspects of the task leadershi p style are considered to be independent of the interpersonal leadership style, the democratic (participative)/ autocratic (directive) leadership style dimension (e.g., Lewin, 1937) is not considered to be independent of the task person dimension. For exam ple, Eagly and Johnson (1990) suggested that the democratic style describes a more narrow aspect of the interpersonal style, while the autocratic style describes precise behaviors of the more general task oriented style. Western culture generally associate s the task and autocratic style with male gender stereotypes (e.g., masculine, dominant, aggressive, ambitious, controlled), while the personal and democratic styles are closely associated with female gender stereotypes (e.g., warm, friendly, submissive, c oncerned for others). General beliefs about a competent leader are most closely associated with masculine gender stereotypes, while female gender stereotypes are not seen as being essential attributes for a good

PAGE 14

2 leader (e.g., Broverman, Vogel, Broverman, Clarkson & Rosendrantz, 1972). While most researchers agree that gender stereotypes, in part, play a role in female under representation in leadership roles, there are several major competing theories explaining the reasons for this disparity. For example gender role spillover theory (e.g., Heilman, Block, Martell, & Simon, 1989) states that gender roles carry over from (Nieva & Gutek, 1980) into organizational settings. According to role congruence theory leaders who behave inconsistently with their gend er role expectations are more negatively evaluated than leaders who behave in a manner that is consistent with their gender roles. For this reason, Deaux and Major (1987) proposed that it is important to examine the degree to which a job emphasizes gender typed skills and behaviors. Empirical evidence of gender bias in leadership evaluation is not consistent. While meta analytic research found that women consistently were more democratic in their leadership style (Eagly et al., 1990), research evaluating th e perception of leadership outcomes most consistently found that women were negatively evaluated when (a) assessed by males, and (b) when supervising in a male dominated environment (e.g., military) (e.g., Eagly., 1987). In addition, studies consistently s howed evidence that female leaders who used stereotypically autocratic styles were evaluated most negatively by male raters, while female raters did not show this rating bias (e.g., Eagly et al., 1995). Eagly, Makhijani and Klonsky (1992), in their meta a nalytic review, found a multitude of methods used to measure leadership outcomes. They commented that these methods increased the complexity of outcome comparisons. For example, leader competence was frequently measured using leader productivity, expertise effort, and effectiveness. Similarly, leader satisfaction was also assessed in terms of leader

PAGE 15

3 likeability, desire to work with the leader, and group cohesiveness. Leadership ability was measured by some studies in terms of ability to organize, plan, and execute managerial duties. In addition to leadership outcome measures, studies frequently included scales to assess leader masculinity/femininity personality traits (e.g., Rojahn & Willemsen, 1994). However, little factor analytic research has examined th e measures of leadership. And even though an extensive amount of research has explored the link between leader gender, leadership styles, and leader ratings, only one study attempted to integrate both task/personal and autocratic/democratic leadership dime nsions in their examination of leader outcome ratings (Remland, Jacobson, & Jones 1983). This particular study was of interest because the authors integrated both leadership dimensions into their design (along with participant gender, participants psychol ogical gender, and gender of manager described in the scenario). Problem Statement and Purpose of the Study The literature review revealed that many studies have used single items to measure leader attributes, or composites of two or more items. The liter ature review of leadership studies has further revealed that even though the two leadership dimensions (task/personal, democratic/autocratic) are generally believed to be equally important as factors influencing leader evaluations, the majority of studies have not examined possible interactions between these dimensions (except for Remland et al.s study). Finally, while most leadership research has been conducted within an organizational setting, leadership studies within academia have been much less freque nt, and more importantly dated from the 70s and 80s and in need of updating.

PAGE 16

4 The purpose of this study was two pronged: (a) develop a psychometrically sound leadership evaluation scale for an academic setting, and (b) create an experimental design that incorporated both leadership style dimensions (task/personal and democratic/ autocratic) in order to examine the predictive accuracy of two well established gender role theories (gender role spillover and gender congruency theory) to explain leader evaluat ions. An experimental design was created using written vignettes describing a professor/student interaction. The researcher manipulated the following variables: professor gender, professor leadership style (democratic or autocratic), and type of situation (task or personal). This study also included participant gender as a variable, which created a 2 (professor gender) x 2 (professor leader style) x 2 (type of situation) x 2 (participant gender) between factors design. The dependent measures consisted of 2 4 leadership items representing four leadership factors: competence, likeability, masculinity/femininity, and professional traits. This study tested five hypotheses that were derived from a combination of two gender role theories (gender role spill over an d gender role congruency theory) and empirical evidence. Research participants were undergraduate students from the College of Arts and Science and the College of Education from one university. Hypothesis 1: Democratic professors will be rated as signi ficantly more competent than autocratic professors. Hypothesis 2: Autocratic professors will be rated as significantly more masculine than democratic professors.

PAGE 17

5 Hypothesis 3: Democratic professors will be rated as significantly more likeable than autocratic professors. Hypothesis 4: There will be an interaction effect between participant gender, professor gender, professor leadership style, and the type of situation on the ratings of leader competence. Hypothesis 5: There will be an interaction effect between participant gender, professor gender, professor leadership style, and the type of situation on the ratings of lead er likeability. Hypotheses 4 and 5 anticipated differences between male and female research participants with regard to competence and likeability ratings. First, it was predicted that female research participants would not be influenced by professor gende r in their ratings of leader competence and likeability, only by leadership style (see hypotheses 1 and 3). On the other hand, competence and likeability ratings for male research participants were anticipated to be influenced not only by leadership style (hypotheses 1 and 3), but also by professor gender and type of situation. First, it was anticipated that male participants would judge autocratic female professors more harshly than their autocratic male counterparts. In addition, these anticipated differe nces were predicted to be largest in the personal condition, that is, male participants would rate autocratic female professors in personal situations the least competent and the least likeably compared to all other conditions. Organization of the Study The remainder of this study is organized in the following manner. Chapter II reviews the literature on gender differences in leadership evaluations within

PAGE 18

6 organizational and academic settings. Specifically, this review summarizes studies that examined the relationships between gender role theories, leadership dimensions (task/personal, democratic/autocratic), and rater gender with regard to leadership evaluations of male and female leaders. Chapter III presents the procedures and methods used in this expe riment. Sample selection, instrument and leadership factors, and experimental procedures are discussed. Chapter IV presents the results of factor analytic procedures used to analyze the psychometric properties of the leadership factors (using the whole sam ple), followed by an in depth presentation of the results that test the five hypotheses. This chapter also presents detailed results of exploratory findings. Hypothesis testing and exploratory findings are presented separately for the College of Arts and S cience and the College of Education. Chapter V discusses the results of the five hypotheses and exploratory findings as they relate to gender role theories and findings of previously published studies. Theoretical and practical implications are evaluated along with limitations concerning population and ecological validity. Finally, suggestions for future research are presented.

PAGE 19

7 Chapter II Review of the Literature The review of the literature is divided into four sections. Using an historic persp ective, the first section discusses the major theories (e.g., Kanter, 1976) that attempt to explain differential distribution of men and women in positions of authority and leadership. The second section examines empirical evidence of gender differences wi th regard to leadership behaviors. Specifically this section will establish the link between leadership style and gender stereotypes with regard to verbal and non verbal behaviors believed to express masculine and feminine leadership styles. The third sec tion summarizes studies examining the relationship between gender and the perception of leadership effectiveness and leadership ability within non academic as well as academic environments. The fourth section discusses methodological issues. Specifically, the development and use of a written vignette/scenario format and other alternative formats are explored along with measurement issues regarding operational definitions that capture the measures of leadership ability and effectiveness. Section I Gender Role Models Based on their research in 1994 on managers and secretaries, Vinnicombe and Colwill found that over 95% of secretaries in the North America and Western Europe were women, whereas over 95% of senior managers were men. In general, women are ove r represented in low status jobs and under represented in high status occupations.

PAGE 20

8 These subspecialties within male oriented professions are usually rated lower in terms of status and prestige by members of that profession. With the exception of nursing, occupations traditionally associated with females tend to have male supervisors in the highest positions. Most researchers agree that one of the most important reasons that women are not equally represented in organizational leadership positions is that women are socialized into specific gender roles (e.g., Eagly, 1987; Payne, Fuqua, & Canegami, 1998). Payne et al., for example, stated that the socialization process promotes differential gender expectations for men and women in many aspects of social inte ractions. With regard to the workforce, it is believed that men are expected to exhibit greater qualifications for leadership than do women. Kruse and Wintermantel (1986), in their review of gender stereotyping in leadership positions, summarized what othe r researchers have stated repeatedly (e.g., Rosenkrantz et al., 1968), which is that stereotypes of how men differ from women match the general perception of how leaders differ from followers. In their review of the gender stereotype research literature, K ruse et al. noted that, in general, men are believed to be aggressive, independent, objective, active, dominant, competitive, and decisive, while women are believed to be gentle, emotional, sensitive, dependent and submissive. These gender beliefs are gene rally accepted to be normal and healthy by mental health professionals and have been established over many generations (Kruse et al., 1986). As a result of gender typing of leadership positions, Terborg (1977) suggested that women needed to adopt masculin e qualities if they wanted to succeed in management.

PAGE 21

9 Extensive research on gender stereotypes in social behaviors (e.g., Broverman et al., 1972) has established that the majority of the beliefs that people hold about the differences between men and women c an be described along the personal/task dimension (Eagly, 1987). The personal dimension involves behaviors that are of a caring and nurturing nature. Such behaviors include concern for the welfare of others, loving children, being affectionate, able to dev ote self completely to others, eager to soothe hurt feelings, helpful, kind, and sympathetic. Additional traits included in the personal dimension describe a persons emotional expressiveness (e.g., expresses tender feelings), interpersonal sensitivity (e. g., aware of feelings of others), and personal style (e.g., social orientation, femininity). It is believed that women exhibit personal oriented behaviors more strongly than men (e.g., Eagly, 1987). In contrast, men are more strongly associated with the t ask dimension. Behaviors consistent with the task style include tendencies to be assertive, controlling, aggressive, ambitious, dominant, forceful, and acting as a leader. Additional attributes include independent, self reliant, self sufficient as well as self confident and feeling superior. A task oriented person acts in a direct and adventurous manner and does not give up easily. The task/personal dimension has also been labeled communal/agentic, and socially/instrumentality oriented. The current study wi ll refer to this dimension as task/personal. Physical Characteristics as They Relate to Task/Personal Dimension Ecological theory has suggested that male and female physical attributes may be directly linked to task and personal qualities. Deaux and Lewi s (1984) found empirical evidence that masculine descriptions such as strong and sturdy were associated with the

PAGE 22

10 task style, while feminine descriptors such as graceful and soft were linked to the personal style. Deaux et al. stated that traditionally, wom en in society have been socialized to be nurturing, likeable, affectionate, soft spoken, warm, yielding, selfless, gentle, compassionate and dependent, while men have been traditionally socialized to be ambitious, aggressive, dominant, self reliant, strong individualistic, and independent. According to Klenke (1996), attributes and behaviors associated with women typically are inconsistent with leadership behaviors. Schein (1972) was one of the pioneers who explored gender stereotypes in organizational se ttings by presenting a list of 92 male and female characteristics to a sample of managers. Results of her study suggested that both men and women managers believed that a successful manager possessed a higher degree of masculine traits as compared to femin ine traits. Gutek (1993), in their review of stereotypes of women in management, found that Scheins findings in the 1970s were still valid in the late 1980s. In the late 1980s, women in management were perceived as less aggressive and independent than me n in management. However, studies also showed that women were seen as having higher degrees of interpersonal skills (e.g., Brenner, Tomkiewicz, & Schein, 1989; Frank, 1989). Gutek concluded that these results were compatible with the fact that women are so cialized to be cooperative while men are socialized to be competitive. Gender role spillover Phillips and Lord (1982) proposed that in organizational settings people develop specific expectations about appropriate leadership behaviors which take preceden ce over more gender based expectations. Gender role spillover theory (Gutek & Morasch, 1982) rejects that notion, and instead stipulates that gender

PAGE 23

11 roles continue to carry over into organizational settings. Specifically, Gutek et al. maintain that women a nd men are not regarded as generic managers, but instead, are evaluated by both their gender and their position within the organization. Gender role spillover affects women in leadership positions differently than men because of peoples expectations about leader appropriate behaviors more closely match prototypical masculine qualities. According to numerous researchers (e.g., Kruse & Wintermantel, 1986; Ragins & Sundstrom, 1989) women who want to be successful in their leadership positions have to adopt ma le like qualities in their leadership style. This managerial model (Terborg, 1977) however forces women to violate conventions concerning appropriate female behavior, and as a consequence, women are evaluated less favorably than their male counterparts. In addition to being more negatively evaluated, women in leadership roles, who adopt a masculine style, may also be perceived to be more extreme in their behavior than male leaders. Gender role congruence Nieva and Gutek (1981) proposed a more detailed set of predictions in terms of leadership evaluation. Their theory maintains that leaders who behave consistent with their gender role expectations are evaluated more favorably than leaders who behave in a manner inconsistent with their gender role expectation The gender role congruency model predicts that women who choose a feminine leadership style (gender congruent) are predicted to be more positively evaluated than women who choose a masculine or counter stereotypic leadership style. Empirical studies show mixed results. For example, Eagly, Makhijani and Klonski (1992) reported that women who displayed a masculine leadership style were more negatively evaluated than women who displayed a more feminine leadership style. However, Petty and Buning (1980) faile d to

PAGE 24

12 find support for the gender congruency hypotheses and found that male and female leaders who showed consideration were evaluated as equally effective. Nieva and Gutek propose that women who lead in a style congruent with their gender role may reduce t he role conflict that is experienced when occupying a leadership position, as well as escape the negative evaluations that are experienced when behaving counter to gender role expectations. Similarly, Eagly and Karau (1991) suggested that gender congruent leadership behaviors may ease the role conflict women experience in managerial positions and lead to a less stressful situation. Social role theory Social Role theory as it pertains to leadership positions (e.g., Baron & Kenny, 1986), stated that in orde r to consistently account for gender differences in leadership, moderator and mediator variables have to be considered. For example, Deaux and Major (1987) suggested that the degree to which a job emphasized gender differentiated skills and abilities could act as a moderator variable. Specifically, the authors predicted that women may be attracted to a job that requires primarily personal oriented activities, whereas men may be more attracted to jobs requiring predominantly task oriented activities. Some su pport for this notion was reported by Wood (1987) who devised a group of tasks that were either categorized to require complex personal oriented activity, or that called for predominantly task oriented activities. These tasks were then assigned to all male and all female groups. Results indicated that female groups were superior in their performance of tasks requiring personal oriented skills while male groups excelled at task oriented projects. Social role theory predicts that women will experience role co nflict when taking on leadership roles because conventions

PAGE 25

13 regarding appropriate female behaviors conflict with the general expectation that leaders behave in a masculine, or task oriented style (Eagly, 1987). Section II Observed Gender Differences in Leadership Styles Eagly and Blair (1990) noted two approaches to the study of gender differences in leadership positions. The first approach is represented in books and articles written by management experts who base their information on their own experi ences in organizations, as well as on personal interviews with men and women employed in leadership roles. The second approach is based on formal inquiries by social scientists. Most experts who base their information on experience maintain that men and wo men in leadership positions behave in accordance to their gender role. For example, books on the practice of management (Loden, 1985) have argued that women prefer to manage in a style that allows for cooperation and collaboration between managers and sub ordinates. This style gives less control to the leader, and encourages problem solving based on intuition, empathy, as well as rationality. Sargent (1981) agrees that even though men and women in management behave according to gender stereotypes, he argued that a movement towards androgyny would improve both men and womens performance as leaders. On the other hand, empirical research data do not show conclusive evidence of that claim. Reviews from empirical research (e.g., Bartol & Martin, 1986) have led to the general consensus that there are few differences between males and females with regard to leadership style. Eagly and Johnson (1990) point out, however, that Bartol and Martin (1986), for example, based their generalizations on a sample of only eigh t studies. Furthermore, relatively informal methods were used to draw these conclusions

PAGE 26

14 (summarizing research findings). To remedy this approach, Eagly et al. used meta analytic techniques in their review of the leadership literature in order to come up wi th a systematic, quantitative integration of available research. The authors found that the majority of the studies examined four aspects of leadership styles. These four aspects were usually organized into the leader style dimensions of task/personal and democratic/autocratic. The task/personal leadership dimension was first coined by Bales in 1950 and further developed by the Ohio State studies on leadership (e.g., Halpin & Winer, 1957). These studies maintained that the task dimension included behaviors such as maintaining high standards for performance, and making leader and subordinate roles explicit, while the personal leadership dimension included behaviors such as being friendly and helpful to subordinates. The democratic/autocratic leadership dimen sion has its roots in experimental leadership studies (e.g., Lewin & Lippitt, 1938), and is generally considered to define a narrower aspect of the task/personal dimension (Bass, 1981). Eagly et al. along with other researchers state that the democratic/au tocratic dimension relates to gender stereotypes in the same manner as the task/personal dimension, that is, men are believed to be more dominant and controlling (i.e., more autocratic) than women. Other variables coded for each study were publication for mat, possible confounding of gender with age, education, and management seniority. Finally, three types of study settings were examined in the meta analysis: (a) organizational studies within educational, business and government settings; (b) assessment st udies using college undergraduate and graduate students as research participants; and (c) experimental studies using college undergraduate and graduate students.

PAGE 27

15 Eagly et al. examined all instances in the literature that pertained to the assessment of the task/personal dimension, as well as the democratic/autocratic dimension. Measuring these dimensions included the use of direct observation (coding leadership behaviors as they occurred), as well as assessing leadership styles indirectly using measures of a ttitude or personality questionnaires. Other important criteria called for the samples to be drawn from a typical population of the United States or Canada (i.e., adults). Finally, meta analytic research called for sufficient information in order to calcul ate effect size estimates for gender with regard to the leadership dimensions. Using these criteria, Eagly et al.s meta analytic review included 162 studies. All studies, published and unpublished, were conducted between 1961 and 1987 with a median date o f 1981. Eagly et al. predicted that gender differences in leadership styles would be less pronounced in studies using organizational settings compared to experimental environments. Beyond this prediction the authors maintained that the purpose of their in quiry was primarily descriptive and exploratory. Eagly et al. found that in general, leadership styles as computed by weighted means across all types of style were statistically significantly stereotypic, however the effect sizes of those results were very small. When computing the means for type of leadership style separately, no statistically significant gender difference was found for the task dimension as well as for the personal dimension. Calculated means revealed a significant gender difference when comparing men and women on the personal leadership dimension such that women tended to use the personal leadership style more often than their male counterpart. The calculated effect size for this difference was very

PAGE 28

16 small ( d= 0.03) compared to the average effect size range of 0.00 to 1.2 within the area of social and personality psychology research (Eagly, 1987). Gender differences were also found along the democratic/autocratic dimension revealing that women tended to act in a more democratic style. The c alculated effect size for this estimate after outlier removal was small ( d = 0.27). Eaglys hypothesis that gender differences with regard to leadership style are less pronounced in organizational settings than in experimental and assessment studies was su pported in the meta analytic findings. Specifically, with regard to the task style, gender comparisons in organizational studies were statistically significantly less stereotypic than results obtained in assessment and laboratory studies. However, when com paring the personal and task leadership styles to the democratic and autocratic leadership styles, study settings had no impact on the outcomes. Based on the results of their meta analytic findings, Eagly et al. concluded that gender differences in leaders hip behaviors follow a more complex pattern than previously suggested. Results suggested that, consistent with findings from research on gender differences in social behaviors (e.g., Eagly, 1987; Hall, 1984), womens leadership styles were more democratic than mens regardless of research setting which may reflect underlying gender differences in either social skills, or may indicate subtle status differences men and women occupy within the same organizational role (Eagly et al., 1990). On the other hand, w ithin organizational settings men and women did not differ with regard to task leadership styles, which may indicate that within organizational hierarchies, leadership roles take precedence over gender roles.

PAGE 29

17 Section III Perception of Gender Difference s in Leadership Positions and Leader Evaluation Many researchers speculate that the lack of representation of women in higher levels of leadership in organizations may be because womens credentials and performance are not fairly evaluated (e.g., Nieva & Gutek, 1981). Fiske, Bersoff, Borgida, Deaux, and Heilman (1991) argued that gender discrimination of women in leadership roles is due to the influence of gender stereotypes on leadership performance. In an effort to examine gender discrimination, Eagly, M akhijani and Klonsky used meta analytic procedures to evaluate the issue of whether women are more negatively evaluated than their male counterparts. Their meta analytic research was confined to experimental research that controlled leadership characterist ics and behaviors while manipulating leader gender. Eagly et al. reported that experimental studies reviewed, either presented their information in a written vignette format, or trained confederates to act out roles. Eagly et al.s predictions for the meta analysis were based on gender role theory (Eagly, 1987), that is, people develop sets of expectations of what is considered appropriate behavior for roles within organizational settings (i.e., leader or subordinate). However, this theory maintains that ge nder based expectations have carryover effects similar to the gender role spillover theory (Gutek & Morasch, 1982) in that peoples expectations of leaders are influenced by their expectations of gender appropriate behaviors. Based on this theoretical fr amework, Eagly et al. predicted that woman leaders would be evaluated more negatively than their male leaders. Based on this framework the authors also predicted that the negative evaluation of female leaders would be most pronounced when female leaders us e an autocratic and directive style. Finally, the authors

PAGE 30

18 predicted that male research participants would be more prone to evaluate female more negatively than their male counterparts because this pattern was evident in numerous studies. This meta analysi s included 61 experimental studies with a median publication date of 1980. The majority of research participants in the included studies were college undergraduates, and business management and other graduate students. Most of the studies portrayed the org anizational context using examples of business, manufacturing, educational, or small groups vignettes. Leadership styles portrayed in the reviewed studies included examples of the task/personal dimension and democratic/autocratic leadership dimension. Th e majority of outcome measures of those studies consisted of rating scales for leader competence, leadership style, as well as satisfaction with leader. Results indicated a significant gender effect when comparing leader competence and leader satisfactio n with perceptions of leadership style. Specifically, Eagly et al. found that when measuring leader satisfaction and leader competence, research participants showed a greater tendency to evaluate female leaders more negatively compared to male leaders. As predicted, male research participants showed a greater preference for male leaders than female research participants. Inconsistencies emerged because results indicated that female leaders were perceived to be significantly more task style oriented than mal e leaders. A priori contrasts indicated that men who displayed an autocratic style were perceived more favorably than task oriented women, providing support for the gender role congruency theory that women are negatively evaluated when they display a male oriented leadership style. The fact that women tended to be perceived more negatively when displaying an autocratic or task type leadership style fits with the

PAGE 31

19 notion that a directive and autocratic style most contradicts the traditional female patterns of accepted behaviors. Comparing men and women along the masculine/feminine dimension produced a larger difference between men and women with regard to the masculine dimension compared to the feminine dimension. Men leading in a feminine manner were not perc eived negatively relative to women, however masculine women were more negatively evaluated than masculine men. Finally, studies showed a trend that male leaders were preferred over female leaders most often when men were a majority within that environment. This trend did not emerge when there was equal male and female representation by both genders. This result is consistent with gender role spillover theory that states that women who violate gender role expectations are more negatively evaluated. Inconsist ent findings however indicate some inadequacies of gender theories. For example, results of this meta analysis indicated that the negative evaluation of female leaders occurred not when directing male subordinates but when directing female subordinates. Ma le and female research participants favored female leaders with male subordinates and preferred male leaders with female subordinates. Eagly et al. concluded that in general men were evaluated more favorably than women. However, the calculated effect si ze estimate for this trend was weak (0.07). More specifically, the effect size estimate for gender differences in leader satisfaction was 0.10, while gender differences with regard to leadership competence produced an estimated effect size of 0.09. Gender and Leader Competence. Even though a number of social and organizational psychologists (e.g., Denmark, 1993) have argued that women are not less effective than men in leadership roles, some

PAGE 32

20 authors (e.g., Hollander, 1993) have stated that there may be som e situations when female leaders appear to be more effective, while male leaders appear to be more effective in other situations. Bass (1990) argued that male leaders are more favorably evaluated with regard to competence because of existing gender stereot ypes. Hunt (1991) defined leader competence in terms of the ability of a leader to facilitate group or organizational goals. There is some degree of consensus among organizational psychologists (e.g., Hunt, 1991) that competence should be regarded as an ou tcome of a leaders ability. In order to assess the level of leader competence numerous criteria have been suggested. For example, empirical studies have assessed the level of competence by measuring: (a) group or organizational productivity, (b) subordina te leader satisfaction, and (c) leadership performance as well as reputational ratings using the perspectives of superiors, subordinates, peers, and leaders themselves (Eagly et al., 1995). In their review of the literature on leadership competence, Eag ly, Karau and Makhijani (1995) found that the majority of studies were conducted within an organizational setting examining male and female managers in comparable leadership positions. A smaller body of literature investigated leadership competence using e xperimental designs with college students as research participants. The purpose of Eagly et al.s meta analysis was to provide a systematic and quantitative synthesis of existing research in order to address the general question of whether male and female leaders differed in their degree of leadership competence, as well as to evaluate possible moderating conditions that could influence competence ratings for men and women. In order to predict gender differences in leadership competence the authors used soc ial role theory (Eagly, 1987). That theory states that women experience increased

PAGE 33

21 degrees of conflict in their leadership roles because existing social pressure to behave consistent with the female gender role is inconsistent with the characteristics of th e masculine behaviors the leadership role traditionally demands. The more leadership roles are male dominated, the higher the degree of role conflict in women. According to this theory, female managers are predicted to be viewed more competent if they adop t a feminine leadership style, than women who adopt a masculine leadership style. Support for social role theory is evident in the results of Eagly and Johnsons meta analysis (1990), which showed evidence that female leaders led in a more democratic and p articipatory leadership style than their male counterpart. Meta analytic findings provided evidence that women were most negatively evaluated when they occupied male dominated positions, or when women were evaluated by men (Eagly, Makhijani, & Klonsky, 199 2). The authors included 96 studies with a median publication date of 1980 (range: 1962 1992) in their meta analysis of which 70% were organizational studies. Careful consideration was given to the classification of measures of competence. Competence mea sures were defined as objective or subjective. Objective measures included information about the managers expert knowledge in his/her field, as well as measures of production goals met by subordinates. Subjective competence measures included rating scales assessing leader performance, ability, effectiveness, satisfaction with leader, and effort and motivation. In order to examine the interaction between the perceived degree of gender stereotypic nature of leadership roles with perceived leadership effectiv eness of men and women, the authors conducted a separate study to measure the degree of perceived gender stereotypic content of leadership roles. The resulting match between gender roles and leadership roles was then defined along the

PAGE 34

22 dimension of gender c ongeniality. Specifically, leadership roles that were judged to require female stereotypic qualities (e.g., interpersonally oriented qualities of the role) were defined as female congenial while the male congenial aspects of leadership roles had been judge d to require masculine qualities (e.g., task oriented qualities). Results of this meta analysis indicated that in general women and men did not differ in their leadership effectiveness. No gender differences were found when comparing objective performance measures with subjective measures, or when comparing results from organizational studies with experimental studies. However, studies conducted within the military organization produced statistically significant outliers, that is, military studies deviated strongly from all of the studies in the meta analysis. Within the military organization men fared statistically significantly better with regard to the competence measure. The mean weighted effect size estimate was 0.42. A statistically significant model was produced when leadership type was examined according to the categories of first level (line level) and second level (middle level) management. Specifically, men were perceived to be more competent when occupying line level leadership roles (effect siz e estimate: 0.19), while women were perceived to be more competenct when managing at the middle level (effect size estimate: 0.18). When considering the gender congeniality dimension, results of this meta analysis showed a relationship between the perceive d degree of male stereotypic content of a leadership role and the perceived leadership competence rating. Thus, women who occupied increasingly male congenial leadership roles were perceived to be less competent than their male counterparts. Analogously, w omen were judged to be

PAGE 35

23 more competent than their male counterpart when the leadership role was perceived to require female stereotypic qualities. In their conclusion, Eagly et al. underscored that overall women did not differ from men in terms of competen ce evaluations. However, the authors also point out that even though there were no differences found overall, men were judged to be more competent leaders in leadership roles considered to possess a high degree of masculine qualities, while female leaders tended to be judged as more competent in leadership roles considered to possess a high degree of feminine qualities. Results of the military setting provided evidence for the gender role spillover model that states that gender based expectations carry over into the work place (Nieva & Gutek, 1981). Eagly et al. argued that men were rated to be more competent in line management while women were preferred in middle management. These results fit the gender congeniality dimension when examining the skills invo lved for line and middle management. Thus, line level management can be described to be male congenial because it requires the involvement of a relatively high degree of technical skills, while middle management fits the dimension of female congeniality be cause of the emphasis on human relations skills. Gender and Leader Likeability Rojahn and Willemsen (1994) examined the sex role congruency theory in the context of leadership. In their pilot study they found that male participants were more likely to st ereotype behaviors than female participants. Thus, the authors hypothesized that gender role appropriate behaviors would be evaluated favorably by men. The experimental scenario consisted of two male and two female students assigned to do a team class pro ject. Via a written description, the authors created the positions of team

PAGE 36

24 leader for each female and male member of the group. They also manipulated the scenario to include descriptors of task and personal leadership style behaviors. Four versions were cr eated: two conditions representing a male task leader (gender appropriate) and two conditions representing a female task leader (gender inappropriate). Similarly, two personal male conditions (gender inappropriate), and two personal female conditions (gend er appropriate) were created. Examples of the task condition described the leader to be pragmatic, suggesting that differences be settled by drawing lots, and a suggestion by the leader to encourage the weakest member (always opposite gender to leader) to drop out. The personal condition described the leader to be sensitive to the conflicts among group members, making considerable efforts to maintain good relations with members and offer to help the weak member of the group in order to avoid having the pers on drop out. Participants included 342 female and 154 male Dutch undergraduate psychology students who were randomly assigned to one of four experimental conditions. After reading the scenario, participants filled out questionnaires concerning leader compe tence, leader likeability, and leader personality traits. Three questions were asked for leader competence (1. Will the task be finished on time? 2. Will the manager maintain group morale? 3. How does the manager contribute to the goal?), and leader likeab ility (1. How much would you like to work for the manager? 2. How comfortable would you feel working for him/her? 3. How likeable do you find the manager?). Leader personality traits consisted of 29 items describing stereotypical male and female traits. Ma le adjectives included active, ambitious, cool, dominant, independent, influential, persistent, taking charge, determined, forceful and self confident, balanced, convincing, demanding,

PAGE 37

25 efficient, and full of initiative. Female adjectives included appreciat ing, considerate, cooperative, helpful, social, sympathetic, tolerant, understanding, warm, reliable and talkative. All dependent variables were arranged on a scale ranging from 1 (not at all) to 100 (extremely). The between subjects factors included leade r gender, leadership style, and participant gender. Results revealed that men judged leader competence differently depending upon leader gender. Specifically, female task leaders were judged to be less competent than their male counterparts. Similarly, fem ale personal leaders were judged to be more competent by male participants than her male counterpart. The fact that male research participants rated both male and female gender role inappropriate leaders more negatively with regard to leader competence was unexpected. Overall, results for leader competence supported the gender role congruency theory for male participants only. Female research participants were not influenced by leader gender in their evaluation of leadership measures. Other findings showed that overall, personal style leaders were judged to be as competent as the task style leaders. Personal leaders were rated to be more likable and more competent with regard to maintaining group morale. Estimated effect sizes for leader competence and likea bility were comparable to the calculated effect size estimates of meta analytic research (e.g., Eagly, Makhijani, & Klonsky, 1992; Swim, Borgida, Maruyama, & Myers, 1989). Many organizations provide women with leadership training in an effort to overcome the status disadvantage women face when taking a leadership position (e.g., Brown, Dovidio & Ellyson, 1990). However, some researchers claim that these strategies can potentially backfire because they may cause men to like women less as leaders, and thus may be less likely influenced by women who underwent leadership training. For

PAGE 38

26 example, Carli (1990) found that when women were perceived to be likeable they were also perceived to exert more influence over men than when they were perceived to be less likea ble. Carli, LaFleur, and Loeber (1995) examined different strategies that optimized the relationship between perceived likeability and influence given a male or female audience. Specifically, the authors examined whether the dominant nonverbal style was a more effective strategy for men than women when addressing a male audience. Further, the authors examined whether the combination of projecting warmth and competence produced better results especially for a male audience. Carli et al. also evaluated possib le interaction effects between gender, influence, perceived competence, and likeability. Based on the premise that women in the Western culture are more often associated with a lower social status than men (e.g., Berger, Rosenholtz, & Zelditch, 1980), Car li et al. hypothesized that male research participants would perceive a dominant female leader to be more threatening and less likeable compared to her male counterpart. Based on previous research the authors predicted that the degree of influence for wome n would depend on whether she was perceived to be likeable, while they predicted that the perception of likeability of male leaders would not impact the perception of influence. A preliminary study (95 male and 114 female undergraduates) was conducted to d etermine a gender neutral topic that was interesting to both men and women. As a result, the topic of the universitys current student meal plan was selected. The authors composed an argument in favor of the current meal plan and trained four (two male and two female) student confederates as a part of a dyad member. Confederates were trained to deliver a positive argument in favor of the student meal plan using either a dominant, submissive, task, or social style. The dominant style was characterized by spe aking in a loud voice,

PAGE 39

27 pointing at the other person in an intrusive manner, staring at the other person while talking, and maintaining stern facial expression. The submissive style was characterized as speaking in a soft pleading tone, using many hesitatio n and stumbles in the speech, slumped body posture, nervous hand gestures and avoiding eye contact. The social style was characterized as speaking in a moderately loud voice, having relaxed body posture with a body leaning towards the listener, friendly fa cial expression, and moderately high amount of eye contact. The task style was defined as speaking rapidly in a firm tone of voice, using few hesitations in the speech, having an upright body posture, using calm hand gestures, and displaying a moderately h igh amount of eye contact. The other student of the confederate dyad posed as a student listening to the speaker. For the study, 80 male and 80 female students from introductory psychology classes were recruited. Students were informed that the purpose of the study was to examine first impressions and subsequent group interactions. Thus, before meeting their selected partner for the subsequent campus issue debate, students were asked to watch a video of that student member and to provide information about t heir impression of the student they watched in the video. Using a nine point Likert type scale, research participants rated the following statements: (a) how well they could work with the speaker and (b) the extent to which the speaker was likeable, trustw orthy, competent, persuasive, powerful, knowledgeable, confident, condescending, influential, anxious, intelligent, intimidating, threatening, group oriented, friendly, and believable. Research participants were also asked to indicate whether the person li stening to the speaker on the tape was male or female. This question was asked to determine whether participants held stereotypes about nonverbal styles in conjunction with listeners. A factor analysis on the adjectives extracted five

PAGE 40

28 factors: likeability ( a =.81), competence ( a =.78), power ( a =.88), threat ( a =.85) and anxiety ( a =.72). A factorial ANOVA using gender of participant, gender of speaker and nonverbal style was used to analyze the data. The influence of the speaker with regard to participants opi nion change was measured by how well participants agreed with the speakers message. Results revealed that speakers using the task or social nonverbal style were more influential than those using the submissive or dominant styles in changing participants attitudes. Male participants rated female task speakers to be less influential than their male counterparts. Regression analysis used opinion change as the dependent variable and the estimated factor scores as predictor variables. Results showed that highe r levels of competence and likeability were associated with greater influence, while higher levels of threat were associated with reduced influence. Male participants did not weigh competence differentially with regard to male and female speakers. However, likeability seemed to be more important for female speakers in terms of influencing a male audience. Female participants were not influenced by the likeability variable. Other results indicated that social speakers were liked more than task speakers, who were liked more than submissive speakers. Dominant speakers were liked the least. Dominant speakers were rated to be more threatening and higher in power than task and social speakers. Male participants liked male task speakers better than female task spe akers. However, male participants did not differ in their likeability ratings of dominant male and female speakers. Finally, contrary to predictions, men did not perceive dominant female speakers to be more threatening than dominant male speakers. Interest ingly, female participants rated male task speakers to be more threatening than female task speakers, but did not differentiate between dominant male and female speakers. Also, a statistically

PAGE 41

29 significant association was found between the assigned gender o f listener and speaker style. Research participants were more likely to believe that submissive and dominant speakers were addressing men, and that social and task speakers were addressing women. The authors suggested that persons are generally threatened by competent opposite sex speakers more so than by competent same sex speakers. And even though likeability and competence were important predictors of influence, likeability was a more important predictor of influence with a male audience when dealing wit h female speakers. Results demonstrated that both male and female speakers benefited from the use of a social oriented style. Interaction Between Rater Gender and Leader Gender Evidence in meta analytic reviews of leadership evaluation indicate that male research participants have a greater tendency to evaluate female leaders more negatively than female research participants (e.g., Eagly et al., 1992). However, a recent study by Luthar (1996) found evidence contradicting these previous research findings. The authors experiment examined the relationship between research participant gender, leader gender, and leader style (democratic or autocratic) on the evaluation of leadership ability and leadership performance. Based on published research, Luthar hypoth esized that male research participants would tend to evaluate male leaders more favorably than female leaders, while female research participants were predicted not to show any gender bias in their evaluation of leadership ability and performance. In his s tudy, Luthar randomly assigned 130 female and 160 male undergraduate business students to the following vignette conditions: a democratic or autocratic male or female general manager described as facing the problem to increase the number of membership hold ers in a health

PAGE 42

30 club. After reading one version of the vignette, research participants were asked to rate, using nine point Likert type rating scales, management style and leadership ability. Consistent with previously published studies, Luthar found that democratic managers were rated higher on leadership ability and performance than their autocratic counterparts. However, contrary to previous empirical evidence, this study revealed a rater gender bias for both men and women. Specifically, the interaction effect revealed an opposite rater gender trend, that is, male research participants rated male managers more positively, while female research participants rated female managers more favorably. Male participants did not perceive a difference between male and female autocratic managers on leadership ability, while women rated autocratic male managers significantly lower compared to autocratic female managers on the leadership ability scale. Gender and Leadership Issues in Academia This group of studies exa mined perceptions of performance of teaching faculty within the academic environment. Studies examined the effects of instructor gender and teaching style on performance ratings. Other studies used experimental designs to manipulate teacher behaviors such as friendliness and smiling to examine their impact on teacher performance. After reviewing results of studies examining student evaluation of teacher performance, Unger (1979) concluded that there was no correlation between student achievement and evaluat ion of instructors. Instead, ratings appeared to be a reflection of student affective reactions to personal characteristics of instructors. For example, Wittrock and Lumsdaine (1977) suggested the non task related instructor characteristics

PAGE 43

31 of warmth, popu larity, reputation and enthusiasm had a significant impact on student ratings of teacher performance. Wittrock et al. suggested that a students own set of values and frame of reference impacted the way instructors were evaluated. Research regarding gender role stereotypes and evaluation bias (e.g., Broverman, Vogel, Boverman, Clarkson, & Rosenkrantz, 1972) provided evidence that instructor gender influenced student evaluations of teacher performance. Harris (1976) found that instructors with a feminine st yle were perceived to be less competent than instructors with a masculine style. Unger (1979) examined the effect of instructor gender on responses to course evaluation questionnaires. The author recruited 40 members of a psychology department of an urban northeastern liberal arts college. After elimination of minority faculty, the sample included 12 women and 26 men ranking from instructor to full professor. Evaluations were administered during the last few weeks of the semester for each class that instruc tors taught. For each instructor averages were calculated for teaching effectiveness and perceived grading difficulty. Ranks for grading difficulty were correlated with ranked items for teaching performance. Results revealed that while no correlation was f ound between teaching effectiveness and difficulty of grading among male instructors, a negative correlation between the two variables was established for female instructors. Women who were perceived as difficult graders received lower teacher effectivenes s scores than women instructors who were perceived as easy graders. Unger concluded that more demanding female professors were perceived as acting in a gender inappropriate manner, and thus were more negatively evaluated than their gender appropriate femal e counterparts.

PAGE 44

32 Similarly, a study by Martin (1984) revealed that students exhibited ambivalent expectations about female faculty. The highest rated female instructor in Martins research had a combination of feminine qualities such as friendliness, suppor tiveness and warmth as well as the traditionally masculine quality of preparedness. Kierstead, DAgostino, and Dill (1988) explored how traditionally female qualities such as warmth and friendliness influenced students in their rating of the instructor. In their first study, the authors constructed a course vignette that described behaviors of friendliness such as out of class socializing (i.e., having lunch with students from the class). While the educational characteristics (such as type of course, office hours kept and classroom behavior) of the vignette remained constant, the gender of the instructor was manipulated along with a set of friendly behaviors. A sample of 20 male and 20 female college students were asked to read through one version of the sce nario and to evaluate the instructor. Results revealed an instructor gender main effect, that is, male instructors were perceived more favorably than female instructors. In their second study, Kierstead et al. manipulated smiling behavior. They created a s lide tape presentation of a lecture on the anatomy of the eye. One male and one female model posing as instructors were trained for the smiling or non smiling condition. While the lecture was held constant, the smiling condition was manipulated during the presentation of slides. After 20 male and 20 female research participants watched the presentation, they were asked to rate the performance of the instructor, and were asked to indicate whether they would take a course with this instructor. Results reveale d an interaction between instructor gender and smiling; whereas smiling did not affect ratings for the male instructor, a smiling female instructor received statistically significantly more favorable ratings than the non smiling female

PAGE 45

33 counterpart. Female instructors were perceived more negatively when they deviated from stereotypical expectation (gender role congruency theory), whereas smiling behaviors did not impact ratings of male instructors. Section IV Conceptual Issues Leadership Dimensions The r eview of the literature on the differential impact of leadership style on the evaluation of leadership performance of male and female leaders revealed that most studies examined leadership styles along two dimensions: (a) task/personal, and (b) democratic/ autocratic (e.g., Eagly et al., 1992). Even though these dimensions are usually treated as independent of each other, Eagly et al. noted that the democratic/autocratic leadership style described a more precise aspect of the more general task/personal dimen sion. Luthar (1996) pointed out that despite different terminologies used in published leadership studies, the concept of the democratic/autocratic leadership style was virtually contained in all of them. Luthar pointed out that most leadership classificat ions used the democratic/autocratic leadership dimension either directly or indirectly. Many management consultants do not favor the autocratic leadership style (e.g., Naisbitt, 1982) because of its potentially demoralizing effect on employees. Even so, ma ny experts in the field of leadership believe that the effectiveness of a leadership style may be contingent upon situational cues (e.g., Wood, 1987). For example, House (1971) suggested that the task style could be more effective when tasks are not well d efined. A task style thus could overcome ambiguity by imposing structure. A personal style, on the other hand, could be more beneficial when work conditions are routine (Drenth & Koopman, 1984). However, Eagly et al. in their 1995 meta analytic review on

PAGE 46

34 l eadership effectiveness, pointed out that the empirical literature rarely investigated possible interaction effects between the task/personal and the democratic/autocratic leadership styles on leadership evaluations. One study conducted in 1983 by Remland, Jacobson and Jones manipulated leader gender, leadership style, and type of situation the exchange occurred to measure possible interactions among those variables. Specifically, Remland et al. examined the effect of research participants psychological ge nder (measured by Spence and Helmreichs Personal Attributes Questionnaire) on the performance evaluation of nontraditional gender role behaviors of male and female managers. The authors hypothesized that psychological gender (masculine, feminine, and andr ogynous) of male and female research participants would interact with leader gender, leadership style, and type of situation on leadership measures. Research participants were 139 male and 150 female undergraduate students enrolled in a communication clas s. Prior to the experiment, participants filled out Spence and Helmreichs (1974) 24 item psychological gender orientation questionnaire. After participants were categorized according to their psychological gender (gender typed masculine or feminine, andr ogynous, and undifferentiated) they were randomly assigned to one of six experimental conditions. The authors described the six conditions as versions of a written vignette in which a manager (male or female) interacted with a male subordinate about a task problem (how to improve productivity output), followed by a personal problem (how the subordinate could deal with a family conflict). The authors created a total of six experimental conditions. In versions one and two, a task vignette was presented follow ed by a personal vignette, with a supportive acting male (female) manager in both scenarios (the authors called these versions SS). Versions three and four described a non supportive

PAGE 47

35 male (female) manager in the task oriented vignette, and a supportive mal e (female) manager in the personal problem (NS for not supportive/supportive versions). Versions five and six described a non supportive male (female) manager in both the task and the personal vignettes (NN for non supportive versions in both scenarios). T he order of vignette placement in all six experimental conditions remained constant, with the task vignette presented first, followed by the personal vignette. All experimental conditions held subordinate gender within the dyad constant (male). This exper iment did not include an SN version to test a supportive manager (male or female) for the task problem, or a version portraying a non supportive manager (male or female) for the personal vignette. The nonverbal behaviors depicting a supportive or non suppo rtive manager for the vignettes consisted of nonverbal descriptions. The supportive manager was described as leaning forward, touching the subordinate, speaking in a soft voice, smiling, nodding, and maintaining high eye contact, while the non supportive m anager was described as leaning back, keeping physical distance, speaking in a firm voice, having a serious facial expression, interrupting, no eye contact, and turning away from the subordinate. Remland et al.s nonverbal descriptions of managers within d yads coincide with behaviors identified in the literature as democratic and autocratic (e.g., Eagly et al., 1990). The experimental design was a 2 (manager gender) x 3 (nonverbal behavior pattern: SS, NS, NN) x 3 (participant psychological gender: androgyn ous, gender typed masculine or feminine, and undifferentiated) x 2 (participant gender) between factors design. Research participants, after having been categorized according to their psychological gender, were randomly assigned to one of the six condition s. Manager performance in the vignettes was measured using 18 items that were rated on a seven point bipolar scale. These items

PAGE 48

36 captured two factors: task behavior and consideration of manager performance. It is not clear what type of rotation was used (o rthogonal or oblique) with the factor analysis, or which items were included for each scale in the final factor score estimates. Examples for the consideration factor as best discerned included the following items: Lowering employees confidence/raising em ployees confidence, concerned with employees approval/ not concerned with employees approval, degrading/upgrading, boosting employees ego/damaging employees ego, treating like an inferior/treating like a superior, poor/excellent, and appropriate/inapp ropriate. Items for the task factor included: clear/unclear, inefficient/efficient, organized/unorganized, and concerned with detail/unconcerned with detail. The internal consistency estimate for the consideration factor as assessed by Cronbach alpha was 91 and .75 for the task factor. Remland et al. hypothesized that psychological gender of the research participants would interact with the ratings of non traditional manager performance. Results, however, did not support this prediction. Instead, results o f this study revealed that managers were rated most considerate in the SS condition followed by the NS version. Managers in the NN condition were rated as the least considerate. Surprisingly, no statistical difference in terms of consideration was found be tween manager dyads in the SS and NS conditions. Manager dyads in the SS conditions were rated to be more task oriented than manager dyads in the NN conditions. Again, no statistical difference was found in terms of the perception of task style between the SS and the NS conditions. Finally, male research participants rated manager dyads as more considerate than female research participants. Although this study had several strong points, such as an experimental design, as well as a combination of leadership behaviors with task and personal situations, this study

PAGE 49

37 had several weaknesses. First, it is not clear why the authors combined the task oriented vignettes with the person oriented vignettes, thus reducing the conditions to SS, NS, and NN. Furthermore, af ter having combined the task vignette with the person vignette, the authors did not add a condition needed to test for a supportive manager in the task condition, and a non supportive manager in the personal condition (i.e., the SN condition). The order ef fect problem may be very prominent in this study because the authors always presented the task vignette before the personal vignette. Another potential problem was found with the vignette manager subordinate dyad. Specifically, the authors failed to vary s ubordinate gender within that dyad. With regard to the 18 item scale measuring the dimensions of consideration and task behavior there were several weaknesses. First, most published leadership research studies have measured leadership performance in terms of leadership productivity, effectiveness, likeability, competence, and influence (e.g., Eagly et al., 1992). The measures of this study were inconsistent with current measures and theories of leadership performance. It is unclear why items such as effecti ve/ineffective and inefficient/efficient were not used in the analysis of leadership effectiveness. The relatively weak internal consistency estimate for the task factor (.75) may be an indication that the items used do not form a strong cluster. Even thou gh it is not of interest for the present study, there is a question regarding how research participants were distributed into the categories of psychological gender. And finally, this study was conducted in the early 80s and may not be relevant in todays society, because as Gergen (1986) has pointed out, research results in social psychology are in part a reflection of societal values at a given point in history.

PAGE 50

38 Summary and Purpose of Current Study A large number of studies on gender differences in leadership behaviors and leadership evaluation have been summarized in the literature using meta analytic techniques (e.g., Eagly et al., 1990). Meta analyses have evaluated three important questions: 1) Is there a statistical difference with regard to the evaluation of leadership behavior in men and women, 2) do these differences vary as a function of various moderator variables that include types of situations such as social settings (e.g., Baron & Kenny, 1987), and 3) how important are those differen ces estimated via the calculation of effect sizes (e.g., Hedges & Olkin, 1985). Results of meta analyses (Eagly et al., 1990) have found that women in management positions tend to behave more democratically compared to men in those positions. Meta analytic results also indicated that female leaders were more negatively evaluated than their male counterparts, especially in highly masculine typed leadership roles such as the military (Eagly et al., 1995). In general, however, calculated effect size estimates with regard to gender differences in leadership behavior, and leadership evaluation were consistently small. Eagly and Wood (1991) pointed out that these effect sizes were comparable to those usually obtained in social behavior research. They argued that d ismissing these findings of gender differences as trivial could be misleading. Similarly, Eagly, Karau, and Makhijani (1995) have pointed out that even though there were no gender differences with regard to the evaluation of leadership competence, more in depth analysis revealed that gender did matter because gender appropriate leaders were rated to be more competent than leaders with gender inappropriate behaviors (these findings were not consistent, however). Small effect sizes may indicate a change in a ttitude towards women in leadership roles, due to the fact that

PAGE 51

39 more and more women enter all levels of managerial positions. Nevertheless, Eagly et al. cautioned not to ignore gender as a relevant issue since women still fare much worse than men when they occupy increasingly masculine typed leadership positions. Deaux and Major (1987) speculated that gender differences in social behavior are context dependent. Specifically, the authors argued that the context in which social interaction occurs influences the resultant behavior. In the review of the literature, gender differences in leadership behavior and leadership evaluation were examined most commonly along the task/personal dimension without integrating the context in which those behaviors were occurri ng. One experiment (Remland et al., 1983) manipulated the context as well as gender and leader behavior in their evaluation of gender differences in leadership evaluation. Even though the studys methodological weaknesses may have limited its results, Rem land et al. addressed a number of important questions related to the evaluation of male and female managers leading democratically or autocratically in a task or personal environment. By improving upon Remland et al.s study design, the present study aim s to contribute to the empirical and theoretical literature on the perception of leader gender incorporating the task/personal as well as the democratic/autocratic leadership dimensions. Furthermore, since studies have consistently found that male particip ants perceived female leaders more negatively than female participants, it was important to include rater gender into the design. This study also attempted to create a set of psychometrically sound leadership outcome measures, because most leadership studi es measured leadership outcomes using single items, or, if more than one item was used, they usually did not undergo any psychometric evaluation. This study psychometrically evaluated a set of leadership outcomes to come up with a set

PAGE 52

40 of psychometrically s ound factors. With regard to the study setting, most research on gender differences in leadership evaluations were conducted in non academic organizational settings. This study examined leadership evaluation within an academic setting, because most researc h of gender differences in teacher evaluation is dated (mid 70s to 80s). By operationally defining leadership roles along the leader dimensions of task/personal and democratic/autocratic, using a written vignette format, this study maximized the manipula tion of important leadership variables in order to examine hypothesized complex interaction effects, that otherwise could not have been possible.

PAGE 53

41 Chapter III Method This chapter is divided into three sections. The first section presents an overview of the study with the research hypotheses. Section II describes the research participants and the rationale for their selection. Section III describes the process used in the instrument development, along with the development of the outcome mea sures. Section I Overview The purpose of this experiment was to examine the effect of professor (leader) gender, professor nonverbal behavior (democratic/autocratic), and type of situation (task/person), on the evaluation of leader performance by male a nd female research participants. Written scenarios of a professor student interaction were created in this experimental design, to manipulate professor gender, professor nonverbal behavior (democratic/autocratic), and the situation in which the interaction occurred (task versus person oriented). Leadership evaluations (i.e., leader competence, masculinity/femininity, leader likeability) were provided by male and female research participants. The design of this experiment was a 2 (professor gender) x 2 (prof essor leadership style) x 2 (type of situation) x 2 (participant gender) between factors design. Hypotheses The following hypotheses were based both on theory and empirical evidence from previously published research.

PAGE 54

42 Main Effects Dimension of leader competence It was predicted that male and female undergraduate student research participants would rate democratic professors (male or female) higher on the competence scale (Table) than autocratic professors regardless of the situation (task or personal) See Figures 1 4 Dimension of leader masculinity/femininity. It was predicted that male and female research participants would rate autocratic professors more masculine than democratic professors. Dimension of leader likeability This study predic ted that all research participants (male and female) would rate the democratic professors higher on the likeability scale than the autocratic counterpart regardless of type of situation (task or personal). Interaction Effects Interaction between professo r gender, professor leadership style, type of situation, and participant gender with regard to competence and likeability ratings Figures 1 4 represent a visual depiction of the predicted four way interaction effects for the competence and likeability f actors. Specifically, the visual display illustrates the differential predictions for male and female research participants. Comparing Figures 1 and 2 with Figures 3 and 4 illustrates the prediction that male research participants will be similar to female research participants in their competence and likeability ratings in the democratic condition. In the autocratic conditions, however, male and female research participants are predicted to differ, that is, male research participants (but not female resea rch participants) would also be influenced by professor gender and type of situation when evaluating leader competence and leader likeability. Using gender role

PAGE 55

43 spill over theory, it was predicted that male research participants would rate autocratic femal e professors in a personal situation the least competent, followed by autocratic female professors in a task situation (Figures 1 4 ). Figure 1 Competence Ratings By Male and Female Participants for Professors Acting Democratically in a Task-Oriented Situation 1 2 3 4 5 Male Professor Female Professor Competence Female Research Participants Male Research Participants

PAGE 56

44 Figure 2 Competence Ratings By Male and Female Participans for Professors Acting Democratically in a Personal-Oriented 1 2 3 4 5 Male Professor Female Professor Competence Female Research Participants Male Research Participants Figure 3. Democratic Personal 3.9 3.5 3.8 4 1 2 3 4 5 Male Professor Female Professor Leadership Ratings Female Research Participants Male Research Participants

PAGE 57

45 Section II Research Parti cipants Because past leadership and gender research has generally revealed small to medium effect sizes it was determined that a large sample size was needed to achieve a level of power of .80 (Keppel, 1991). The sample size for this study, a 2 (professor gender) x 2 (professor leadership style) x 2 (type of situation) x 2 (participant gender) between factors design, was estimated using Stevenss power tables. Based on these power calculations it was determined that a minimum of 100 research participants p er cell (16 cells total) was needed, resulting in a sample size of 1600 participants. The research was conducted at the second largest metropolitan state university in the southeast. Out of the 37,500 enrolled students, approximately 27,000 students are en rolled in undergraduate programs in 10 Colleges (Table 1). This study selected the College of Arts and Science and the College of Education because of their size, range of departments and their willingness to participate. Table 1 Colleges Within the Parti cipating University Colleges Architecture and Community Design Arts & Sciences Business Administration Education Engineering Marine Science Medicine Nursing Public Health Visual and Performing Arts Note Research participant s were selected from Colleges marked with

PAGE 58

46 The sample consisted of 1653 undergraduate students from the College of Art and Science (n=931) and the College of Education (n=722). Table 2 compares the student race/ethnicity profile of the sample used in the study with the race/ethnicity profile of the total student population of the university. Racial distribution of the sample was similar to the University population. Table 2 Student Race/Ethnicity Profile of Sample and University Population Study Sampl e University Total Student Population N=1653 % N=37500 % African/American 147 9 1305 11 American/Indian 14 1 146 1 Asian 60 4 1924 5 Caucasian 1148 70 26798 71 Hispanic 164 10 3528 9 Other 119 7 1146 3 Note percentages we re rounded to whole numbers The study consisted of 1049 (63%) females and 604 (37%) males, as compared to 59% female and 41% male for the total student population. The average age for the sample was 21.92 years (SD=4.92 years) compared to 24 years (unknow n SD) for the undergraduate student population as reported for the Fall of 2001. Of the 931 research participants from the College of Arts and Science, 464 (50%) were males and 467 (50%) were females. Of the 722 research participants from the College of Education 140 (19%) were males, and 582 (81%) were females.

PAGE 59

47 Three departments within the College of Arts and Science, and three departments within the College of Education were used for data collection (Table 3). The goal was to access introductory classes in order to maximize research participant volunteers. Class sizes ranged from 30 to 200 students. Exact class sizes could not be established because student attendance varied. Table 3 Source Information of Study Sample College Department Number of Classes Class Size Range Arts & Science Geography Language/Linguistic Communication 3 10 10 100 200 10 40 20 40 Education Language/Acquisition Social Foundation Measurement 8 15 6 20 100 15 30 10 30 Total 52 Note. Class size ranges are given because exact numbers were not available Because most classes where data collection occurred were introductory classes for both colleges, the reported majors were more reflective of the college than the department where testing took place. Tables 4 and 5 list reported majors by gender separately for each College. The most frequent major for male and female research participants for the College of Arts and Science was Communication. Both genders were comparable in their reported majors.

PAGE 60

48 The majority of female research participants from the College of Education were Elementary education majors, whereas male research participants were not concentr ated in any one area. Table 4 Reported Majors for Male and Female Arts and Science Students (N=931) Males N=464 N % Females N=467 N % Communication 133 34 119 30 Criminology 47 12 19 5 ISS 44 11 32 8 Psychology 25 6 47 12 Languages 14 4 34 8 Biology 30 8 56 14 Geography 16 4 12 3 Geology 4 1 6 2 Mass Communication 18 5 37 9 Sociology 8 2 2 1 History 13 3 11 3 Classics 10 3 6 2 Chemistry 18 5 8 2 Political. Science 9 2 0 0 English 9 2 15 4 Note. Percenta ges were rounded to whole numbers.

PAGE 61

49 Table 5 Reported Majors for Male and Female Education Students (N=722) Males N=140 N % Females N=582 N % Elementary Ed 16 12 293 5 2 Special Ed 20 14 71 13 Early Child 3 2 65 12 Physical Ed 26 19 20 4 Math Ed 11 8 16 3 Social Sciences 19 14 24 4 Secondary Ed 16 12 25 5 Ed Leadership 2 1 2 1 Language Ed 5 4 5 1 English Ed 10 8 24 4 Adult Ed 1 1 5 1 Music Art Ed 10 8 12 2 Note. Percentages were rounded to whole numbers. There were differences between the samples from the College of Arts and Science and the College of Education in terms of age distribution, race/ethnicity, and distribution of school standing. With regard to age, results from an ANOVA revealed a main effect for college and gender. Specifically, research participants for the College of Education were statistically significantl y older than their Arts and Science counterparts, F (1, 1345)=93.53, p <.01. The mean age as reported by the research participants for the College of Arts and Science was 20.86 ( SD =3.79), while the mean age of research participants from the College of Educat ion was 23.36 ( SD =5.84). The effect size estimate for this difference was medium ( d =0.5). Results further revealed a statistically significant gender effect for age, F (1, 1345)=21.14, p <.01. Male research participants had a mean age of 22.20 ( SD =5.68), whi le female research participants had a mean age of 21.76 ( SD =4.41). The effect size estimate for this difference was

PAGE 62

50 very small ( d =0.09). No interaction effect between college and gender was found, F (1, 1345)=0.02, p >.05. Curiously, female participants from the College of Education had a higher tendency of omitting the question of age. Table 6 lists age summary statistics for male and female research participants for each college. Table 6 Summary Statistics for Age by College and Participant Gender Arts & Science Education Males (N=464) Females (N=467) Males (N=140) Females (N=582) Mean Age 21.55 20.20 24.53 23.10 Range 18 56 18 27 18 57 18 48 SD 5.1 1.67 7.10 5.50 Missing 79 69 31 120 Differences between the two samples of this study were also f ound in the distribution of school standing. Results of testing the differences in proportions indicated that the College of Arts and Science had a significantly higher proportion of freshmen, ? 2 (1, N =1653)=108.63, p <.01. The College of Arts and Science had a statistically significant higher proportion of sophomores than the College of Education, ? 2 (1, N =1653)=24.91, p <.01. On the other hand, the College of Education had a statistically significant higher proportion of juniors in this study than the Col lege of Arts and Science, ? 2 (1, N =1653)=47.36, p <.01. The College of Education also had a statistically significant higher proportion of seniors in this study than the College of Arts and Science, ? 2 (1, N =1653)=19.22, p <.01. Table 7 lists the frequency d istributions for class standing separately for each college.

PAGE 63

51 Table 7 Frequency Distribution of School Standing by College and Gender Arts & Science Education Class Standing Males N=464 N % Females N=467 N % Males N=140 N % Fema les N=582 N % Freshman 114 25 82 17 5 4 20 4 Sophomore 116 25 118 25 20 14 89 16 Junior 105 23 149 32 55 40 259 45 Senior 124 27 117 25 58 42 201 35 Missing 5 1 2 13 N=1653 Differences were found in the race/ethnicity profile of the two colleges in this study. Results of testing the differences in proportions indicated no statistically significant difference between the colleges in the frequency distrib ution of the American/Indian category, ? 2 (1, N =1653)=2.4.3, p >.05. A statistically significant difference was found in the proportion of African/American students, ? 2 (1, N =1653)=12.39, p <.01. The proportion of African/Americans in Arts and Science was statisticall y significantly higher than the College of Education counterpart. The proportion of Caucasians was statistically significantly higher in the College of Education than in the College of Arts and Science, ? 2 (1, N =1653)=62.74, p <.01. The College of Arts and S cience had a statistically significantly higher proportion of Asians than the College of Education, ? 2 (1, N =1653)=20.81, p <.01. Finally, the College of Arts and Science had a statistically significantly higher proportion of Hispanics than the College of E ducation, ? 2 (1, N =1653)=19.51, p <.01. Table 8 lists the frequency distribution of race/ethnicity by gender and separately for each college.

PAGE 64

52 Table 8 Racial/Ethnicity Frequency Distribution by College and Gender Arts & Science Education Race/Ethnicity Males N=464 N % Females N=467 N % Males N=140 N % Females N=582 N % African/American 47 10 56 12 10 7 34 6 American/Indian 3 1 2 1 2 1 7 1 Asian 22 5 29 5 1 1 8 1 Caucasian 296 64 277 59 114 81 461 79 Hispanic 62 14 57 12 3 2 42 7 Other 33 7 46 10 10 7 30 5 N=1 653 Section III Instrument Development Experimental Conditions Most research evaluating the effects of gender stereotypes on the evaluations of leaders have typically used written scenarios or vignettes (Rojahn & Willemsen, 1994). Consistent with th is paradigm, the present study used the vignette format in order to manipulate type of situation, professor gender, and professor leadership style, while keeping all other characteristics of the scenario constant. Specifically, two scenarios were created, one describing a task oriented situation, and the second describing a person oriented situation. Embedded within these scenarios was a professor student dyad, with the professor described as either male or female, and the professors interaction style with the student described as either democratic or autocratic. All other descriptions of the scenario were held constant. Crossing all possible combinations of the independent variables (situation, professor gender, professor leadership style) resulted in eigh t

PAGE 65

53 vignette versions (see Appendices C and D for complete set of booklet versions). Prior to the experiment, a pilot was conducted to obtain a measure of construct validity for the scenario manipulations. Appendix A details the procedures and results suppo rting the validity of the scenario manipulations. Task Situation A175 word vignette was created of a professor student dyad interaction (Appendix C) that involved a task oriented problem. Specifically, the scenario described a student, enrolled in a cours e taught by this professor, entering the professors office in order to discuss a solution to increase student participation during class. The student suggests access to the professors notes before class as a solution to the problem. Personal Situation A 175 word vignette was created that described a professor student dyad interaction that involved a personal situation (Appendix D). Specifically, the scenario described a student of the professor, entering the professors office in a state of emotional di stress revealing despair over a recent relationship breakup. Professor Nonverbal Democratic Leadership Style for Task Situations The nonverbal democratic leadership style between professor and student was manipulated using the following nonverbal descrip tors: smiling encouragingly, nodding his/her head while the student talks, leaning forward towards the student while listening to the student, gently tugging at his/her collar while listening to the student, and accompanying the student to the door at the end of the conversation (Appendix C).

PAGE 66

54 Professor Nonverbal Autocratic Leadership Style for Task Situations The nonverbal autocratic leadership style between professor and student was manipulated using the following nonverbal descriptors: frowning, sitting straight in his/her chair and staring at the student, leaning back in his/her chair, lowering his/her eye brows, shaking his/her head, pointing his/her finger at the student, and abruptly standing up and motioning the student to the door at the end of the conversation (Appendix C). Professor Nonverbal Democratic Leadership Style for Personal Situations The nonverbal democratic leadership style between professor and student for this condition was manipulated using the following nonverbal descriptors: noddin g encouragingly, sitting in a chair opposite the student, leaning forward towards the student, and folding his/her arms in his/her lap while the student tells the story. Finally, the professor remains seated and waits for the student to finish talking befo re offering to make an appointment (Appendix D). Professor Nonverbal Autocratic Leadership Style for Personal Situations The nonverbal autocratic leadership style between professor and student for this condition was manipulated using the following nonverba l descriptors: not paying attention to the student, displaying impatience, sighing, crossing his/her arms over his/her chest while the students speaks, and staring at the student (Appendix D). Professor Student Dyads Professor gender within every profes sor student dyad was manipulated while student gender remained unidentified. As a result, every type of scenario combination had a version for male and female professor student dyads.

PAGE 67

55 Development of Dependent Measures Meta analytic studies (e. g., Eagly, Makhijani, & Klonsky, 1992) have revealed that most studies measured leader evaluation using the following aspects of leadership performance: (a) competence, expertise, effort, productivity, and general evaluation, (b) leader likeability, desire to work with him or her, and group cohesiveness, and (c) leader power, authority, and influence. Most recent studies (e.g., Carli, LaFleur, & Loeber, 1995; Luthar, 1996) also included lists of adjectives measuring leader effectiveness, likeability, and pe rsonality traits. For example, Rojahn and Willemsen (1996) used a list of 26 gender stereotypical traits in their assessment of the perception of leader character traits. Based on measures of previously published studies, the current study ini tially selected 26 items for four leadership dimensions: leader competence (7 items), leader likeability (4 items), leader masculinity/femininity (4 items), and leader professional traits (11 items) (Table 9). These measures were adjusted to reflect an aca demic context. A draft version of the 26 items was initially reviewed by 10 experts (two male and three female professors in the College of Education, as well as three male and two female professionals in management positions). Reviewers were asked to chec k the 26 items for content and clarity, as well as to identify possible alternatives and suggestions for improving these leadership measures. Based on their feedback, two items (skillful/not skillful, productive/unproductive) under the leader competence fa ctor were dropped. Further, three items were reversed and the placement of items were randomly ordered rather than ordered under their concept. Table 9 lists the leadership items for each leadership construct.

PAGE 68

56 An initial pilot study was conduc ted within the College of Education (see Appendix B for a detailed description of procedures and results) to examine internal consistency reliability of the four leadership factors: leader competence (5 items), leader likeability (4 items), masculinity/fem ininity (4 items), and leader professional traits (11 items). The final 24 leadership items were arranged on a five point semantic differential rating scale. Low scores (1) were designed to represent the negative aspects of three of the factors (i.e., inco mpetence, not likeable, and negative professional traits). A low score on the masculinity/femininity scale represented femininity, while the high end of the scale (5) reflected masculinity. In addition to the 24 leadership items, four exploratory items (se e Appendix C or D) were presented about the professors behaviors in the scenario. These items asked the research participants to rate (using a 5 point Likert type scale) the professor regarding his/her potential for promotion, communication skills, level of respect, and level of likeability. The 5 point rating scale ranged from 1 (strongly agree) to 5 (strongly disagree), which were then reversed scored to reflect a high score with agreement. Eight booklet versions were created to accommodate the eight sc enario versions. In order to maximize ease of handling, a booklet was created by folding 17x11in paper to create a four page booklet of 8 1/2 x 11 in. The front page contained a reminder that participation was voluntary, followed by directions. Upon open ing the booklet, page two contained one version of the professor student interaction, followed on the next page (page three) by the 24 randomly ordered leadership items. Finally, the back page (page four) of the booklet contained four short comments abou t the professor, each paired with

PAGE 69

57 a five point rating scale. Finally, background information (participants gender, age, major, school standing, and race/ethnicity) was requested. Table 9 Initial Items for the Four Leadership Dimensions Leader Competenc e (7 items) Masculinity/Femininity (4 items) Leader Likeability (4 items) Professioanl Traits (11 items) Competent/incompetent Timid/forceful Critical/tolerant Powerful/powerless Effective/ineffective Soft/tough Considerate/Inconsiderate Hardwo rking/lazy Qualified/not qualified Aggressive/not aggressive Popular/unpopular Persistent/gives up easily Influential/not influential Dominant/submissive Likeable/not likeable Fair/not fair Capable/not capable Responsible/irresponsible Skillful/Not skillful* Not helpful/helpful Productive/unproductive* Cooperative/not cooperative Trustworthy/Untrustworthy Independent/dependent Objective/subjective Unprepared/prepared Note. Items marked with * were dropped in the final version Procedure Instructors from the two colleges were contacted via email with a letter introducing the researcher and the purpose of the study and a request for an appointment (Appendix E). Most appointments secured the permission of the researcher to ent er the classroom to administer the instruments. The researcher entered the classroom only by appointment and permission of the instructor at the beginning of class, or in some cases at

PAGE 70

58 the end of class. The majority of data collection was conducted by the researcher. However, if an instructor wanted to hand out the booklets him/herself, he or she was given an approximate 10 minute training session in an effort to standardize data collection. Before each testing session the researcher randomly ordered the bo oklet forms appropriate for that class. Generally, data were collected at the beginning of class. After being introduced by the instructor as a graduate student working on her dissertation, the researcher gave her standard verbal introduction with regard t o the purpose of the study, and the need for volunteer participants, and the anonymity of responses. Booklet forms were then distributed to research participants who indicated their willingness to volunteer by raising their hand. The researcher also stress ed that the nature of the scenario was a matter of personal opinions on the part of the student, and that there were no right or wrong answers. Participants were reminded not to put their names on the booklet. Upon receipt of the booklet, participants were instructed to first read the front page of the booklet for instructions, then to read the scenario, and fill out all the questions including the back page of the booklet that contained general background information about the participant. While working th rough the booklet, participants were asked not to communicate with their fellow students, and to keep their comments or questions until all the booklets had been collected. The majority of instructors granted extra time for the researcher to answer questio ns that were posed after the booklets had been filled out. Generally, students took about 10 minutes to fill out the booklets.

PAGE 71

59 Chapter IV Results This chapter is divided into three sections. Section I presents the analysis of the psychometric prope rties of the measures of leadership. Section II reports the results of the five hypotheses posed by this study. Section III presents exploratory findings of this study. Section I Psychometric Analysis of the Four Leadership Dimensions Based on results from the literature review and pilot study (Appendix B), 24 leadership items were created that used a five point semantic differential response scale. These items were designed to measure the following four leadership dimensions: Competence (5 items), Masculi nity/Femininity (4 items), Likeability (4 items), and Professional Traits (11 items). Competence Five items were proposed for this factor. Table 10 lists the summary statistics. For these items the low end of the five point scale reflected incompetence, while the higher end of the scale reflected competence (all five items were reversed for this purpose). Masculinity/Femininity This factor was represented by four items (Table 11). The lower range of the five point scale reflected femininity, while the h igh end of the scale reflected masculinity. Two items were reversed scored.

PAGE 72

60 Table 10 Summary Statistics for Proposed Competence Factor Item Mean SD Skewness Kurtosis Competent/Incompetent* 3.36 0.99 0.27 0.21 Effective/ineffective* 2.71 1.16 0 .25 0.77 Qualified/not qualified* 3.21 0.95 0.01 0.03 Capable/not capable* 3.45 0.96 0.23 0.19 Influential/not influential* 3.28 1.06 0.25 0.36 N=1652 Note: Items with an were reversed scored. Table 11 Summary Statistics for Proposed Masculi nity/Femininity Factor Item Mean SD Skewness Kurtosis Timid/forceful 3.59 0.96 0.46 0.02 Soft/tough 3.58 1.00 0.46 0.21 Aggressive/not aggressive 3.01 1.21 0.16 0.67 Dominant/submissive 3.72 0.98 0.37 0.40 N=1652 Note: Items with an wer e reversed scored. Likeability Four items were developed for this factor (Table 12). A low score on the five point response scale indicated not likeable, while the high end of the scale reflected a high degree of likeability. Three of the four items wer e reversed scored. Professional Traits This factor consisted of 11 items (Table 13). A low score on the professional traits items reflected the negative side of the trait, while a high score reflected the desired professional trait (nine items were rever sed scored).

PAGE 73

61 Table 12 Summary Statistics for Proposed Likeability Factor Item Mean SD Skewness Kurtosis Critical/tolerant 2.64 1.16 0.26 0.76 Considerate/inconsiderate 2.61 1.29 0.35 1.00 Popular/unpopular 3.18 1.14 0.16 0.68 Likeable/not likeable 2.81 1.27 0.13 1.00 N=1652 Note: Items with an were reversed scored. Table 13 Summary Statistics for Proposed Professional Traits Factor Item Mean SD Skewness Kurtosis Powerful/powerless* 3.59 0.92 0.27 0.11 Hardworking /lazy* 3.31 1.00 0.07 0.13 Persistent/gives up easily* 3.19 1.10 0.25 0.54 Fair/not fair* 3.10 1.00 0.05 0.45 Responsible/irresponsible* 3.36 1.00 0.26 0.48 Not helpful/helpful 2.64 1.20 0.25 0.98 Cooperative/not cooperative* 2.82 1.20 0.08 0.90 Trustworthy/untrustworthy* 3.21 1.00 0.04 0.45 Independent/dependent* 3.66 1.00 0.52 0.00 Objective/subjective* 3.10 0.98 0.07 0.13 Unprepared/prepared 3.23 1.10 0.16 0.67 N=1652 Note: Items with an were reversed scored. Principal ax is factoring specifying four factors Exploratory factor analysis was used to analyze the underlying structure of the 24 leadership items. Results of a previously conducted pilot study, in addition to the results from an extensive review of the literature, suggested a four factor solution (Competence, Masculinity/Femininity, Likeability, and Professional Traits). Based on these results, the initial exploratory factor analysis (principal axis procedure) used a four factor solution with an oblique rotation ( promax). This procedure involved the whole sample (n=1652).

PAGE 74

62 Initial estimates of communality were the squared multiple correlations of the correlation of each item with the remaining items. In order to determine whether the four factor solution was reas onable, the initial unrotated eigenvalues were inspected, and compared to (a) the expected eigenvalues generated by the null model with the same parameters, (b) the Kaiser criterion of retaining eigenvalues greater than one, (c) the average eigenvalue of t he four factor solution, and (d) logical interpretation of the factors. Table 14 lists the unrotated eigenvalues, the explained proportion of that eigenvalue based on the common variance solution, and the expected eigenvalues of the null model. Table 14 Eigenvalues for Principal Factor Extraction Method compared to Null Model Component Eigenvalue (n=1652) Proportion Null Model Eigenvalues 1 8.30 0.72 0.21 2 3.18 0.27 0.17 3 0.44 0.03 0.15 4 0.32 0.02 0.13 5 0.19 0.01 0.11 6 0.15 0.01 0.10 7 0.11 0.009 0.08 8 0.08 0.007 0.07 9 0.03 0.003 0.06 10 0.02 0.002 0.04 Average Eigenvalue=0.480

PAGE 75

63 Inspection of the eigenvalues did not support a four factor solution based on the above stated criteria. Specifically, the third and fourth eigenvalues of th is four factor solution were below the Kaiser criterion of one, as well as just below the average eigenvalue of 0.480 (of the common solution). Yet, the values of the third and fourth eigenvalues were slightly higher than the eigenvalues calculated by the null model (Table 14). Since the four factors correlated moderately high with one another, both the pattern matrix (standardized regression coefficients) and structure matrix (zero order correlations) were examined (see Table 15). In general, items with factor loadings larger than 0.30 are considered strong factor loadings (Rummel, 1988). In light of this general rule, several of the leadership items that exhibited a high amount of cross correlations in the structure matrix (correlations), retained their item cross loading pattern when examining the standardized regression coefficients (pattern matrix in Table 15). Specifically, nine out of the 24 items exhibited moderately high cross loadings (items marked with an in Table 15). The factor loadings o f the 24 leadership items were examined for (a) the magnitude of loadings on the factors, and (b) the magnitude of cross loadings on secondary factors. Using .30 or larger as an indication of a strong factor loading, 11 of the 24 items loaded most strongly on the first factor. Similarly, the second factor consisted of 11 items. Factor three had two items with strong loadings, while none of the 24 items loaded strongly on the fourth factor. Further, as stated earlier, nine of the 24 items had strong cross l oading tendencies on at least one other secondary factor.

PAGE 76

64 Item clustering of the four factor solution was compared with the originally proposed clustering of the four leadership factors (competence, masculinity/femininity, likeability, and professional tr aits). Factor one combined four out of the five competence items with seven of the professional traits items. Factor two combined all four masculinity/femininity items along with all four items originally suggested to belong to the likeability construct. I n addition, factor two also retained three items originally proposed to belong to the professional traits cluster. Finally, two professional trait indicators loaded strongly on factor three. None of the leadership items had a strong association with factor four. The originally proposed four dimensions of leadership were not supported by the results of the four factor solution using the current sample of 1652 research participants. Instead, these results suggested a two factor solution underlying the 24 le adership items. Thus, given these results the principal axis factor analysis with oblique rotation (promax procedure) was rerun with two factors specified. Principal axis procedure with two factors The two factors extracted accounted for approximately 99 % of the common variance (see Table 16 for information on pattern and structure matrices). Two leadership items (independent/dependent, objective/subjective) were dropped from both factors even though their factor loadings satisfied the factor loading crit erion of .30 (.34 for independent/dependent and .38 for objective/subjective). Specifically, these items were dropped because their factor loadings were substantially lower than the factor loadings of the remaining 22 leadership items. In addition, one ite m (critical/tolerant) was eliminated because of its strong negative factor loading ( .60) on factor two. This

PAGE 77

65 item was originally proposed to belong to the likeability factor (Table 12), and reversing the direction would have seriously impeded the interpre tation of this factor such that the high score would have linked likeable, popular, and considerate with critical. As a result, 21 leadership items remained in the solution. Table 16 summarizes the pattern matrix and structure matrix from the two factor s olution. Table 15 Factor Pattern and Structure Matrices for Four Factor Solution Pattern Matrix (Standardized Regression. Coefficients) Structure Matrix (Correlations) Item 1 2 3 4 1 2 3 4 Incompetent/competent 56 .10 9 29 65 43 51 30 Ineffective/effective 64 24 4 12 71 52 47 15 Qualified/not qualified 75 9 2 31 69 24 45 9 Capable/not capable 59 6 26 7 73 28 63 5 Influential/not influential 18 10 48 0 54 33 64 5 Critical/tolerant 5 67 3 6 21 63 15 9 Con siderate/inconsiderate 41 61 0 11 67 76 47 1 Popular/unpopular 4 61 32 5 42 71 50 20 Likeable/not likeable 19 67 10 14 53 81 45 28 Timid/forceful 2 67 32 10 5 58 10 23 Soft/tough 4 73 1 9 33 76 26 25 Aggressive/not aggressive 20 75 7 2 7 65 5 19 Dominant/submissive 8 76 36 2 1 61 16 12 Powerful/powerless 21 58 45 3 27 35 40 13 Hardworking/lazy 49 6 23 1 61 21 53 3 Persistent/gives up easily 67 37 4 11 55 10 35 21 Fair/not fair 40 45 13 2 67 66 54 11 Responsible/irrespo nsible 68 9 0 31 71 44 50 31 Not helpful/helpful 54 46 3 16 72 64 47 8 Cooperative/not cooperative 32 55 9 11 62 69 48 0 Trustworthy/untrustworthy 41 29 13 30 61 57 51 36 Independent/dependent 6 29 41 5 20 12 35 1 Objective/subjective 34 1 4 1 36 15 26 0 Unprepared/prepared 78 21 19 1 57 5 26 9 N=1652 Note: Items with an items have strong cross loadings.

PAGE 78

66 The rotated eigenvalue of factor 1 was 6.8, and explained 59% of the common varianc e in the solution. Out of the 21 leadership items, 14 items loaded primarily on this factor. This factor captured the following leadership items: (a) all five items from the competence factor (Table 10), (b) eight items from the professional traits factor (Table 13), and (c) one item from the likeability factor (Table 12). In order to name this factor appropriately, item factor loadings were judged in terms of their strength, as well as their level of cross loading on the second factor. Based on these crit eria, the five original competence items loaded most prominently (unique factor loadings ranged from .62 to .80) on this factor, while at the same time exhibiting minimal cross loading for the second factor (Table 16). Next, the considerate/inconsiderate l eadership item from the originally proposed likeability factor loaded strongly on this factor (factor loading of .55), while at the same time loading almost as strongly on the second factor (factor loading .47). Finally, of the eight leadership items origi nally proposed to belong to the professional trait factor, four items (persistent/gives up easily, fair/not fair, not helpful/helpful, and cooperative/not cooperative) exhibited strong factor loadings on both factors, while the other four professional trai ts items (hardworking/lazy, responsible/irresponsible, trustworthy/untrustworthy, and unprepared/prepared) had factor loadings for only the primary factor (unique factor loading ranged from .59 to .72 ). The purpose of factor rotation is to obtain factors that consist of items having strong and meaningful factor loadings on only one factor. Results of this factor analysis were mixed. Specifically, out of the 14 items loading on factor 1, the five competence items (Table 10) exhibited strong and meaningful factor loadings for this factor only. Similarly, four out of the eight professional trait items also had strong primary factor

PAGE 79

67 loadings and minimal secondary factor loadings. On the other hand, the remaining four professional traits indicators (persistent/ gives up easily, fair/not fair, not helpful/helpful, cooperative/not cooperative), as well as the one likeability indicator (considerate/inconsiderate) had strong factor loadings on both factors (Table 16). Table 16 Factor Pattern and Structure Matric es for the Two Factor Solution Pattern Matrix (Standardized Regression Coefficients) Structure Matrix (Correlations) Item 1 2 1 2 Incompetent/competent 67 8 7 0 33 Ineffective/effective 67 18 73 43 Qualified/not qualified 73 14 67 13 Capable/not capable 80 16 74 14 Influential/ not influential 62 3 28 62 Critical/tolerant 6 60 72 68 Considerate/inconsiderate 55 47 55 64 Popular/unpopular 35 5 0 64 75 Likeable/not likeable 42 59 64 75 Timid/forceful 15 69 11 63 Soft/tough 18 68 44 75 Aggressive/not aggressive 11 72 16 68 Dominant/submissive 23 76 6 67 Powerful/powerless 48 64 24 46 Hardworking/lazy 68 17 62 9 Persistent/gives up easily 64 46 47 22 Fair/not fair 60 33 73 56 Responsible/irresponsible 72 7 75 34 Not helpful/helpful 62 31 73 54 Cooperative/not cooperative 52 40 67 60 Trustworthy/untrustworthy 59 26 69 48 Independent/dependent 34 33 22 20 Objective/subjective 38 6 36 9 Unprepared/prepared 60 26 50 3 N=1652

PAGE 80

68 Because of their highly ambiguous cross loadings, these five items were dropped from facto r 1. The remaining nine items in the factor solution were further examined in terms of their conceptual contribution. Based on this criterion the content of the five competence items (incompetent/competent, ineffective/effective, qualified/not qualified, capable/not capable, influential/not influential) demonstrated a clear and cohesive structure. Next, of the four professional traits items that loaded highly on this factor (i.e., hardworking/lazy, responsible/irresponsible, trustworthy/untrustworthy, unpr epared/prepared) one item was eliminated (unprepared/prepared) because its content did not contribute to a meaningful interpretation of this factor (it described a personal style rather than the desired general quality of leader competence). To summarize, factor 1 originally had 14 items, of which six were eliminated from the solution. The theme of leader competence was clearly expressed in the content of these items, thus, factor 1was named competence (Table 17). The competence factor response scale ranged from 1 (not competent) to 5 (competent). The rotated eigenvalue of factor 2 was 4.69 and explained approximately 40% of the common variance in the solution. Of the 21 leadership items, seven items loaded primarily on this factor (see pattern matrix in Table 16). Specifically, the following leadership items identified this factor: (a) all four items originally proposed as the masculinity/femininity factor (timid/forceful, soft/tough, aggressive/not aggressive, dominant/submissive), (b) two of the items o f the originally proposed likeability factor (popular/unpopular, likeable/not likeable), and one item from the original professional traits factor (powerful/not powerful). Like in factor 1 (competence), items for this factor were examined in terms of the strength of their unique factor loadings, as well as

PAGE 81

69 their factor loadings on the secondary factor (Table 16). Based on these criteria, the four original items from the masculinity/femininity factor exhibited the highest factor loadings (ranging from .68 t o .76), while loading minimally on the competence factor. Next, the two items from the likeability factor (popular/unpopular, likeable/not likeable) had relatively strong factor loadings on this factor, but also exhibited strong negative factor loadings on the competence factor. Similarly, the powerful/not powerful item, originally part of the professional traits cluster, showed a similarly strong double loading (Table 16). In order to maximize factor interpretation, the seven items for this factor were ex amined not only for their factor loadings but also for their contribution to factor interpretability (i.e., high cross loading on the other factor and content). Based on these criteria, the four items from the masculinity/femininity scale (timid/forceful, soft/tough, aggressive/not aggressive, dominant/submissive) not only had high factor loadings on this factor, but in addition had minimal cross loadings on the competence factor. Unfortunately, the remaining three items (popular/unpopular, likeable/not lik eable, powerful/not powerful) strongly loaded on this factor as well as on the competence factor, which at best would have complicated the interpretation of this factor. Therefore, these three items were dropped from this factor. As a result, Factor 2 emer ged with the original core items from the masculinity/femininity factor. As a result, this factor was named masculinity/femininity. The five point scale ranged from (1) femininity to (5) masculinity. Table 17 lists the summary statistics for these two fact ors.

PAGE 82

70 Table 17 Factor Names and Indicator Items Factor 1 Competence (8 items) Factor 2 Masculinity/Femininity (4 items) Incompetent/competent Timid/forceful Ineffective/effective Soft/tough Qualified/not qualified Aggressive/not aggressive Capable/ not capable Dominant/submissive Influential/not influential Hardworking/lazy Responsible/irresponsible Trustworthy/untrustworthy N=1652 Table 18 Summary Statistics for Competence and Masculinity/Femininity Factors Total N=1653 Males N=604 Fe males N=1049 Competence M (8 items) SD Skewness Kurtosis Alpha 3.24 0 .76 0.01 0.14 .88 3.29 0.75 0.03 0.08 ---3. 22 0.77 0.00 0.26 ----Masculinity/ M Femininity SD (4 items) Skewness Kurtosis Alpha 3.50 0.81 0.10 0.66 0.79 3.43 0.77 0.0 3 0.49 ----3.54 0.83 0.16 0.74 ----Note Correlations between competence and masculinity/femininity for the total sample was .24, for males the correlation was .16, and the correlation for females was .28.

PAGE 83

71 Comparison of two factor result s with originally proposed four factors Originally, four leadership factors were proposed: competence (five items), masculinity/femininity (four items), likeability (four items), and professional traits (eleven items). Results from the current study revea led evidence for two factors: competence (eight items) and masculinity/femininity (four items). The competence factor included all originally proposed competence items. In addition, three leadership items from the originally proposed professional traits f actor loaded strongly on this factor. Overall, the five competence items along with the three items originally attributed to professional traits represented the construct of leadership competence. The masculinity/femininity factor of the current study incl uded all four of the original items developed for this factor. Even though two items from the original likeability factor and one item from the original professional trait factor also loaded strongly on this factor, they were dropped because of strong fac tor cross loadings on the competence factor. To conclude, four factors underlying the leadership construct were initially proposed. However, results from this study did not support four factors underlying the leadership construct. Instead, leadership was represented by two factors: leader competence (eight items) and leader masculinity/femininity (four items). These two factors consisted of all the originally proposed core items. In addition, three items from the professional traits concept emerged as it ems clustering with the competence factor. Surprisingly, the likeability factor in its proposed form did not hold up in this study.

PAGE 84

72 Overview of Confirmatory Factor Analysis Procedure (CFA) The purpose of the CFA was to examine the fit of the two factor solution and to evaluate sources of misfit (e.g., correlated error for the items). Figure 5 illustrates the hypothesized two factor leadership model that was developed based on the result of statistical (i.e., exploratory factor analysis) and conceptual c riteria. The measurement portion of the model consists of the causal arrows that go from the latent factors (competence and masculinity/femininity) to the manifest variables that measure them. Specifically, competence had eight manifest indicators, while m asculinity/femininity was represented by four indicator variables. While the two latent factors were specified to intercorrelate (as noted by the double headed arrow in Figure 5) the indicators for each factor were specified to only load on their designate d factor. In order to scale both latent variables, the loading for the first item for each factor was set to 1.0. For each of the 12 indicator variables measurement error variances were estimated (independence of error was assumed). CFA was conducted usi ng the PROC CALIS procedure available within the SAS statistical package. The raw data were used to create a covariance matrix by using the COV option. The data were analyzed using the maximum likelihood estimation procedure. Examination of skewness and ku rtosis (Tables 10 & 11) indicated that most of the variables showed some degree of kurtosis, however their levels of deviation stayed within the moderate range of +/ 1. As a rule of thumb, levels of kurtosis exceeding 1.5 could compromise the validity of t he maximum likelihood procedure (Bollen, 1989).

PAGE 85

73 Results of the ? 2 goodness of fit test indicated a statistical significant lack of fit, ? 2 (1615, 53)=850.16, p <.001. Since the, ? 2 goodness of fit test is greatly influenced by sample size (Bollen, 1989), sample independent fit indices have become better indicators for model fit. Of those, one of the most frequently reported and discussed fit indices is the root mean square error of approximation (RMSEA), which is an inferential approach assessing model dis crepancy relative to the degrees of freedom (MacCallum, 1986). Guidelines suggest a good fit when RMSEA <.05, a moderate fit to range between .05 and .08, and a poor fit when exceeding .10. The RMSEA of this analysis was .09, which indicated a moderate to poor fit. Another sample independent index routinely reported is the comparative fit index (CFI). This fit index is based on the noncentrality parameter and is scaled from 0 to 1, with values greater than .9 indicating an acceptable fit. The CFI result o f .90 indicated a minimally acceptable fit. Results showed that leader competence correlated moderately negative with leader masculinity (r= .35). The standardized loading estimates for the eight competence items ranged from .61 to .81, while the four masc ulinity/femininity items had standardized loading estimates ranging from .67 to .76. The Lagrange/Wald test modification indices suggested that one source of model misfit was that five indicator items for the competence factor had secondary loadings on th e masculinity/femininity factor, while two of the masculinity/femininity indicators had secondary loadings on the competence factor (see Table 19).

PAGE 86

74 Table 19 Summary Statistics of the Seven Largest Lagrange Chisquare Change Statistics Original Factor Indicator Item Reduction in ? 2 If Crossloading With Factor Competence Ineffective/effective Responsible/irresponsible Influential/not influential Capable/not capable Hardworking/Lazy 112.56 61.15 51.80 47.70 37.90 Masculi nity/ Feminity Masculinity/ Femininity Timid/forceful Aggressive/Not aggressive 224.53 76.70 Competence Four Factors Revisited The purpose of the extensive series of exploratory and confirmatory factor analytic procedures was to derive leadership fac tors that had a high degree of construct validity and that were internally consistent. Based on pilot testing and results from the literature review, four factors underlying the leadership construct were proposed: (a) competence (five items), (b) likeabili ty (four items), (c) masculinity/femininity (four items), and (d) professional traits (eleven items). However, empirical evidence did not support a four factor leadership model, instead, a two factor leadership structure emerged: (a) competence (eight item s), and (b) masculinity/femininity (four items). The empirically derived competence factor consisted of the originally proposed competence items. In addition, three items from the professional traits factor loaded on this construct as well. The masculinity /femininity factor that emerged as a result of testing consisted of the original four masculinity/femininity items. As for the proposed professional traits factor, three out of the eleven items loaded strongly on the competence factor.

PAGE 87

75 The likeability fact or as proposed did not hold as a distinct factor (see Table 12 for summary statistics). Three of the four items (considerate/inconsiderate, popular/unpopular, likeable/not likeable) had strong dual loadings on the competence and masculinity/femininity fac tors, while one item (critical/tolerant) was eliminated because of its lack of conceptual fit. Despite the dual loadings for the three likeability items on the competence and masculinity/femininity factors, these three items also showed moderately strong intercorrelations (Table 20). Further the internal consistency of the three items was mo derately strong (a= .84). Given this moderate high alpha and the fact that likeability was a focus of this study, the researcher decided to combine these items (likeable/not likeable, popular/not popular, considerate/inconsiderate) to form a likeability sc ale. The scale ranged from (1) not likeable to (5) very likable. Table 20 shows summary statistics for the three leadership factors. Table 20 Likeability Indicator Item Intercorrelations Item 1 2 3 1. Likeable/notlikeable 1.0 2. Popular/not popular .66 1.0 3. Considerate/inconsiderate .70 .57 1.0 N=1652

PAGE 88

76 Table 21 Summary Statistics for Leadership Factors (N=1652) Factor Number of Items Mean SD Skewness Kurtosis Alpha Competence 8 3.24 0.76 0.01 0.14 .88 Masculinity/ Femininity 4 3.49 0.81 0.10 0.66 .78 Likeability 3 2.75 1.10 0 .23 0.87 .84 Note. The correlation between competence and masculinity/femininity was .24, and between competence and likeability was .69. The correlation between masculinity/femininit y and likeability was .63. To conclude, extensive factor analytic procedures established two internally consistent factors of leadership: competence ( a =.88) and leader masculinity/femininity ( a =.78). In addit ion, a three item likeability factor was created that demonstrated reasonable psychometric properties ( a =.84).

PAGE 89

77 P2 P 3 P4 P5 P 6 P7 P8 1 P10 P11 P12 Figure 5. Two factor confirmatory factor analysis model for leadership measures Competence F1 S 2 F 1 Incompetent/competent Ineffective/effective k 1 S 2 k 1 k 2 S 2 k 2 Qualified/not qualified Capable/not capable k 3 S 2 k 3 k 4 S 2 k 4 Influential/not influential k 5 S 2 k 5 Hardworking/laz y k 6 S 2 k 6 Responsible/irresponsible k 7 S 2 k 7 Trustworthy/untrustworthy k 8 S 2 k 8 Masculinity/Femini nity F2 F 1 F 2 Timid/forceful k 9 S 2 k 9 Soft/tough Aggressive/not aggressive k 10 S 2 k 10 k 11 S 2 k 11 Dominant/submissive k 12 S 2 k 12 S 2 F 2 1

PAGE 90

78 Section II Hypotheses Testing The purpose of this experiment was to examine a set of five hypotheses in relation to the three leadership constructs of compe tence, masculinity/femininity, and likeability. The five hypotheses were tested using a 2 (professor gender) x 2 (professor behavior) x 2 (type of situation) x 2 (participant gender) between factors design. To enhance the interpretation of the statisticall y significant main and interaction effects, Cohens effect sizes were calculated for statistically significant main and interaction effects as measured by Cohens (1988 ) f metric: f effect = (df effect )F effect /N G uidelines for the f metric describe effect sizes of 0.10 as small, 0.25 as medium, and 0.40 as large. Effect sizes were also calculated for follow up pairwise comparisons using Cohens d metric : d =( Sample Mean 1 Sample Mean 2 ) /SD pooled Guidelines for Cohe ns d metric describe the effect sizes of 0.2 as small, 0.5 as medium, and 0.8 as large. Finally, in order to enhance external validity the hypothesis factors were examined separately for the College of Arts and Science (N=931) and the College of Educatio n (N=722). Before testing hypotheses using analysis of variance, assumptions for this procedure have to be examined. The first assumption of the analysis of variance is that

PAGE 91

79 the outcome scores are independent within treatment groups, as well as independe nt between treatment groups (Keppel, 1991). This is not only an assumption for the analysis of variance, but a requirement of an experimental design. It is reasonable to assume that this study did not violate the independence assumption because (a) test booklets were randomly assigned to students within a classroom, and (b) students completed the instruments independently and were exposed to the test for about 10 minutes. The second assumption of the analysis of variance procedure states that individual t reatment populations, from which the sample is drawn, are assumed to be randomly drawn and have a normal distribution indicated by the level of skewness (0) and kurtosis (0) (Keppel, 1991). This study examined this assumption for the three leadership outc ome variables separately for the samples from the College of Arts and Science and the College of Education, as well as by booklet form and participant gender (Tables 22 33). Results show that all three leadership factors in both colleges had a level of ske wness ranging from a minimum of 0.05 to 1.0, while the level of kurtosis ranged from 0.01 to 2.23. These ranges indicate that the normality assumption was likely violated. However, effects of leptokurtic or platykurtic distributions on the nominal alpha level (Type I error) are slight (Winer, Brown, & Michels, 1991).

PAGE 92

80 Table 22 Descriptive Statistics for Male Research Participants in the College of Arts and Science for Leader Competence Form N Mean SD Skewness Kurtosis 1 60 3.27 0.78 0.12 0.14 2 54 3.38 0.85 0.12 0.59 3 61 3.03 0.80 0.19 0.14 4 57 3.45 0.66 0.35 0.19 5 58 3.25 0.77 0.21 0.59 6 63 3.46 0.88 0.10 0.89 7 56 3.14 0.59 0.81 2.23 8 55 3.58 0.63 0.43 0.80 Note. Form 1 = Autocratic Female Professor in Personal Condi tion; From 2 = Democratic Female Professor in Personal Condition; Form 3=Autocratic Female Professor in Task Condition; Form 4 =Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form6=Democratic Male Professor in Personal Condition; Form7=Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition. Table 23 Descriptive Statistics for Female Research Participants in the College of Arts and Science for Leader Competence Form N Mean SD Skewness Kurtosis 1 58 3.04 0.91 0.11 0.68 2 54 3.49 0.71 0.59 0.07 3 60 3.14 0.55 0.02 0.17 4 60 3.37 0.71 0.17 0.01 5 58 3.14 0.83 0.21 0.24 6 59 3.68 0.71 0.08 0.37 7 61 3.06 0.59 0.42 0.02 8 57 3. 44 0.77 0.83 2.03 Note. Form 1 = Autocratic Female Professor in Personal Condition; From2=Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Cond ition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition.

PAGE 93

81 Table 24 Descriptive S tatistics for Male Research Participants in the College of Education for Leader Competence Form N Mean SD Skewness Kurtosis 1 14 3.18 0.54 0.45 0.98 2 20 3.28 0.64 0.30 0.33 3 14 2.80 0.68 0.69 0.29 4 20 3.44 0.62 0.43 0.14 5 20 2.66 0.68 0.0 7 0.22 6 16 3.42 0.78 0.18 0.60 7 19 3.24 0.66 0.39 2.12 8 17 3.41 0.45 0.33 1.45 Note Form 1 = Autocratic Female Professor in Personal Condition; From 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form4=Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form6=Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democ ratic Male Professor in Task Condition. Table 25 Descriptive Statistics for Female Research Participants in the College of Education for Leader Competence Form N Mean SD Skewness Kurtosis 1 77 3.00 0.73 0.15 0.67 2 75 3.38 0.77 0.45 0.19 3 69 2.88 0.63 0.34 0.35 4 71 3.35 0.63 0.40 0.51 5 72 2.88 0.76 0.08 0.52 6 73 3.50 0.91 0.31 0.64 7 72 2.91 0.66 0.53 0.77 8 73 3.27 0.78 0.25 0.01 Note : Form 1 = Autocratic Female Professor in Personal Condition; From 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form4=Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition ; Form7=Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition.

PAGE 94

82 Table 26 Descriptive Statistics for Male Research Participants in the College of Arts and Science for Leader Masculinity Form N Mean SD Skewness Kur tosis 1 60 3.59 0.65 0.56 0.01 2 54 3.01 0.59 0.20 0.32 3 61 4.01 0.56 0.37 0.35 4 57 3.37 0.68 0.52 0.04 5 58 3.35 0.63 0.34 0.04 6 63 2.80 0.63 0.31 0.01 7 56 4.06 0.56 0.02 0.97 8 55 3.16 0.77 0.12 0.69 Note Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professo r in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition. Table 27 Descriptive Statistics for Female Research Partici pants in the College of Arts and Science for Leader Masculinity Form N Mean SD Skewness Kurtosis 1 58 3.85 0.62 0.38 0.32 2 54 2.78 0.67 0.60 0.26 3 60 4.24 0.54 0.68 0.55 4 60 3.32 0.66 0.38 0.52 5 58 3.59 0.68 0.11 0.08 6 59 2.76 0.55 0. 34 0.45 7 61 4.21 0.62 0.83 0.45 8 57 3.37 0.73 0.16 0.42 Note. Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor i n Task Condition.

PAGE 95

83 Table 28 Descriptive Statistics for Male Research Participants in the College of Education for Leader Masculinity Form N Mean SD Skewness Kurtosis 1 14 3.78 0.46 1.31 2.31 2 20 2.60 0.40 0.65 1.91 3 14 4.37 0.62 0.63 1.06 4 20 3.41 0.57 0.04 1.26 5 20 3.38 0.81 0.14 0.44 6 16 2.90 0.40 0.25 1.00 7 19 4.25 0.46 0.05 0.57 8 17 3.23 0.75 0.73 0.21 Note. Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democratic Female Professor in Personal Condi tion; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic M ale Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition. Table 29 Descriptive Statistics for Female Research Participants in the College of Education for Leader Masculinity Form N Mean SD Skewness Kurtosis 1 77 3.83 0.6 6 0.63 1.11 2 75 2.71 0.72 0.38 0.08 3 69 4.24 0.60 0.74 0.12 4 71 3.36 0.67 0.01 0.23 5 72 3.76 0.66 0.30 0.14 6 73 2.81 0.58 0.16 0.01 7 72 4.23 0.55 0.79 0.66 8 73 3.38 0.64 0.11 0.86 Note. Form 1 = Autocratic Female Professor i n Personal Condition; Form 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition.

PAGE 96

84 Table 30 Descriptive Statistics for Male Research Participants in the College of Arts and Sc ience for Leader Likeability Form N Mean SD Skewness Kurtosis 1 60 2.58 1.10 0.47 0.76 2 54 3.29 1.00 0.23 0.47 3 61 2.13 0.94 1.13 1.08 4 57 3.16 0.85 0.19 0.91 5 58 2.74 1.01 0.28 0.36 6 63 3.58 0.85 0.46 0.10 7 56 2.11 0.70 0.28 0.78 8 55 3.28 0.93 0.19 0.48 Note. Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form6=Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition. Table 31 Descriptiv e Statistics for Female Research Participants in the College of Arts and Science for Leader Likeability Form N Mean SD Skewness Kurtosis 1 58 2.34 1.12 0.59 0.57 2 54 3.61 0.87 0.85 0.39 3 60 1.94 0.66 0.78 0.27 4 60 2.95 0.89 0.02 0.04 5 58 2 .58 0.89 0.63 0.22 6 59 3.97 0.88 0.58 0.50 7 61 1.90 0.63 0.45 0.91 8 57 3.14 0.99 0.37 0.29 Note Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Fema le Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Conditio n; Form 8 = Democratic Male Professor in Task Condition

PAGE 97

85 Table 32 Descriptive Statistics for Male Research Participants in the College of Education for Leader Likeability Form N Mean SD Skewness Kurtosis 1 14 2.16 0.60 1.14 1.72 2 20 3.58 0.68 0.31 0.06 3 14 1.73 0.74 0.71 0.20 4 20 3.23 0.70 0.42 0.90 5 20 2.06 0.95 1.03 0.50 6 16 3.66 1.00 0.65 0.27 7 19 2.26 0.88 0.59 0.25 8 17 3.13 0.76 0.99 1.22 Note: Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democr atic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form4=Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Professor in Personal Condition; Form 6 = Democratic Male Professor in Perso nal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition Table 33 Descriptive Statistics for Female Research Participants in the College of Education for Leader Likeability Form N Mean SD S kewness Kurtosis 1 77 2.11 0.77 0.81 0.25 2 75 3.56 0.87 0.52 0.02 3 69 1.83 0.57 0.42 0.20 4 71 30.1 0.95 0.01 0.50 5 72 2.16 0.85 0.21 0.78 6 73 3.55 0.97 0.23 0.74 7 72 1.88 0.66 0.89 1.40 8 73 2.84 0.86 0.06 0.68 Note : Form 1 = Autocratic Female Professor in Personal Condition; Form 2 = Democratic Female Professor in Personal Condition; Form 3 = Autocratic Female Professor in Task Condition; Form 4 = Democratic Female Professor in Task Condition; Form 5 = Autocratic Male Profess or in Personal Condition; Form 6 = Democratic Male Professor in Personal Condition; Form 7 = Autocratic Male Professor in Task Condition; Form 8 = Democratic Male Professor in Task Condition

PAGE 98

86 The third assumption of the analysis of variance proce dure is homogeneity of variance. Violations of this assumption could impact the F test in the positive direction, especially if heterogeneity is associated with one deviant group, or when treatment groups have unequal sample sizes (Keppel, 1991). Heterogen eous variances have a very slight effect on nominal alpha, and no theoretical power value exists with heterogeneous variances (Winer et al., 1991). This study used the modified Levenes test to check the three leadership factors for homogeneity of variance s by booklet form and participant gender, separately for the College of Arts and Science and Education. The homogeneity assumption held for the masculinity factor in both colleges, but was violated for the competence and likeability factors (Table 34). Ta ble 34 Modified Levenes Test Summary Statistics for the College of Arts and Science and Education by Booklet Form and Participant Gender Study Sample DF MS F p College of Arts & Science N=931 Competence Masculinity/Femininity Likeability 15 15 15 1.42 0.40 3.38 2.41 1.21 2.87 <.01 >.05 <.01 College of Education N=722 Competence Masculinity/Femininity Likeability 15 15 15 0.98 0.5 0 1.89 1.89 1.55 2.40 <.05 >.05 <.01

PAGE 99

87 Hypothesis 1 Democratic professors will be rated as significantly more competent than autocratic professors Tables 35 and 36 show the ANOVA source table for both colleges with regard to the competence facto r. The prediction of the hypothesis that democratic professors would be rated more competent than autocratic professors was supported by the results of this study for the College of Arts and Science, F (1, 915)=52.9, p <.01, and for the College of Education F (1, 706)=39.7, p <.01. Research participants from both colleges rated the democratic professor as more competent (Arts and Science: M=3.5, SD=0.74; Education: M=3.4, SD=0.7) than the autocratic professor (Arts and Science: M=3.1, SD=0.7; Education: M=2. 9, SD=0.7). Estimated effect sizes for this competence main effect were medium for both colleges (Arts and Science=0.24, Education=0.23). Tables 35 and 36 list the ANOVA source information for each college. Hypothesis 2 Autocratic professors will be rated as significantly more masculine than democratic professors. This hypothesis was supported in this experiment in both the sample from the College of Arts and Science, F(1, 915)=352.2, p <.01, and the sample from the College of Education, F(1, 706)=241.0, p <.01. The calculated effect sizes for this main effect were large for both colleges ( f =0.60) (see ANOVA source Tables 37 and 38). Research participants for both colleges rated the autocratic professors as more masculine (Arts and Science: M=3.9, SD=0.7; Education: M=4.0, SD=0.7), than democratic professors (Arts and Science: M=3.1, SD=0.7; Education: M=3.1, SD=0.7).

PAGE 100

88 Table 35 ANOVA Source Table for the College of Arts and Science for Competence Source DF MS F f Professor Gender (PG) 1 1.4 2.5 Profe ssor Behavior (PB) 1 29.2 52.9** 0.24 PG x PB 1 0.3 0.7 Situation (S) 1 1.0 1.8 PG x S 1 0.1 0.1 PB x S 1 0.2 0.3 PG x PB x S 1 0.0 0.0 Participant Gender (PaG) 1 0.1 0.2 PG x PaG 1 0.0 0.0 PB x PaG 1 0.5 0.8 PG x PB x PaG 1 0.1 0.2 S x PaG 1 0.1 0.2 PG x S x PaG 1 0.8 1.4 PB x S x PaG 1 3.3 6.0* 0.08 PG x PB x S x PaG 1 0.1 0.2 Error 915 0.5 N=931; Note: p<.05, ** p<.01.

PAGE 101

89 Table 36 ANOVA Source Table for the College of Education for Competence Source DF MS F f Professor Gender (PG) 1 0.0 0.0 Professor Behavior (PB) 1 21.0 39.7** 0.23 PG x PB 1 0.2 0.3 Situation (S) 1 0.0 0.0 PG x S 1 0.9 1.8 PB x S 1 0.1 0.1 PG x PB x S 1 4.0 7.4** 0.10 Participant Gender (PaG) 1 0.1 0.2 PG x PaG 1 0.0 0.0 PB x PaG 1 0.1 0.1 PG x PB x PaG 1 0.0 0.0 S x PaG 1 0.8 1.6 PG x S x PaG 1 1.2 2.3 PB x S x PaG 1 0.0 0.1 PG x PB x S x PaG 1 1.0 1.9 Error 706 0.5 N=721; Note: ** I <.01.

PAGE 102

90 Table 37 ANOVA Source Table for the College of Arts and Science for Masculinity/Femininity Source DF MS F f Professor Gender (PG) 1 2.8 6.9** 0.09 Professor Behavior (PB) 1 144.2 352.2** 0.60 PG x PB 1 0.0 0.1 Situation (S) 1 59.0 144.2** 0.40 PG x S 1 1.3 3.2 PB x S 1 0.2 0.6 PG x PB x S 1 0.8 2.0 Participant Gender (PaG) 1 2.1 5.2* 0.07 PG x PaG 1 0.5 1.2 PB x Pa G 1 3.6 8.8** 0.10 PG x PB x PaG 1 1.1 2.6 S x PaG 1 0.3 0.8 PG x S x PaG 1 0.0 0.0 PB x S x PaG 1 1.0 2.6 PG x PB x S x PaG 1 0.1 0.2 Error 915 0.4 N=931; Note: p<.05, ** p<.01.

PAGE 103

91 Table 38 ANOVA Source Table for the College of Education for Masculinity/Femininity Source DF MS F f Professor Gender (PG) 1 0.2 0.5 Professor Behavior (PB) 1 96.0 241.0** 0.60 PG x PB 1 1.2 3.1 Situation (S) 1 38.4 96.4** 0.40 PG x S 1 0.1 0.3 PB x S 1 0.0 0.0 PG x PB x S 1 1.4 3.6 Participant Gender (PaG) 1 0.3 0.8 PG x PaG 1 0.3 0.8 PB x PaG 1 0.0 0.1 PG x PB x PaG 1 0.4 0.9 S x PaG 1 0.4 1.0 PG x S x PaG 1 0.1 0.1 PB x S x PaG 1 0.7 1.8 PG x PB x S x PaG 1 0.6 1.6 Error 706 0.4 N=722; Note: ** p <.01. Hypothesis 3 Democratic professors will be rated as significantly more likeable than autocratic professors. This hypothesis was supported in both college samples (see ANOVA source Tables 39 and 40). A statistically significant main effect was found for professor leader style for the College of Arts and Science sample, F (1, 915)=333.7, p <.01, and for the College of Education sample, F (1, 706)=274.2, p <.01. The calculated effect sizes for this main effect were equally large in both colleges ( f =0.60). Research participants from both colleges rated the democratic professor as more likeable (College of Arts and Scien ce:

PAGE 104

92 M=3.4, SD=0.9; College of Education: M=3.3, SD=0.9), than the autocratic professor (College of Arts and Science: M=2.3, SD=0.9; College of Education: M=2.0, SD=0.8). Table 39 ANOVA Source Table for the College of Arts and Science for Likeability S ource DF MS F f Professor Gender (PG) 1 6.2 7.6** 0.09 Professor Behavior (PB) 1 273.0 333.7** 0.60 PG x PB 1 1.5 1.4 Situation (S) 1 60.8 74.3** 0.30 PG x S 1 2.2 2.7 PB x S 1 0.3 0.3 PG x PB x S 1 0.1 0.0 Participant Gender (PaG) 1 0.7 0.9 PG x PaG 1 0.1 0.1 PB x PaG 1 4.4 5.3* 0.08 PG x PB x PaG 1 0.0 0.0 S x PaG 1 3.6 4.4* 0.07 PG x S x PaG 1 0.0 0.0 PB x S x PaG 1 4.1 5.0* 0.07 PG x PB x S x PaG 1 0.1 0.1 Error 915 0.7 N=930; Note: p<.05, ** p<.01.

PAGE 105

93 Table 40 ANOVA Source Table for the College of Education for Likeability Source DF MS F f Professor Gender (PG) 1 0.2 0.3 Professor B ehavior (PB) 1 186.0 274.2** 0.60 PG x PB 1 0.9 1.3 Situation (S) 1 15.0 22.0** 0.20 PG x S 1 0.1 0.2 PB x S 1 3.2 4.7* 0.08 PG x PB x S 1 1.7 2.5 Participant Gender (PaG) 1 1.4 2.0 PG x PaG 1 0.4 0.6 PB x PaG 1 0.3 0.4 PG x PB x PaG 1 0.0 0.1 S x PaG 1 0.9 1.3 PG x S x PaG 1 0.6 0.9 PB x S x PaG 1 0.0 0.0 PG x PB x S x PaG 1 0.7 1.0 Error 706 0.7 N=721; Note: p <.05, ** p <.0 1. Hypothesis 4 There will be an interaction effect between participant gender, professor gender, professor leader style, and the type of situation on the ratings of leader competence. Specifically, male research participants were hypothesized to percei ve autocratic female professors to be significantly less competent than autocratic male professors. Further, this difference was hypothesized to be most pronounced in the personal condition such that competence ratings for this condition (autocratic female professor in personal situation) were hypothesized to be the lowest compared to all other conditions. Female research participants were not influenced by professor gender in their perception of leader competence.

PAGE 106

94 This hypothesis was not supported by the results from the College of Arts and Science, F (1, 915)=0.19, p >.05, and the College of Education, F (1, 706)=1.90, p >.05. The effect size estimates for these results were very small ( f =0.01 for the College of Arts and Science and f =0.05 for the College of Education). Hypothesis 5 There will be an interaction effect between participant gender, professor gender, professor leader style, and the type of situation on the ratings of leader likeability. Specifically, male research participants were hypothesize d to perceive autocratic female professors to be significantly less likeable than autocratic male professors. Further, this difference was hypothesized to be most pronounced in the personal condition such that likeability ratings for this condition (autocr atic female professor in personal situation) were hypothesized to be the lowest compared to all other conditions. Female research participants were hypothesized to be influenced by professor gender in their perception of leader likeability. This hypothesi s was not supported by the results from the College of Arts and Science, F(1, 915)=.12, p >.05, and the College of Education, F(1, 706)=.98, p >.05. The effect size estimates for these results were very small ( f =0.01 for the College of Arts and Science and f =0.03 for the College of Education). Additional Findings for the Competence Factor Interaction effects A statistically significant three way interaction effect between professor leader style, type of situation, and participant gender was found for the Co llege of Arts and Science sample, F(1, 915)=5.97, p <.02. Calculated effect size for this three way interaction effect was small (f= 0.08). Figures 6 and 7 depict this interaction effect

PAGE 107

95 separately for male and female research participants. Specifically, mal e and female participants did not differ in their competence ratings for the autocratic professor/ task condition (male participants: M=3.1, SD=0.7; female participants: M=3.1, SD=0.6). However, male participants rated professors in the autocratic personal condition as more competent (M=3.3, SD=0.8) than female participants (M=3.1, SD=0.9). The effect size was small ( d =0.24). This trend was reversed in the democratic condition. That is, male participants rated professors in the democratic/task condition mor e competent (M=3.6, SD=0.6) than female participants (M=3.5, SD=0.7). This effect was small ( d =0.15) However, male participants rated the democratic professor/personal condition less competent (M=3.4, SD=0.8) than female participant ratings for that condit ion (M=3.6, SD=0.7). The effect size estimate was small for this effect ( d =0.27). This particular three way interaction effect was not statistically significant in the College of Education sample. Figure 6 Competence Ratings for theCollege of Arts Science:3Way Interaction Effects for Professor Leader Style, Situation, and Rater Gender (Male Participants) 1 2 3 4 5 Democratic Autocratic Leader Style Competence Ratings Task Personal

PAGE 108

96 Figure 7 Competence Ratings for the College of Arts & Science: 3-way Interaction Between Professor Leader Style, Situation, and Rater Gender (Females) 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leader Style Competence Ratings Task Personal Instead, results from the College of Education sample revealed a statistically significant three way interaction effect between professor gender, professor leader style, and type of situation, F (1, 706)=7.41 p <.01 (Figures 8 and 9). The effect size for t his interaction effect was small ( f =0.10). Research participants in the College of Education rated autocratic male professors more competent than female autocratic professors in the task condition (Male professor: M=3.0, SD=0.7; Female professor: M=2.9, SD =0.6, d =0.15) while the competence ratings were reversed for the autocratic/personal condition (Male professor: M=2.8, SD=0.7; Female professor: M=3.0, SD=0.7, d =0.29). This trend was reversed in the democratic condition. Specifically, research participant s rated democratic male professor in the task condition less competent than the female counterpart (Male

PAGE 109

97 professor: M=3.3, SD=0.7; Female professor: M=3.4, SD=0.6; d =0.15), while the democratic male professor in the personal condition was rated more compet ent than his female counterpart (Male professor: M=3.5, SD=0.9; Female professor: M=3.4, SD=0.7; d =0.13). Figures 8 and 9 illustrate this relationship. Figure 8. Competence Ratings for the College of Education: 3way Interaction between Prof Gender, Prof Leader Style, and Situation (Ratings for Male Professors) 1 2 3 4 5 Democratic Autocratic Leader Style Competence Ratings Task Personal

PAGE 110

98 Figure 9 Competence Ratings for the College of Education: 3-way Interaction between Professor Gender, Professor Leader Style, and Situation (Ratings for Female Professors) 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leader Style Competence Ratings Task Personal Additional Findings for the Masculinity Fac tor A statistically significant main effect was present for the situation condition in both the College of Arts and Science, F (1, 915)=144.2, p <.01, and the College of Education, F (1, 706)=96.4, p <.01. The effect sizes for this main effect were equally l arge for both colleges ( f =0.40). Research participants rated the professor as more masculine in the task condition (Arts and Science: M=3.7, SD=0.8; Education: M=3.8, SD=0.8) than in the personal condition (Arts and Science: M=3.1, SD=0.7; Education: M=3. 3, SD=0.8). Other findings suggested that there were some statistically significant differences between the two samples on the masculinity/femininity ratings. Results of this study found a statistically significant main effect for professor gender only in the College of

PAGE 111

99 Arts and Science sample, F (1, 915)=6.9, p <.01. Calculated effect size estimate for this main effect was small ( f =0.09). Interestingly, research participants from the College of Arts and Science perceived female professors to be more masculi ne (M=3.6, SD=0.8) than their male counterpart (M=3.4, SD=0.8). This main effect was not statistically significant in the College of Education sample. Further, the main effect for participant gender was statistically significant only for the sample from t he College of Arts and Science, F (1, 915)=5.3, p <.03. The calculated effect size for this main effect was small ( f =0.07). Female research participants from the College of Arts and Science perceived professors in the scenarios as slightly more masculine (M= 3.5, SD=0.8) than male research participants (M=3.4, SD=0.8). Finally, a statistically significant interaction effect between professor leader style and participant gender was found in the Arts and Science sample, F (1, 915)=8.8, p <.01 with a small effect size ( f =0.10) (Figure 10). This statistically significant interaction effect was not present in the College of Education sample. Male participants from the College of Arts and Science perceived autocratic professors as less masculine (M=3.7, SD= 0.7) than female participants (M=4.0, SD=0.7). The effect size for this result was small to medium ( d =0.43). Both male and female research participants did not differ in their masculinity ratings for the democratic professor (Males: M=3.1, SD=0.7; Females: M=3.1, SD =0.7). Figure 10 illustrates this interaction effect.

PAGE 112

100 Figure 10 Masculinity Ratings for the College of Arts & Science: 2way Interaction Effect between Professor Leader Style and Participant Gender 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leader Style Masculinity Ratings Male Raters Female Raters Additional Findings for the Likeability Factor A statistically significant main effect for the situation condition was present in both the College of Arts and Science, F (1, 915)=74.3, p <.01), and the College of Education, F (1, 706)=22.0, p <.01. The calculated effect sizes for this main effect were medium in both colleges (Arts and Science=0.30, Education=0.20). However, further results indicated different statistical ly significant trends for the two colleges. A statistically significant main effect for professor gender was obtained for the College of Arts and Science, F (1, 915)=7.6, p <.01. This effect was not present in the

PAGE 113

101 College of Education sample. Male profess ors were perceived more likeable (M=2.9, SD=1.0) than female professors (M=2.7, SD=1.0). Further, a two way statistically significant interaction effect between type of situation and participant gender was present in the College of Arts and Science, F(1, 915)=4.4, p <.04. The effect size calculated for this two way interaction effect was small ( f= 0.07). Male participants perceived professors in the task condition as more likeable (M=2.7, SD=1.0) than female participants (M=2.5, SD=1.0) ( d =0.20), while both male and female participants were similar in their likeability rating for the professor in the personal condition (M=3.1, SD=1.0 for both genders). Figure 11 shows this relationship. Figure 11. Likeability Ratings for the College of Arts & Science: 2way Interaction Effect between Type of Situation and Rater Gender 1 2 3 4 5 Personal Situation Task Situation Likeabiity Ratings Male Raters Female Raters

PAGE 114

102 A two way statistically significant inte raction effect between professor leadership style and participant gender was present in the College of Arts and Science, F(1, 915)=5.3, p <.03. The effect size calculated for this two way interaction effect was small ( f =0.08). Male participants perceived t he autocratic professor as more likeable (M=2.4, SD=1.00) than female participants (M=2.2, SD=0.9). This result had a small effect size ( d =0.21). This trend was reversed in the democratic condition, where male participants perceived the democratic profess or as less likeable (M=3.3, SD=0.9) than female participants (M=3.4, SD=1.0) ( d =0.11). Figure 12 is a visual presentation of this relationship. A three way statistically significant interaction effect between professor leader style, type of situation, and participant gender was found for the College of Arts and Science, F (1, 915)=5.0, p <.03. The calculated effect size for this three way interaction effect was small ( f =0.07; see Figures 13 and 14 ). While both male and female participants perceived the auto cratic professor as the least likeable, female participants were consistently lower in their likeability ratings for the autocratic/task condition (male participants: M=2.1, SD=0.8; female participants: M=1.9, SD=0.7 d =0.27), and autocratic/personal condi tion (male participants: M=2.7, SD=1.0; female participants: M=2.5, SD=1.0, d =0.20) than male participants. This trend was in part reversed in the democratic condition, where male participants perceived the democratic professor in the task situation as mor e likeable (M=3.2, SD=0.9) than female research participants (M=3.0, SD=0.9). This effect was small ( d =0.20). However, female research participants rated democratic professors in the personal condition more likeable (M=3.8, SD=0.9) than male research part icipants (M=3.5, SD=0.9). This effect was small ( d =0.33).

PAGE 115

103 Figure 12 Likeability Ratings for the College of Arts & Science: 2way Interaction between Prof Leader Style and Rater Gender 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leadership Style Likeability Ratings Male Raters Female Raters

PAGE 116

104 Figure 13. Likeability Ratings for the College of Arts & Science: 3way Interaction Effect between Professor Leader Style, Situation, and Rater Gender (Male Raters) 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leader Style Likeability Ratings Task Personal

PAGE 117

105 Figure 14 Likeability Ratings for the College of Arts & Science: 3way Interaction Effect between Prof Leader Style, Situation, and Rater Gender (Female Raters) 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leader Style Likeability Ratings Task Personal Finally, a two way statistically significant interaction effect for likeability between professor leader sty le and type of situation was found for the College of Education, F (1, 706=4.7), p <.04. The calculated effect size for this interaction effect was small (f=0.08). Research participants rated autocratic professors in the task condition less likeable (M=1.9, SD=0.7) than autocratic professors in the personal condition (M=2.1, SD=0.8). This effect was small ( d =0.26). This trend was more pronounced in the democratic condition, that is, democratic professors in the task condition were perceived less likeable (M= 3.0, SD=0.9), than democratic professors in the personal condition

PAGE 118

106 (M=3.6, SD=0.9). The calculated effect size for this result was medium ( d =0.67). Figure 15 illustrates this interaction effect. Figure 15 Likeability Ratings for the College of Education: 2way Interaction Effect between Professor Leader Style and Type of Situation 1 1.5 2 2.5 3 3.5 4 4.5 5 Democratic Autocratic Leaderhip Style Likeability Ratings Task Personal

PAGE 119

107 Chapter V Discussio n This chapter is divided into four parts. The first section summarizes the background and the purpose of the experiment. The second section discusses the psychometric properties of the leadership measures and the implications of these measures for interp reting the results of the hypothesis testing. The third section discusses the results from this experiment and how they relate to previous findings. All findings of this experiment will be discussed in light of several major gender role theories. Section f our lists the limitations of this study, and discusses recommendations for future research. Section I Summary The purpose of this experimental study was to (a) examine the psychometric properties of the factors underlying a set of 24 leadership items and (b) test five hypotheses that were derived from two popular gender role theories as well as from empirical findings of previous studies. This study used an academic context in order to update academic leadership research. An experimental design was creat ed using written vignettes depicting a professor student interaction. Experimental manipulations resulting in eight booklet versions included professor gender, professor leadership style (democratic/autocratic), and the type of situation (task/personal) in which the interaction took place. In addition, the design included participant gender as a variable because female participants were

PAGE 120

108 predicted to be less influenced by leader gender than male participants (as tested in hypotheses 4 and 5). Samples were selected from the College of Arts and Science and from the College of Education in an effort to strengthen population validity. A total of 1654 undergraduate student from the College of Arts and Science (N=932) and the College of Education (N=722) voluntee red for this experiment during the Fall Semester of 2001. Section II Psychometric Properties of the Leadership Construct One factor that has hindered research on gender differences in leadership positions is the lack of psychometrically sound instrume nts that measure the multidimensional nature of leadership. In view of limitations of previous research this study began with the development of a multidimensional measure of leadership. This measure was developed under the assumption that in order to adva nce gender role theory as it relates to leadership, sound measurement of leadership is required. In turn, good theory guides researchers in selecting important variables noted in previous research (Alumbaugh, 1995). As an empirical strategy, leadership outcome measures that were used in previous research were selected and sorted into four leadership outcome dimensions (i.e., competence, masculinity/femininity, likeability, and professional traits). Items reflecting these dimensions were then entered int o a series of exploratory and confirmatory factor analyses in order to evaluate the psychometric properties of these measures. Results of the principal axis factor analysis provided evidence of two strong factors representing competence and masculinity/fe mininity.

PAGE 121

109 Results for the masculinity/femininity dimension revealed that all four originally proposed items retained their internal cohesiveness (moderately high internal consistency), and formed an independent factor. This factor was important because it links gender roles in Western society with general beliefs about leadership qualities (i.e., masculine attributes are associated with leadership roles, and feminine attributes are associated with subordinate roles). Leader competence is an important lead er outcome measure, and has been defined by organizational psychologists (e.g., Hunt, 1991) to be the ability of a leader to facilitate organizational goals. Eagly et al. (1995) in their meta analytic review found that even though many of the studies eval uated leader competence, their methods of assessing this outcome varied. The current study attempted to create a leader competence factor that incorporated five competence items used by previous studies (competent/incompetent, effective/ineffective, qualif ied/not qualified, capable/not capable, influential/not influential). Results showed that all five items clustered with the competence factor. In addition though, three items from the professional traits (hardworking/lazy, responsible/irresponsible, trustw orthy/untrustworthy) loaded on the competence factor. When pondering the qualities of what makes a leader competent, it seemed conceptually reasonable that the qualities of hard working, sense of responsibility, and level of trustworthiness were important leadership behaviors. Although professional traits were originally proposed as a separate dimension of leadership, results from this study showed that this dimension did not hold up as an internally consistent and independent factor. This was not unexpect ed because the originally proposed professional traits consisted of items that had been used in previous

PAGE 122

110 studies and were considered to be miscellaneous. Even though this study dropped the professional traits dimension, additional research is needed to eva luate the generalizability of this finding. For example, this study encountered difficulties with items that had a tendency to crossload on other factors. Generating a set of new items that cross load minimally, may yield improved results. Finally, the li keability factor posed a challenge not uncommon when using analytic procedures with verbally similar yet theoretically different themes (e.g., Tracy, 1995). Even though the likeability items had high internal consistency, the likeability construct did not cluster into an independent factor. Instead, selected items tended to merge with the competence factor (1 item), and the masculinity/femininity factor (2 items). Even though the likeability items (critical/tolerant, popular/unpopular and likeable/not likea ble) were theoretically defined to be a distinct dimension from the masculinity/femininity factor (timid/forceful, soft/tough, aggressive/not aggressive, dominant/submissive), results of the factor analysis provided evidence that research participants did not make that distinction. Instead, two of the likeability items loaded highly onto the masculinity/femininity factor, and one item loaded on the competence factor. It is possible that in spite of pre testing for item quality, the descriptors for the likea bility construct may have been too ambiguous to distinguish them from the other two dimensions. Even though results from this study could not clearly establish likeability as a separate factor, items from the likeability cluster were viewed as potentially adding important information not provided by the competence or masculinity/femininity factors.

PAGE 123

111 To summarize, empirical evidence provided partial support for the multidimensional leadership construct. Results of this study provided evidence for at least a two factor structure (competence and masculinity/femininity), and reasonable evidence to support the likeability dimension. As a result these three factors were used in the testing of the hypotheses. Section III Hypotheses Testing This study tested a set of five hypotheses separately for a sample from the College of Arts and Science (N=931) and for a sample from the College of Education (N=722). Hypothesis 1: Democratic professors will be rated as significantly more competent than autocratic professors. Results of this study provided strong support for the notion that leadership style has an impact on the perceived competence of the leader. Research participants from both colleges viewed the democratic professor to be more competent than the autocratic pr ofessor. Medium effect sizes found in both colleges support this difference in the perception of competence between autocratic and democratic professors (Stevens, 1990). This result echoes the results of previous studies that found that in an organization al setting, democratic managers were evaluated much higher on performance and leadership abilities than autocratic managers (e.g., Luthar, 1996). The higher rating tendency of the democratic professor over the autocratic professor suggests that students wi thin an academic setting show similar rating tendencies for a democratic leader as employees who work in business organizations. Interestingly, Luthar (1996) pointed out that this preference should be expected since North American social values have a demo cratic basis.

PAGE 124

112 Hypothesis 2: Autocratic professors will be rated as significantly more masculine than democratic professors. The results of this study found that research participants from both colleges perceived autocratic professor s to be significantly more masculine than democratic professors. The large effect sizes indicate that research participants very clearly perceived the democratic professor to be more feminine, and the autocratic professor to be more masculine. This result seems to contradict culturally held beliefs that competent leaders display a high degree of masculine traits. The obtained effect size for this finding was large, which is rarely observed in social sciences research (Stevens, 1990), and indicates that part icipants clearly perceived democratic professors to be much less masculine than autocratic professors regardless of professor gender. How can this result be interpreted? First, most of the gender role theories maintain that culturally defined gender roles influence peoples expectation about what makes a leader competent. The concern this study raises is whether gender roles as defined by those theories actually are still an accurate reflection of currently held gender beliefs, especially since most of the gender role theories were proposed during the 70s and 80s. On the other hand, results may reflect Kanters (1977) notion that negative evaluation of women in leadership positions occur because of the differential distribution of men and women in manageri al positions, that is, women are more visible because of their minority group membership. This study was conducted within an academic environment believed to have an approximate equal distribution of male and female professors, which could explain why the results did not show a gender difference that was hypothesized using gender role spill over theory. The results further reflect Phillip et al.s theory (1982) that leaders are evaluated by specific

PAGE 125

113 beliefs and expectations about leader behaviors that overr ide expectations about gender role behaviors. As already mentioned, it would be interesting to examine if cultural gender role beliefs have changed compared to the beliefs generally held about 30 years ago, and how these changes may impact gender role theo ries and their implications. Finally, replicating this study within an organizational environment, may help develop a more precise understanding of how and when gender matters in leader evaluation. Hypothesis 3: Democratic professors will be rated as s ignificantly more likeable than autocratic professors. As expected, research participants from both colleges rated the democratic professor to be much more likeable than the autocratic professor irrespective of gender. The magnitude of this effect size was large for both samples. Past research (e.g., Bartol & Butterfield, 1976) found that female managers were evaluated to be less effective, as well as less likeable if they displayed an autocratic leadership style compared to democratic female managers. Bart ol et al.s findings support gender role congruency theory that punishes gender role incongruent behaviors of leaders. However, results of this study did not find evidence in support of the gender congruence theory. Instead, research participants rated bot h male and female democratic professors as more likeable than their autocratic counterparts. This study showed that research participants clearly perceived the democratic professor as much more likeable (large effect size), more competent (medium effect si ze), and as more feminine than their autocratic counterparts. The perception of a competent leader in this study was associated with possessing a higher degree of femininity, regardless of leader gender. Results indicated that leader style rather than lead er gender was crucial in the evaluation of the leadership qualities. This finding is

PAGE 126

114 consistent again with Phillips et al. (1982) who maintain that leader behavior of managers becomes more important than leader gender in the evaluation of leaders. Hypothe sis 4: There will be an interaction effect between participant gender, professor gender, professor leadership style and type of situation on the ratings of leader competence This hypothesized four way interaction effect was complex in that it combined pred ictions from gender role spill over and gender congruency theories, while incorporating previous empirical evidence that men were more negative than women when evaluating a female leader. Specifically, it was predicted that male professors would not be sub jected to the level of scrutiny when described as behaving inconsistently with their gender role expectation due to the societal association between male gender roles and leadership (gender spill over theory). Predictions were very different for women (gen der congruency theory), in that female professors who behave in a manner that is perceived to be inconsistent with their gender role were predicted to be rated less competent than gender congruent female professors. This devaluation was expected to happen with male raters (past research). Results of this study did not support this four way interaction effect. First, male research participants did not perceive female professors who behaved in a gender incongruent manner (autocratic) while dealing with eithe r task oriented or interpersonally oriented situations more negatively than female research participants. Further, research participants did not favor male professors over female professors when female professors behaved in a gender incongruent manner. How do the results of this study relate to the results from meta analytic procedures? First, even though this study manipulated a condition which portrayed a female professor

PAGE 127

115 behaving in a manner that is considered to be gender incongruent (autocratic) under a condition that calls for feminine oriented skills (interpersonal situation), the manipulation may not have been strong enough to elicit the perception of gender incongruent behaviors especially in male participants since the scenarios were designed with in the academic environment, which is generally believed to be more gender neutral. This is important because the magnitude of this interaction as calculated by meta analytic research has been consistently very weak (i.e., Eagly et al., 1990; Swim et al., 1989). Eagly et al. further stressed that this differential evaluation of male and female leaders most consistently occurred within the following contexts: (a) when overall leadership measures were used (i.e., leader competence, satisfaction with leader), (b) when few independent variables were examined, and (c) when female leadership was evaluated within a male dominated field (i.e., athletic coaches), especially by male raters. The fact that the results of this study did not support a similar trend may b e due to several reasons. Again, as already mentioned, the study examined leadership behaviors in an academic context that is believed to have a fairly equal gender distribution (as described by the general employment statistics of men and women within the College of Arts and Science and the College of Education as opposed to the more traditionally male oriented College of Engineering). Further, this experiment manipulated a complex set of variables simultaneously within the context of a scenario in order t o allow the interplay among conditions (as opposed to manipulating only one variable). Nevertheless, the lack of an interaction effect was consistent across both samples of this study, and should not be dismissed, especially since effect sizes have consist ently been very small in meta analytic research. It is crucial to also consider the publication date of studies investigating

PAGE 128

116 perception of gender differences in leader positions because societal changes or shifts may occur very rapidly over time. This bel ief follows Gergens (1986) notion that gender roles reflect values of society at a given point in time and must be evaluated within the context of time. For example, Rohjahn et al. (1994) quoted the 1975 research of Costrich et al. that found that men s howed a tendency to evaluate female leaders who behaved in a masculine manner much less favorably than female leaders who displayed feminine behaviors. Perhaps results from this study as well as other recently published research underscore the importance o f using updated research when examining the impact of societal gender roles on the evaluation of male and female leaders. Gender roles within the Western society may be changing more quickly, thus contributing to the inconsistencies in research findings ( especially when current results are compared to results that are more than ten years old). Hypothesis 5: There will be an interaction effect between participant gender, professor gender, professor leader style and type of situation on the ratings of lead er likeability. This hypothesized four way interaction effect for likeability was based on the same premise as hypothesis 4. Again, results failed to show support for the gender congruency theory. The results of this study do show that male participants in both colleges failed to be influenced by the gender incongruent behaviors of the female professor. Instead, both male and female participants were influenced by professor leadership style when evaluating leader likeability. How does this finding compare t o the findings of previous research? Bartol et al. (1976) found that male and female leaders were less liked and perceived to be less effective when they behaved in a gender

PAGE 129

117 incongruent manner. More recent research (Rojahn et al., 1995) found support for t he gender congruency theory only from male research participants. It appears that results of this study may indicate a weakening trend of gender role influence on the evaluation of leadership measures. Results of this study certainly seem to support the no tion that it is the leadership style a leader (male or female) displays that influences how he or she is perceived in terms of competence and likeability (as well as level of masculinity). Gender of the professor clearly had a negligible influence in the p erception of leader competence and likeability for male and female participants alike. This finding is encouraging because it supports a trend found in most recent research that indicates a lessening impact of social gender roles in the evaluation of male and female leaders. Additional Findings The results reported in this section are exploratory in nature because they were not predicted beforehand. Although these findings were interesting, it is important to point out that (a) the effects were not cons istent across the two samples, and (b) the calculated effect sizes for those effects were very small. Still, reporting these results underscore the difficulties in grasping underlying mechanisms of gender differences in leadership research. College of Arts and Science A three way interaction effect with regard to competence revealed that male participants rated the democratic professor in the task condition as most competent, followed by the democratic professor in the personal condition and the autocrati c professor in the personal condition. The autocratic professor in the task conditions was rated the least competent. Female participants on the other hand rated the democratic professor in the personal condition as most competent,

PAGE 130

118 followed by the democrat ic professor in the task condition. Female participants rated autocratic professors in the task and personal conditions equally low. This result is difficult to explain because it was not repeated in the College of Education sample. Nevertheless it appea red that male participants perceived democratic professors in the task condition to be most competent, while female participants rated democratic professors in the personal conditions as most competent. This could be a product of the vignette manipulation, or, perhaps that male participants felt most comfortable with the democratic task vignette, because the task situation reflected masculine qualities, while female participant felt most comfortable with the feminine overtones portrayed by the democratic pe rsonal situation An interesting main effect with regard to the masculinity factor was found in that research participants perceived female professors to be more masculine than male professors. In addition, a two way interaction effect revealed that female participants perceived autocratic professors to be more masculine than male participants, while all participants rated democratic professors similarly in terms of masculinity. What does this result mean? First, it needs to be pointed out that this finding was not replicated in the College of Education, and the calculated effect size was very small. Yet, the fact that female professors were perceived to be more masculine than their male counterparts may indicate that participants may have perceived both male and female professors as behaving in a somewhat gender incongruent manner. Yet, this result was tempered by the two way interaction effect that indicated that female participants identified autocratic behavior as acutely masculine. Even though these findi ngs were not consistent, they are intriguing because they hint at undercurrents of gender differences concerning the

PAGE 131

119 inclusion and importance of external cues in the evaluation of a leader. Future research is needed to explore these issues in more depth. With regard to the likeability factor, participants slightly preferred the male professor over the female professor. In addition, a three way interaction effect between professor behavior, situation and participant gender indicated that female participant s were more extreme in their ratings, that is, while all participants liked the autocratic professor in the task situation the least, female participants were more extreme in their low rating. Similarly, while all participants liked the democratic professo r in the person oriented situation the best, female participants also rated the democratic professor much higher than male participants on the likeability score. Again, these results were not replicated in the education sample. Still, a pattern emerges in that female participants may weigh external cues differently in the level of importance when considering leader likeability. It is recommended that future research should explore this avenue further before coming to any conclusions. College of Education A three way interaction with regard to competence ratings was found between professor gender, professor behavior, and situation. The democratic male professor was rated the most competent, followed by the democratic female professor in both personal and t ask conditions. The lowest competence rating was given to the autocratic male professor in the personal condition, followed by the autocratic female professor in the task condition. While there was a difference in competence rating between male and female research participants in the Arts and Science sample, the results from the education sample produced a gender difference only in terms of the professor gender. This difference might have been produced due to the low numbers of male

PAGE 132

120 participants available i n Education (80% female participants). Yet, it is intriguing, that the democratic male professor in a personal situation received slightly higher competence ratings than his female counterpart. This result counters gender role congruency theory that punish es gender role incongruent behavior by both genders. Luthar (1996) has noted that if the behavior of a person is perceived to be in extreme contrast to stereotypical expectation, the reaction to it could become more unpredictable. This result was not repli cated in the Arts and Science sample. Nevertheless, future research should explore possible effects that extreme violations of stereotypical expectation can have on the evaluation of a leader. Section IV Future Research Design In their meta analytic review Eagly et al. (1992) found two major research paradigms researchers used to investigate gender differences in leadership evaluation: laboratory experiments usually conducted with college students as research participants and organizational studies using e mployees and management to fill out rating scales. Organizational studies have the advantage of examining gender differences in a natural setting, whereas experimental studies have greater control of variable manipulation by the use of written vignettes, o r by playing out a scenario with trained confederates. The use of written vignettes in experiments (a technique introduced by Rosen & Jerdee in 1973) is most common because of the relative ease of multiple variable manipulations within a scenario. The stre ngth of an experimental design usually lies within the high degree of internal validity, that is, it is reasonable to assume that the potentially confounding effects of extraneous variables are controlled (e.g., Kennedy et al., 1985). The internal validit y

PAGE 133

121 of this study can be assumed to be satisfactory because of (a) the systematic manipulation of the independent variables embedded in a written scenario which remained constant and (b) the random assignment of the eight booklet version to research particip ants. External validity, the process of ensuring that all members of a population have the same chance of being selected (i.e., random selection) enhances the generalizability of the results to that larger population (Kennedy et al., 1985). It was not poss ible to employ random selection for this study. In order to maximize external validity, two independent samples were used, one from the College of Arts and Science (N=932) and one from the College of Education (N=722). Most Introductory classes from these two colleges are attended by students with varied majors because these classes are part of the college requirement. The results of this study were replicated using two independent samples of undergraduate students who varied in terms of age, racial makeup and school standing, suggesting a moderately confident level of generalizability to the undergraduae student population. An increasingly important issue of validity is the question of ecological validity (i.e., how appropriate is it to generalize the res ults of a study from one context to another?). Kerlinger (1973) described ecological validity as part of external validity, in that it concerns whether the findings derived from an experimental study conducted with a sample drawn from a particular populati on can be generalized to other settings or conditions. One major disadvantage of experimental designs is the byproduct of artificiality created by the experimental control. For example, Eagly et al. (1990) found that when participants in experiments were g iven the role of a leader, they tended to behave more stereotypically than managers observed within organizational settings.

PAGE 134

122 Many researchers support the relevance or ecological validity contribution of experimental studies in the field of leadership studi es. For example, Locke (1986) found that results of laboratory studies were equivalent to results found in field studies. Similar findings derived from different methodologies indeed present a cogent argument to support the notion that experimental designs have a high degree of internal validity as well as ecological validity. It is difficult to evaluate the level of ecological validity of the present study. Even though the researcher is confident that this study achieved a reasonable level of internal val idity, the issues of ecological validity are more complex. Most importantly, ecological validity focuses on the realism of the experiment. This study had several strengths as well as weaknesses regarding the issue of realism. First, this study conducted p ilot studies that investigated the plausibility of the content of the scenarios. Results indicated that both male and female participants rated both task and personal scenarios as plausible regardless of professor gender or professor leadership. Yet, the r elationship between an employee and manager, or between a professor and student as it occurs in a realistic setting could not be simulated by one written paragraph. Thus, one scenario can only portray one incident. The realism of a relationship between lea der and subordinate (professor and student) is impossible to simulate. More research is needed that focuses on how to make experimental studies more realistic without sacrificing the level of internal control. Pilot results of this study, for example, pr ovided evidence that the scenarios were rated to be realistic, providing evidence for scenario realism. However, different methodologies may yield different results. For example, this study only used verbal descriptors of the professor. A degree of realism could be enhanced if visual images (photographs) of the professor were added to the verbal

PAGE 135

123 descriptors. The impact of visual depiction may lead to different results. Extensive research, for example, has provided evidence that physically attractive people are generally evaluated more positively than their less attractive counterparts (e.g., Efran, 1974). Finally, one of the most common sources of bias in attitudinal studies is the halo effect (Pedhazur & Schmelkin, 1991). This type of bias occurs when a gen eral impression of the person to be rated is formed, which then systematically influences the ratings for that person on unrelated dimensions. For example, Nisbett and Wilson (1977) found that students who watched a videotape of a friendly professor rated that professor more favorably on other traits than students who watched the same videotape but with an arrogant professor. The results of this study found that democratic professors were rated to be more competent and more likeable than their autocratic co unterparts. Could a systematic bias have contributed to this result? This is a difficult question to answer in hindsight. It is possible, that democratic professors may have received inflated likeability ratings due to the halo effect. However, results of this study found interaction effects with regard to competence and likeability ratings (see additional findings) that suggest a minimal halo effect on the outcome measures. Nevertheless, future research should investigate the impact of halo effects on the ratings of leadership qualities. Implications This study generated a list of male and female traits based on currently widely accepted theoretical frameworks of gender stereotypes (e.g., Eagly et al., 1984). These gender stereotypes (e.g., dominant and aggressive for men, submissive and friendly for women) have changed very little since Halpins (1957) research on gender differences in

PAGE 136

124 leadership behavior. It may be appropriate to question whether, for example, men are still believed to be more self ass ertive and master their environment more so than women, while women on the other hand are still believed to be more selfless and concerned with others. Stereotypes people have about gender and leadership roles may have significantly changed since the 70s and 80s to the point of re evaluating those beliefs. Similarly, research should update theoretical frameworks of gender roles (e.g., gender role spillover, gender congruency theory) in order to enhance the predictive power of the model. Gender role theori es are very popular, yet, meta analytic results revealed weak evidence at best in their support. Results of this experiment showed that male and female participants rated democratic professors more competent than autocratic professors regardless of gende r and situation. This information could be helpful for tenure track seeking instructors who depend in part on student ratings for their job promotions. Simply being aware that a democratic teaching style can positively impact teacher evaluations could help improve teaching skills as well as increase their changes for promotions. Further, teaching democratic leadership behaviors seem crucial in developing future leaders within academic and organizational environments. This study provided evidence that democr atic behaviors are conveyed verbally as well as nonverbally. Democratic nonverbal behaviors (e.g., leaning forward, smiling, nodding, moderately high eye contact) should be included when teaching future leaders to achieve democratic leadership style. In ad dition, this study provided evidence that democratic leadership styles are equally effective in task as well as personal situations. This information is important especially for female leaders who may feel pressured to adopt an autocratic style in personal situations so as not to be

PAGE 137

125 perceived as weak. Current and future leaders within academic or organizational arenas should consider de emphasizing an autocratic style and instead emphasize leadership behaviors that reflect a democratic style.

PAGE 138

126 References Andersen, P. A., & Coussoule, A. R. (1980). The perceptual world of the communication apprehensive: The effect of communication apprehension and interpersonal gaze on interpersonal perception. Communication Quarterly, 28 44 54. Alimo Metcalfe, B. (1995). Leadership and Assessment. In A. Buckley, The essence of women in management (pp. 92 109). Englewood Cliffs, NJ: Prentice Hall. Ashmore, R. D., & Del Boca, F. K. (1979). Sex stereotypes and implicit personalit y theory: Toward a cognitive social psychological conceptualization. Sex Roles, 5 219 248. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups Reading, MA: Addison Wesley. Baron, R. M., & Kenny. (1986).The mode rator mediator variable distinctionin social psychological research: Conceptual, strategic, and statistical considerations. Jouranl of Personality and Social Psycholgy, 51, 1173 1182. Bartol, K. M. & Martin, D. C. (1986). Women and men in task groups In R. D. Ashmore & R. K. Del Boca (eds). The social psychology of female male relations: A critical analyses of central concepts (pp. 259 310) Bass, B. M. (1990). Bass & Stogdill's handbook of leadership: Theory, research, and managerial applications (3 rd ed.). New York: Free Press. Bass, B. M. (1981). Stogdill's handbook of leadership: A survey of theory and research (rev. ed.). New York: Free Press. Berger, J., Webster, M., Jr., Ridgeway, C. L., & Rosenholtz, S. J. (1986). Status cues, expec tations and behaviors. In E. Lawler (Ed.), Advances in group processes, (p 3, 1 22). Greenwich, CT: JAI Press. Berger. J., Wagner, D. G., & Zelditch, M., Jr. (1985). Introduction: Expectation states theory. In J. Berger & M. Zelditch, Jr. (Eds.), Statu s, rewards, and influence (pp. 1 72). San Francisco: Jossey Bass. Berger, J., Rosenholtz, S. J., & Zelditch, M., Jr., (1980). Status organizing processes. Annual Review of Sociology, 6 479 508.

PAGE 139

127 Bergmann, W. (ed.) (1988). Error without trial. In Ps ychological Research on Antisemitism New York, NY: Walter de Gruyter. Briton, N. J., & Hall, J. A. (Spring, 1995). Gender based expectancies and observer judgements of smiling. Journal of Nonverbal Behavior, 19 (1), 49 65. Broverman, I. K., Vogel, S. R., Broverman, D. M., Clarkson, F. E., & Rosenkrantz, P. S. (1972). Sex role stereotypes: A current appraisal. Journal of Social Issues, 28 (2), 59 78. Burgoon, J. K., Coker, D. A., & Coker, R. A. (1986). Communicative effects of gaze behavior: A test of two contrasting explanations. Human Communication Research, 12 (4), 495 524. Burgoon, J. K. (1985). Nonverbal signals. In M. L. Knapp & G. R. Miller (Eds.) Handbook of interpersonal communication Beverly Hills, CA: Sage. Burgoon, J. K., Buller, D B., Hale, J. L, & deTurck, M. A. (1984). Relational messages associated with nonverbal behaviors. Human Communication Research, 10 351 378. Burgoon, J. K., & Jones, S. B. (1976). Toward a theory of personal space expectations and their violations. Human Communication Research, 2 131 146. Carli, L. L., LaFleur, S. J., & Loeber, C. C. (1995). Nonverbal behavior, gender, and influence. Journal of Personality and Social Psychology, 68 (6), 1030 1041. Carli, L. L., Martin, C. Leatham, G., Lyons, K. & Tse, I. (1993). Perceptions of nonverbal behavior Paper presented at the annual meeting of the American Psychological Association, Toronto, Ontario, Canada. Colwill, N. L. (1995). Sex differences. In Buckley, A. (1995). The essence of women in ma nagement (pp. 20 34). Englewood Cliffs, NJ: Prentice Hall. Denmark, F. L. (1993). Women, leadership, and empowerment. Psychology of Women Quarterly, 17 343 356. Deaux, & Major. (1987). Deutsch, F. M. (September, 1990). Status, sex and smiling: The effect of role on smiling in men and women. Personality and Social Psychology Bulletin, 16 (3), 521 540. Dorenkamp, A. G., McClymer, J. F., Moynihan, M. M., & Vadum, A.C. (1995). Images of women in American popular culture (2 nd ed). Fort Worth, TX: H arcourt Brace & Co.

PAGE 140

128 Dovidio, J. F., Keating, C. F., Heltman, K., Ellyson, S. L., & Brown, C. E. (1988). The relationship of social power to visual displays of dominance between men and women. Journal of Personality and Social Psychology, 54 (2), 233 242. Dovidio, J. F., Heltman, K., Brown, C. E., Ellyson, S. L., & Keating, C. F. (1988). Journal of Personality and Social Psychology, 55 (4), 580 587. Eagly, A. H., Karau, S. J., Makhijani, M. G. (1995). Gender and the effectiveness of leaders: A meta an alysis. Psychological Bulletin, 117 (1), 125 145. Eagly, A. H., & Johnson, B. T. (1990). Gender and leadership style: A meta analysis. Psychological Bulletin, 108 (2), 233 256. Eagly, A. H., Makhijani, M. G.; & Klonsky, B. G. (1992). Gender and the evaluation of leaders: A meta analysis. Psychological Bulletin, 111 (1), 3 22. Eagly, A. H., & Wood, W. (1991). Explaining sex differences in social behavior: A meta analytic perspective. Journal of Personality and Social Psychology, 17 (3), 306 315. Eagly, A. W., & Mladinic, A. (1989). Gender stereotypes and attitudes toward women and men. Personality and Social Psychology Bulletin, 15 (4), 543 558. Eagly, A. H. (1987). Sex differences in social behavior: A social role interpretation Hillsdale, NJ: Erlbaum. Eagly, A. H., & Steffen, V. J. (1986). Eagly, A. H., & Steffen, V. J. (1984). Gender stereotypes stem from the distribution of women and men into social roles. Journal of Personality and Social Psychology, 46 735 754. Ellsworth, P. C., & Langer, E. J. (1976). Staring and approach: An interpretation of the stare as a nonspecific activator. Journal of Personality and Social Psychology, 33 117 122. Ellyson, S. L., & Dovidio, J. F. (Eds.) (1985). Power, dominance, and nonverbal behavio r New York, NY: Springer Verlang. Exline, R. V., Ellyson, S. L., & Long, B. (1975). Visual behavior as an aspect of power role relationship. In P. Pliner, L. Krames, & T. Alloway (Eds.), Nonverbal communication of aggression (pp. 21 52). NY: Plenum P ress.

PAGE 141

129 Exline, R. V., & Winters, L. C. (1965). Affective relations and mutual glances in dyads. In S. S. Tomkins, & C. W. Izard (Eds.) Affect, cognition and personality NY: Springer. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research Reading, MA: Addison Wesley. Fiske, S. T., Bersoff, D. N., Borgida, E., Deaux, K., & Heilman, M. E. (1991) Social science research on trial: The use of sex stereotyping research in Price Waterhouse vs. Hopkins. American Psychologist, 46 1049 1060. Friesen, E. P., & Ellsworth, P. (1972). Emotion in the human face NY: Pergamon Press Gergen, K. J. (1986). Correspondence versus autonomy in the language of understanding human action. In D. W. Fiske & R. A. Shewder (Eds. ) Metatheory in Social sciences Chicago: University of Chicago Press. Goffman, E. (1976). Gender advertisements. Studies in the Anthropology of Visual Communication, 3 69 154. Gutek, B. A. (1993). Changing the status of women in management. Applied Psychology: An International Review, 42 301 311. Gutek, B. A., & Morasch, B. (1982). Sex ratios, sex role spillover, and sexual harassment of women at work. Journal of Social Issues, 38 (4), 55 74. Halberstadt, A. G., Hayes, C W., & Pike, K. M. (1988). Gender and gender role differences in smiling and communication consistency. Sex Roles, 19 (9 10), 589 605. Halberstadt, A. G., & Saitta, M. B. (1987). Gender, Nonverbal behavior, and perceived dominance: A test of the theory Journal of Personality and Social Psychology, 53 (2), 257 272. Hall, J. A., & Halberstadt, A. G. (1986). Smiling and gazing. In J. S. Hyde & M. Linn (Eds.), The psychology of gender: Advances through meta analysis Baltimore, MD: The John Hopkins U niversity Press. Hall, J. A. (1984). Nonverbal sex differences: Communication accuracy and expressive style Baltimore, MD: John Hopkins University Press. Harris, M. B. (1976). The effects of sex, sex stereotyped descriptions, and institution on evalu ation of teachers. Sex Roles, 2 15 21.

PAGE 142

130 Hedges, L. V., & Olkin, I. (1985). Statistical methods for meta analysis San Diego, CA: Academic Press Heilman, M. E., Block, C. J., Martell, R. F., & Simon, M. C. (1989). Has anything changed? Current chara cterizations of men, women, and mangers. Journal of Applied Psychology, 74 935 942. Henley, N. M. (1997). Body politics: Power, sex, and nonverbal communication Englewood Cliffs, NJ: Prentice Hall. Henley, N. M. (1973). Status and sex: Some touchin g observations. Bulletin of the Psychonomic Society, 2 91 93. Hollander, E. P. (1992). The essential interdependence of leadership and followership. Current Directions in Psychological Science, 1 71 75. Hooff, J. A. R. A. M. van (1962). Facial exp ressions in higher primates. Symposia of the Zoological Society of London, 8 97 125. House, R. J. (1971). A path goal theory of leader effectiveness. Administrative Science Quarterly, 16 321 338. Hunt, J. G. (1991). Leadership: A new synthesis N ewbury Park, CA: Sage. Johnson, C. (1994). Gender, legitimate authority, and leader subordinate conversations. American Sociological Review, 59 122 135. Kanter, R. M. (1977). Men and women of the corporation New York: Basic Books. Kaschak, E. (1 978). Sex bias in student evaluations of college professors. Psychology of Women Quarterly, 2 (3), 235 43. Kierstead, D., D'Agostino, P., & Dill, H. (1988). Sex role stereotyping of college professors: Bias in students' ratings of instructors. Journal of Educational Psychology, 80 (3), 342 344. Kleinke, C. L. (1980). Interaction between gaze and legitimacy of request on compliance in a field setting. Journal of Nonverbal Behavior, 5 3 10. Klenke, K. (1996). Women and leadership: A contextual pe rspective NY, NY: Springer. Kollock, P., Blumsteing, P. & Schwartz, P. (1985). Sex and power in interaction: Conversational privileges and duties. American Sociological Review, 52 34 46.

PAGE 143

131 Kraut, R. E., & Johnston, R. E. (1979). Social and emotion al message of smiling: An ethological approach. Journal of Personality and Social Psychology, 37 9, 1539 1553. Kruse, L., & Wintermantel, M. (1986). Leadership Ms. qualified: I. The gender bias in everyday and scientific thinking. In C. F. Graumann & S. Moscovici (Eds.) Changing conceptions of leadership, pp. 171 197. New York: Springer Verlag. Levin, K., & Lippitt, R. (1938). An experimental approach to the study of autocracy and democracy: A preliminary note. Sociometry, 1 292 300. Lindsey, L. (1990). Gender roles: A sociological perspective Englewood Cliffs, NJ: Prentice Hall. Loden, M. (1985). Feminine leadership or how to succeed in busines without being one of the boys New York: Times Books. MacCoby, E. E., & Jacklin, C. N. (19 74). The psychology of sex differences Stanford, CA: Stanford University Press. Martin, E. (1984). Power and authority in the classroom: Sexist stereotypes in teaching evaluations. Journal of Women in Culture and Society, 9 482 492. MacCallum, R. ( 1986). Specification searches in covariance structure modeling. Psychological Bulletin, 100 107 120. Mehrabian, A. (1967). Orientation behaviors and nonverbal attitude communication. Journal of Communication, 16 324 332. Nieva, V. E., & Gutek, B. A. (1980). Sex effects on evaluation. Academy of Management Review, 5 267 276. Payne, K. E., Fuqua, H. E., & Canegami, J. P. (1997). Women as Leaders. In Rosenbah, W. E., & Taylor, R. L. (1998). Contemporary issues in leadership p. 145 167. Boulde r, CO: Westview Press. Petty, M. M., & Lee, G. K. (1975). Moderating effects of sex of supervisor and subordinate on relationships between supervisory behavior and subordinate satisfaction. Journal of Applied Psychology, 60 624 628. Pedhazur, E. J., & Schmelkin, L. P. (1991). Measurement, Design, and Analysis: An Integrated Approach Hillsdale, NJ: Lawrence Erlbaum.

PAGE 144

132 Phillips, J. S., & Lord, R. G. (1982). Schematic information proccessing andperceptions of leadership in problem solving groups. J ournal of Applied Psychology, 67 486 492. Powell, G. N. (1988). Women and men in management Newbury Park, CA: Sage. Ragins, B. R., & Sundstrom, E. (1989). Gender and power in organizations: A longitudinal perspective. Psychological Bulletin, 105 51 88. Remland, M., Jacobson, C., & Jones, T. (1983). Effects of psychological gender and sex incongruent behavior on evaluations of leadership. Perceptual and Motor Skills, 57 783 789. Rice, R. W., Instone, D., & Adams, J. (1984). Leader sex, le aders success, and leadership process: Two field studies. Journal of Applied Psychology, 69 12 31. Ridgeway, C. (1987). Nonverbal behavior, dominance, and the basis of status in task groups. American Sociological Review, 52 683 694. Ridgeway, C. L., Berger, J., & Smith, L. (1985). Nonverbal cues and status: An expectation states approach. American Journal of Sociology, 90 955 978. Ridgeway, C. L., (1982). Status in groups: The importance of motivation. American Sociological Review, 47 76 88. Rojahn, K., & Willemsen, T. M. (1994). The evaluation of effectiveness and likability of gender role congruent and gender role incongruent leaders. Sex Roles, 30 (1 2), 109 119. Sargent, A. G. (1981). The androgynous manager New York: Amacom. Straus s, H. A., & Bergman, W. (1988). Current research on antisemitism New York, NY: Walter de Gruyter. Swim, J., Borgida, E., Maruyama, G., & Myers, D. G. (1989). Joan McKay versus John McKay: Do gender stereotypes bias evaluations? Psychological Bulleti n, 105 409 429. Tracy, L. (1987). Consideration and initiating structure: Are they basic dimensions of leader behavior. Social Behavior and Personality, 15 (1), 21 33. Tocqueville, A. de. (1840). Democracy in America: Part the Second. The Social I nfluence of Democracy (trans. Henry Reeve). NY: J & H. G. Langley.

PAGE 145

133 Unger, R. (1979). Sexism in teacher evaluation: The comparability of real life to laboratory analogs. Academic Psychology Bulletin, I 163 171. Vinicombe, S. & Colwill, N. L. (1995). Managers and Secretaries. In Buckley, A. (1995). The essence of women in management (pp. 59 73). Englewood Cliffs, NJ: Prentice Hall. Welsh, M. C. (1979). Attitudinal measures and evaluation of males and females in leadership roles. Psychological Rep orts, 45 19 22. Willson, A., & Lloyd, B. (1990). Gender vs. power: Self posed behavior revisited. Sex Roles, 23 (1 2), 91 98. Wittrock, M. C., & Lumsdaine, A. A. (1977). Instructional psychology. In M. R. Rosenzweig & L. W. Porter (Eds.) Annual Revie w of Psychology, 28 417 459. Wolfgang, A. (Ed.) (1984). Nonverbal behavior: Perspective, applications, intercultural insights Lewiston, NY: C. J. Hogrefe. Wood, W. (1987). Meta analytic review of sec differences in group performance. Psychological Bulletin, 102 53 71.

PAGE 146

134 Appendices

PAGE 147

135 Appendix A. Pilot I: Validation of Scenario Manipulation The purpose of this pilot was to statistically validate the experimental manipu lations of two leadership dimensions (task oriented and person oriented), and leader behavior styles (democratic and autocratic), leader gender (male/female) of a set of eight written scenarios depicting an academic environment. Specifically, this pilot st udy examined whether research participants perceived the experimental conditions as they were designed to be perceived. Thus, the task oriented situation was expected to be perceived as less emotional and more objective than the person oriented situation. Similarly, the democratic communication style was expected to be perceived as friendly and supportive, while the autocratic communication style was expected to be rated as non supportive and cold. Dependent Measures Using a sev en point semantic differential scale, three dependent descriptors were used to describe each of the two leadership dimensions (task/person, and democratic/autocratic). Specifically, the task/person dimension was measured using the following anchors: person al/impersonal, emotional/unemotional, and objective/subjective. The democratic/autocratic communication style was measured using the following anchors: warm/cold, non supportive/supportive, and uncaring/caring. Finally, this study examined whether the cont ent of the scenario was perceived to be plausible. The measures for scenario plausibility were measured using the following anchors: plausible/implausible, realistic/unrealistic, and clear/unclear.

PAGE 148

136 Appendix A. (Continued) Procedure An experimental design was used, in which research participants were randomly assigned to one of the eight scenario versions. Data were collected during undergraduate and graduate classes within the Education department of a large urban university. A total of 18 classes were used for this pilot with an average class size of 20. At the beginning of class, the author or the class instructor explained the purpose of the study (judgment of professor/student interaction). It was stressed that there were no correct answers, and that responses were anonymous. Participation was voluntary, and no incentives were offered for participation. Participants completed the experiment within approximately 10 minutes. Data Analysis Three factorial Analyses of Variances (ANOVAs) were used to examine how students perceived the two leadership dimensions (task/person and democratic/autocratic), as well as students = perception of vignette plausibility. For each dimension, the dependent variables were averaged to form one factor score estimate. Internal reliability estimates were calculated using Cronbach alphas. Finally, effect size estimates using Keppels Omega Squared (Keppel, 1991).

PAGE 149

137 Appendix A. (Continued) Results Sample A total of 356 stude nts participated in this pilot. One hundred forty eight (42%) were male, and 208 (58%) were female. Mean age for female students was 27.4 years (range= 17 18 years, SD= 8.7), and mean age for male students was 29.8 (range= 18 59 years, SD= 10.6). Sevent y one percent (n=266) of the sample was Caucasian. 6.4% (n=22) were African American, 4.6% (n=16) were of Hispanic origin, and 8.1% (n=28) were of Asian/Pacific background. 65% (n=225) were undergraduate students, and 28% (n=98) were graduate students. Rat ings of Task Oriented and Person Oriented Dimension (Scenario Situation) Three items (personal/impersonal, emotional/unemotional, objective/subjective) rated on a seven point semantic differential type scale were used. The higher number represents the des criptors personal, emotional, and subjective. Because two descriptors (personal/impersonal and emotional/unemotional) were reversed to avoid response set during the rating process (Crocker & Algina, 1986), they were reversed during statistical analysis. Fu rther, these three items were combined to form one dependent factor score estimate representing the task/person dimension. Calculated internal consistency estimates as measured by Cronbach alpha for the task/person dimension was .95. This resulting factor score estimate was then analyzed using a 2 (gender of research participant) X 2 (task/person situation) X 2 (professor gender) X 2 (professor democratic/autocratic communication style) univariate analysis of variance (ANOVA). Results (see Table 2 for ANOVA summary) revealed a main effect for participant gender,

PAGE 150

138 Appendix A. (Continued) F (1, 353) = 4.40, p <.04). The relative treatment magnitude as measured by omega squared (w 2 ) was .003. Cohen = s guideline for treatment magnitude measure w 2 suggests that a m agnitude of .01 should be considered a small effect, an obtained w 2 of .06 should be labeled a medium effect, and obtaining w 2 of .15 or greater should be considered a large treatment magnitude (Cohen, 1988). Compared to Cohens standard, the obtaine d treatment magnitude of .003 was very small. Another type of effect size estimation for analyses of variance procedures is the f statistic. This measure represents the standard deviation of the standardized means (Stevens, 1990), and characterizes an f ar ound .1 to be a small effect size, an f around .25 to be a medium effect size, and an f larger than .4 to be a large effect size (Cohen, 1988). Using this measure, the effect size estimate for the participant gender main effect was small ( f =.11).

PAGE 151

139 Appendix A. (Continued) Table 2 Obtained ANOVA Source Table for Task/Person Dimension Source df MS F w 2 f Participant Gender (PaG) 1 4.94 4.40* .003 .11 Professor Gender (PrG) 1 0.55 0.49 Democratic/Autocratic (Pr) Style 1 0.20 0.18 Task/Person Situation (Situation) 1 704.1 627.88** .06 1.33 PaG x PrG 1 0.34 0.30 PaG x Pr Style 1 0.01 0.01 PrG x PrStyle 1 0.91 0.81 PaG x Situation 1 0.73 0.65 PrStyle x Situation 1 3.20 2.85 PaG x PrG x PrStyle 1 0.28 0.25 PaG x PrG x Situation 1 0.10 0.08 PaG x PrStyle x Situation 1 1.54 1.38 PaG x PrG x PrStyle x Situation Error 1 338 0.76 1.12 0.67 *p < .04 **p < .0001 Results further revealed th at research participants rated a task oriented situation as more objective and less personal than the person oriented situation (see Table 3 for summary statistics) F (1,353)= 627.88, p < .0001). The calculated treatment magnitude for the situation main effe ct was medium (w 2 =.06), and the effect size estimate was large ( f =1.33).

PAGE 152

140 Appendix A. (Continued) Table 3 Summary Statistics for Task/Person Factor Score a Mean SD Scale Dimension Coefficient Alpha Males Females Males Females Task Situation ( Objective) .95 3.52 3.82 1.38 1.33 Person Situation (Subjective) .95 6.56 6.74 0.82 0.63 a N = 356 Ratings of the Democratic/Autocratic Dimension (Professor Nonverbal Communication Style) The democratic/autocratic dimension was measured usi ng three items on a seven point semantic differential anchored by warm/cold, non supportive/supportive, and uncaring/caring. A lower score represents an autocratic style, while a higher rating score represents a democratic style. One item (warm/cold) was r eversed for statistical analysis. All three items were combined to create a factor score. Internal consistency estimates as calculated by Cronbach alpha for was .95. This factor score estimate was then analyzed using a 2(gender of research participant) X 2 (task/person situation) X 2 (professor gender) X 2 (professor democratic/autocratic communication style) univariate analysis of variance (ANOVA). Results (see Table 4 for ANOVA summary) revealed a statistically significant main effect for professor gender F (1, 355) = 6.07, p <.004). The calculated

PAGE 153

141 Appendix A. (Continued) treatment magnitude as measured by w 2 was very small (.004), and the calculated effect size was very small ( f =.13). Further, the main effect for task/person was statistically significant F (1, 355) =18.16, p <.0001). Calculated treatment magnitude for this main effect was small (w 2 =.01), and its calculated effect size was small ( f =.22). Further, the democratic/autocratic nonverbal communication style dimension was statistically significant F (1, 355) = 532.43, p <.0001). The calculated treatment magnitude was medium (w 2 =.5), and its effect size was large ( f =1.22). Finally, there was a three way interaction effect between participant gender, professor nonverbal communication style, and task /person situation dimension F (1, 355) = 11.82, P <.0007. The calculated treatment magnitude was small (w 2 =.01), and the effect size estimate was small ( f =.19). Table 5 shows how men and women rated the scenarios under different conditions of communication style and task/person situation. It appears that the perception of democratic behavior (e.g., friendly) may depend on the type of situation, as well as the gender of the rater.

PAGE 154

142 Appendix A. (Continued) Table 4 ANOVA Source Table for Democratic/Autocr atic Communication Style Source Df MS F w 2 f Participant Gender (PaG) 1 0.00 0.00 Professor Gender (PrG) 1 6.88 6.07* .004 .13 Professor Communication Style (PrStyle) 1 603.14 532.43** .50 1.22 Task/Person Situation (Situ ation) 1 20.57 18.16* .02 .22 PaG x PrG 1 0.54 0.48 PaG x Pr Style 1 2.47 2.18 PrG x PrStyle 1 2.61 2.31 PaG x Situation 1 0.03 0.03 PrG x Situation 1 2.24 1.98 PrStyle x Situation 1 0.10 0.96 PaG x PrG x PrStyle 1 0.56 0.50 PaG x PrG x Situation 1 1.88 1.66 PaG x PrStyle x Situation 1 13.40 11.82** .01 .19 PaG x PrStyle x Situation 1 1.42 1.26 PaG x PrG x PrStyle x Situation Error 1 340 0.72 385.15 0.68 *p < 01 **p < .0001

PAGE 155

143 Appendix A. (Continued) Table 5 Summary Statistics of Democratic/Autocratic Nonverbal Communication Style Broken Down by Task/Person Situation and Participant Gender w Mean SD Professor Communication Style (a .95) Task/Person Situation Males Females Males Females Democratic Task 4.57 4.31 1.57 0.74 Democratic Person 4.73 5.33 1.26 1.10 Autocratic Task 1.73 1.88 0.82 0.98 Autocratic Person 2.47 1.91 1.31 .98 N = 357 Ratings of Scenario Plausibili ty This pilot study examined whether the written scenarios were perceived as plausible events within an academic setting. Scenario plausibility was measured using two items on a seven point semantic differential an chored by plausible/implausible, and realistic/unrealistic. Higher ratings represent the implausible dimension, while a lower rating represents the realistic aspect of the dimension. Both items were combined to create a factor score. Calculated internal co nsistency for this factor score as estimated by Cronbach alpha was a stable .82. This factor score estimate was then analyzed using a 2 (gender of research participant) X 2 (task/person situation) X 2 (professor gender) X 2 (professor democratic/autocratic communication style) univariate analysis of variance (ANOVA). Results (see Table 6 for ANOVA summary source table). Results revealed a significant main effect for task/person situation dimension F (1, 352) =48.90, p <.0001).

PAGE 156

144 Appendix A. (Continued) The c alculated treatment magnitude was large (w 2 =.11), along with a medium effect size estimate ( f =.37). Further, a significant interaction between participant gender and professor gender was found F (1, 352) =6.58, p <.01). The corresponding treatment magnitude this interaction effect was small (w 2 =.01), and the effect size estimate was small ( f =.17). The mean rating for a task oriented scenario was 2.07 (SD= 1.01) compared to the mean rating of 3.06 (SD=1.43) for the person oriented scenario. Even though this difference produced a statistical significant difference, the value of 3.06 for the person oriented situation still is well within the range of being considered as realistic and plausible (scale ranged from 1(realistic) to 7 (unrealistic). Table 7 depicts the means and standard deviations produced by the interaction effect between participant gender and task/person situation dimension. It appears that male students rated the task situation as less realistic (see Table 7) than female students. However, both men and women rated the person oriented vignette comparably with regard to plausibility.

PAGE 157

1 45 Appendix A. (Continued) Table 6 ANOVA Source Table for Ratings of Scenario Plausibility Source Df MS F w 2 f Participant Gender (PaG) 1 0.00 0.00 Professor Gender (PrG) 1 2.74 1.72 Professor Communication Style (PrStyle) 1 0.19 .012 Task/Person Situation (Situation) 1 77.92 48.90* .11 .37 PaG x PrG 1 10.48 6.58** .01 .17 PaG x Pr Style 1 4.40 2.76 PrG x PrStyle 1 3.86 2.42 PaG x Situation 1 1.51 .095 PrG x Situation 1 0.02 0.02 PrStyle x Situation 1 0.64 0.40 PaG x PrG x PrStyle 1 0.77 0.48 PaG x PrG x Situation 1 0.48 0.30 PaG x PrStyle x Situation 1 0.05 0.04 PaG x PrStyle x Situation 1 1.11 0.70 PaG x PrG x PrStyle x Situation Error 1 337 0.78 1.59 0.49 *p < .01 **p < .0001

PAGE 158

146 Appendix A. (Continued) Table 7 Summary Statistics of Plausibility Ratings of Male and Female Stu dents Under the Scenario Dimension of Task/Person N Mean SD Task/Person Dimension Male Female Males Females Males Females Task Situation 61 91 2.20 1.98 1.04 0.99 Person Situation 78 117 3.02 3.10 1.53 1.36 Summary The purpose of this pilot study was to verify the experimental manipulation. Results of the experiment revealed that the manipulations of the two leadership dimensions of task/person, and democratic/autocratic were perceived as expected. Further, the depictions of scenarios were perceived by men and women to be realistic and plausible.

PAGE 159

147 Appendix A. (Continued) Scenario of Student Professor Interaction DIRECTIONS: Please read inside this booklet the description of an interaction between a student and professor. After reading the scenario go to the next page where y ou will see a list of descriptors for the professor and the scenario. Each descriptor consists of a pair of extreme opposites separated by seven spaces. Consider each descriptor and mark one of the spaces between each pair of opposites that is closest to your perceptions. Please mark every descriptor using your best judgement EXAMPLES If you feel the person is probably competent, you would mark a space near the word competent. Competent ___ T ___ ___ ___ ___ ___ Incompetent If you feel the pers on is definitely not competent, you would mark a space nearest the word incompetent. Competent ___ ___ ___ ___ ___ ___ T Incompetent In addition, please provide the biographical information requested on the last page. Thank you. If at any point you become uncomfortable with this attitude scale, you may discontinue. DO NOT write your name on this form. All your responses are anonymous.

PAGE 160

148 Appendix A. (Continue d) The student knocks on Professor Smiths door. Do you have a minute? Sitting at her computer, Professor Smith gets up from her desk, nods an encouraging hello to the student and says, Come in, come in. Profess or Smith sits in a chair opposite the student and asks, What seems to be the problem? The student slumps into a chair and looks down on the ground. Professor Smith leans forward toward the student, folds her arms in her lap and patiently waits for the s tudent. The student says, My long term relationship just fell to pieces and I am so upset over it that I really dont know what to do. Looking compassionately at the student, Professor Smith gently replies, I dont know how to help you. Professor Sm ith remains seated across from the student while she lets some time pass. I can see this will take more time than we have right now. Let = s make an appointment. Scenario (sample 1 out of 8)

PAGE 161

149 Appendix A. (Continued) How would you describe the professor = s behavior? warm ___ ___ ___ ___ ___ ___ ___ cold nonsupportive ___ ___ ___ ___ ___ ___ ___ supportive uncaring ___ ___ ___ ___ ___ ___ ___ caring Th e scenario consists of a student presenting a problem How would you describe the nature of this problem ? personal impersonal emotional unemotional objective subjective How would you describe the scenario presented here? plausible implausible realistic ___ unrealistic clear unclear Please turn over L Mark the space between each pair of opposites that is closest to your perception

PAGE 162

150 Appendix A. (Continued) Gender _____M _____F Age ______________________________ School Standing _____Freshman _____Sophomore _____Junior _____Senior _____Graduate _____Other (specify ) __________________________________________ Race/Ethnicity: _____American Indian _____Asian/Pacific Islander _____Black (African American) _____ Hispanic _____White (Non Hispanic) _____Other (please specify):_______________________ THANK YOU Please check the following information about yourself.

PAGE 163

151 Appe ndix B. Pilot II: Manipulation Check on Dependent Measures The purpose of the second pilot was to statistically evaluate the internal consistencies of the dependent measures, as well as evaluate the effectiveness of the data collection procedure. Depende nt Measures Each of the eight scenario versions (see Appendix A) was rated for leadership effectiveness, leader likeability, and leader character attributes. Leader effectiveness Using a five point semantic differential scale, three effectiveness desc riptors anchored with the following descriptors were used: competent/incompetent, effective/ineffective, and well qualified/not at all qualified. In order to have a higher number reflect a higher degree of effectiveness, the scale direction was reversed du ring statistical analysis to follow that logic. Leader likeability Four descriptors arranged on a five point semantic differential scale were used anchored by the following descriptors: critical/tolerant, considerate/inconsiderate, popular/unpopular, and likeable/not likeable. A higher rating reflected a higher degree of likeability, thus, some of the items re scaled during statistical analysis to fit that logic. Leader character attributes Fifteen descriptors were arranged on five point semantic differ ential reflecting male and female stereotypes. These 15 character attributes were anchored with the following positive and negative anchors: powerful/powerless, timid/forceful, hardworking/lazy, persistent/gives up easily, soft/tough, fair/not fair, respon sible/irresponsible, not helpful/helpful, cooperative/not cooperative, trustworthy/untrustworthy, aggressive/not aggressive, independent/dependent,

PAGE 164

152 Appendix B (Continued) dominant/submissive, objective/subjective, and unprepared/prepared. A lower rating on this scale corresponded to a negative character evaluation, while a higher rating reflected a positive character attributes. Additional questions In addition, six questions were posed to evaluate professor appropriateness, communication skills, popul arity with students, as well as questions referring to professors chances for future promotions. Procedure An experimental design was used, in which research participants from two undergraduate and one graduate class recruited from the Education depar tment of a large urban university. After one of eight scenario versions was randomly assigned to participating students, they were asked to complete the forms. Participation was voluntary, and no incentives were offered for participation. On average it too k about 8 minutes to complete the experiment. Further, students who participated in the study were encouraged to give verbal or written feedback with regard to the experiment. It was stressed that all responses were anonymous. It was stressed that all resp onses were anonymous. Data Analysis Internal reliability estimates were calculated using Cronbach alphas for the descriptors of leader effectiveness, leader likeability, and leader character attributes. Student comments were evaluated and used to make des ign improvements.

PAGE 165

153 Appendix B (Continued) Results Sample Sixty students participated in this pilot. Twenty nine (48.3%) were male, and 31 (51.7%) were female. Mean age for female students was 32.13 years (SD= 7.5 years, range= 21 48 years), and the mean age for male students was 35.1 (SD= 8.6 years, range= 22 57 years). For this sample, 79.3% (n=37) were Caucasian, 8.6% (n=5) were African American, 3.5% (n=2) were Hispanic, and 3.5% (n=2) marked an Asian/Pacific background. 37% (n=23) of the sample were undergraduate students, and 63% (n=37) were graduate students. Reliability Estimates (Cronbach Alphas) for the Dependent Measures Leader effectiveness The three effectiveness indicators (competent/incompetent, effective/ineffective, well qualified/not at all qualified) were combined equally in order to receive a leader effectiveness factor score estimate. Calculated internal consistency estimates as measured by Cronbach alpha was .84. Leader likeability The four indicators for leader likeability (cr itical/tolerant, considerate/inconsiderate, popular/unpopular, likeable/not likeable) were combined resulting in one likeability factor score estimate with a calculated internal consistency estimate of .86 (Table 8).

PAGE 166

154 Appendix B (Continued) Table 8. Means and Standard Deviations for Factor Scores by Booklet Form Factor Scores Booklet Form Effectiveness (a = .84) Likeability (a = .86) Professional Traits (a = .82) Dominance (a = .82) N Mean SD Mean SD Mean SD Mean SD Perso nal Situation/Autocratic Female 8 3.13 1.5 2.85 1.0 3.24 0.4 3.38 1.0 Personal Situation/Democratic Female 12 3.72 0.9 3.93 0.8 3.79 0.7 3.12 0.6 Task Situation/Democratic Female 5 3.60 0.5 2.95 0.8 3.49 0.7 3.65 0.8 Ta sk Situation/Autocratic Female 6 2.60 0.8 2.13 0.6 3.23 0.5 4.38 0.7 Personal Situation/Democratic Male 7 3.52 0.7 4.02 1.0 3.32 0.7 2.64 0.6 Personal Situation/Autocratic Male 7 3.24 0.9 2.64 0.8 3.29 0.4 3.96 0.5 Ta sk Situation/Autocratic Male 7 3.14 1.1 1.79 0.9 2.90 0.5 4.14 0.4 Task Situation/Democratic Male 8 3.33 0.8 3.25 0.7 3.52 0.7 3.50 0.4 Leader character traits Initially, all 15 character traits (powerful/powerless, timid/for ceful, hardworking/lazy, persistent/gives up easily, soft/tough, fair/not fair, responsible/irresponsible, not helpful/helpful, cooperative/not cooperative, trustworthy/untrustworthy, aggressive/not aggressive, independent/dependent, dominant/submissive, o bjective/subjective, unprepared/prepared) were combined to obtain one factor score estimate. The calculated internal consistency estimate for this factor was .72. After closer inspection, the following four character attributes timid/forceful, soft/tough, aggressive/not aggressive, and dominant/submissive contributed negatively to the item total correlation. As a result, these four items were

PAGE 167

155 Appendix B (Continued) separated from the rest of the items and treated as a separate factor. Internal consistency estimate for the four item dominant factor score was calculated to be .82. The remaining eleven items were combined to form the leader professional traits factor score and the calculated internal consistency estimate for this factor score was .82. Table 8 shows the means and standard deviations for all four factor scores for each of the eight booklet forms. Evaluation of Student Feedback Overall, the directions provided by the instructor/author were adequate, that is, students who participated in this st udy, followed the directions without difficulty. However, the most frequently mentioned comment students had was that the content of the scenario did not hold enough information in order for them to for example answer questions pertaining to professor futu re promotions etc. As a result of these comments, the author will add a neutral statement with regard to the professors career at the University. Summary The purpose of this pilot study was to adjust for possible procedural difficulties in data collectio n, as well as examining internal consistency estimates after combining proposed items for leader effectiveness, leader likeability, leader work attitude, and leader dominance into factor score estimates. Results of this study revealed that the calculated i nternal consistency estimates (Cronbach alphas) were stable. Further, based

PAGE 168

156 Appendix B (Continued) on the results of student feedback, the scenarios were adjusted to include sufficient information about the professor in order to make judgments about his/her future.

PAGE 169

157 Appendix C. Task Vignette Booklets Scenario of Student Professor Interaction DIRECTIONS: Please read inside this booklet the description of an interaction between a student and professor. After reading the scenario go to the next page where you will see a list of descriptors for the professor and the sce nario. Each descriptor consists of a pair of extreme opposites separated by seven spaces. Consider each descriptor and mark one of the spaces between each pair of opposites that is closest to your perceptions. Please mark every descriptor using your bes t judgement EXAMPLES If you feel the person is probably competent, you would mark a space near the word competent. Competent ___ T ___ ___ ___ ___ ___ Incompetent If you feel the person is definitely not competent, you would mark a space nearest the word incompetent. Competent ___ ___ ___ ___ ___ ___ T Incompetent In addition, please provide the biographical information reques ted on the last page. Thank you. If at any point you become uncomfortable with this attitude scale, you may discontinue. DO NOT write your name on this form. All your r esponses are anonymous.

PAGE 170

158 Appendix C (Continued) The student knocks on Professor Smiths door. Do you have a minute? Professor Smith frowns while looking at her watch, All right, what is it? The student steps into her office, Do you remember last class you encouraged our group to participate more during class? Professor Smith, sitting straight in her chair, continues to stare at the student. Well, the student continues, I t hink the problem is that we are so busy writing down information that we cannot concentrate on your lecture. So, if we could make copies of your overheads before each class it would make it much easier for us to participate. Leaning back in her chair, Pr ofessor Smith lowers her eyebrows. Slowly she shakes her head, points her finger at the student and declares, You want me to copy my notes for you before each class? Oh no, the student replies, One of the students would be responsible for all the co pies. Still staring at the student, Professor Smith stands up abruptly and motions the student to the door. Ill let you know by Monday. Scenario Autocratic Female Professor in Task Situation

PAGE 171

159 Appendix C (Continued) The student knocks on Pr ofessor Smiths door. Do you have a minute? Professor Smith smiles and nods while glancing at her watch, All right, what is it? The student steps into her office, Do you remember last class you encouraged our group to participate more during class? Professor Smith leans forward in her chair, nods and smiles encouragingly. Well, the student continues, I think the problem is that we are so busy writing down information that we cannot concentrate on your lecture. So, if we could make copies of your overheads before each class it would make it much easier for us to participate. Professor Smith gently tugs at her collar and nods her head while she is listening to the student. Then she asks, You want me to copy my notes for you before each class? O h no, the student replies, One of the students would be responsible for all the copies. Professor Smith now smiles and gets up to accompany the student to the door while she says, Ill let you know by Monday. Scenario Democratic Female Professor in Task Situation

PAGE 172

160 Appendix C (Continued) The student knocks on Professor Smiths door. Do you have a minute? Professor Smith frowns while looking at his watch, All right, what is it? The student steps into his office, Do you remember last cl ass you encouraged our group to participate more during class? Professor Smith, sitting straight in his chair, continues to stare at the student Well, the student continues, I think the problem is that we are so busy writing down information that we ca nnot concentrate on your lecture. So, if we could make copies of your overheads before each class it would make it much easier for us to participate. Leaning back in his chair, Professor Smith lowers his eyebrows. Slowly he shakes his head, points his fi nger at the student and declares, You want me to copy my notes for you before each class? Oh no, the student replies, One of the students would be responsible for all the copies. Still staring at the student, Professor Smith stands up abruptly and motions the student to the door, Ill let you know by Monday. Scenario Autocratic Male Professor in Task Situation

PAGE 173

161 Appendix C (Continued) The student knocks on Professor Smiths door. Do you have a minute? Professor Smith smiles and nods while glancing at his watch, All right, what is it? The student steps into his office, Do you remember last class you encouraged our group to participate more during class? Professor Smith leans forward in his chair nods and smiles encouragingly. Wel l, the student continues, I think the problem is that we are so busy writing down information that we cannot concentrate on your lecture. So, if we could make copies of your overheads before each class it would make it much easier for us to participate. Professor Smith gently tugs at his collar and nods his head while he is listening to the student. Then he asks, You want me to copy my notes for you before each class? Oh no, the student replies, One of the students would be responsible for all th e copies. Professor Smith now smiles and gets up to accompany the student to the door while he says, Ill let you know by Monday. Scenario Democratic Male Professor in Task Situation

PAGE 174

162 Appendix C (Continued) DEPENDENT MEASURES How would you describe the professor? competent ___ ___ ___ ___ ___ incompetent effective ___ ___ ___ ___ ___ ineffective well qualified ___ ___ ___ ___ ___ not at all qualified critical ___ ___ ___ ___ ___ tolerant considerate ___ ___ ___ ___ ___ inconsiderate popular ___ ___ ___ ___ ___ unpopular likeable ___ ___ ___ ___ ___ not likeable powerful ___ ___ ___ ___ ___ powerless timid ___ ___ ___ ___ ___ forceful hardworking ___ ___ ___ ___ ___ lazy persistent ___ ___ ___ ___ ___ gives up easily soft ___ ___ ___ ___ ___ tough fair ___ ___ ___ ___ ___ not fair responsible ___ ___ ___ ___ ___ irresponsible not helpful ___ ___ ___ ___ ___ helpful cooperative ___ ___ ___ ___ ___ not cooperative trust worthy ___ ___ ___ ___ ___ untrustworthy aggressive ___ ___ ___ ___ ___ not aggressive independent ___ ___ ___ ___ ___ dependent dominant ___ ___ ___ ___ ___ submissive objective ___ ___ ___ ___ ___ subjective unprepar ed ___ ___ ___ ___ ___ prepared In your opinion the student in this scenario is Male Female ( Please circle one )

PAGE 175

163 Appendix C (Continued) Strongly Agree Agree Uncertain Disagree Strongly Disagree 1. The professor is handling the situation well. SA A U D SD 2. The professor has potent ial for future promotions. SA A U D SD 3. The professor = s reaction to the student is appropriate. SA A U D SD 4. The professor has great communication skills. SA A U D SD 5. Students respect this professor. SA A U D SD 6. This professor is well liked by students. SA A U D SD Gender _____M _____F Age ______________________________ School Standing _____Freshman _____Sophomore _____Junior _____Senior _____Gr aduate _____Other (specify) Major: __________________________________________ ( Please write in ) Race/Ethnicity: _____American Indian _____Asian/Pacific Islander _____Black (African American) _____ Hispanic _____White (Non Hispanic) _____Other (please sp ecify):_____________ Directions: Below are statements describing the professor. Please indicate the extent of your agreement o r disagreement by circling your response in the box which best corresponds to your beliefs. Please check the following information about yourself.

PAGE 176

164 Appendix D. Personal Vignette Booklets Scenario of Student Professor Interaction DIRECTIONS: Please read inside this booklet the description of an interaction between a student and professor. After reading the scenario go to the next page where you will see a list of descriptors for the professor and the scenario. Ea ch descriptor consists of a pair of extreme opposites separated by seven spaces. Consider each descriptor and mark one of the spaces between each pair of opposites that is closest to your perceptions. Please mark every descriptor using your best judgemen t EXAMPLES If you feel the person is probably competent, you would mark a space near the word competent. Competent ___ T ___ ___ ___ ___ ___ Incompetent If you feel the person is definitely not competent, you would mark a space nearest the word i ncompetent. Competent ___ ___ ___ ___ ___ ___ T Incompetent In addition, please provide the biographical information requested on the last page. Thank you. If at any point you become uncomfortable with this attitude scale, you may discontinue. DO NOT write your name on this form. All your responses a re anonymous.

PAGE 177

165 Appendix D (Continued) The stu dent knocks on Professor Smiths door. Do you have a minute? Sitting at her computer, Professor Smith does not take her eyes off her computer screen, but continues working on it, as she impatiently says, Come in, come in. Still working on the compute r she asks, What seems to be the problem? The student slumps into a chair and looks down on the ground. Professor Smith glances over at the student, sighs, crosses her arms over her chest and remains silent. The student says, My long term relationship just fell to pieces and I am so upset over it that I really dont know what to do. Staring at the student, Professor Smith finally says, I dont know how to help you. Professor Smith checks to see whether her files on the computer are saved. To be sur e, she hits the save button again. Then she turns back to the student and remarks, I can see this will take more time than we have right now. Lets make an appointment. Scenario Autocratic Female Professor in Personal Situation

PAGE 178

166 Appendix D (Continued) The student knocks on Professor Smiths door. Do you have a minute? Sitting at her computer, Professor Smith gets up from her desk, nods an encouraging hello to the student and says, Come in, come in. Professor Smith sits in a chair opposite the student and asks, What seems to be the problem? The student slumps into a chair and looks down on the ground. Professor Smith leans forward toward the student, folds her arms in her lap and patiently waits for the student. The student says, My long term relationship just fell to pieces and I am so upset over it that I really dont know what to do. Looking compassionately at the student, Professor Smith gently replies, I dont know how to help you. Professor Smith remains seated across from the student while she lets some time pass. I can see this will take more time than we have right now. Lets make an appointment. Scenario Democratic Female Professor in Pe rsonal Situation

PAGE 179

167 Appendix D (Continued) The student knocks on Professor Smiths door. Do you have a minute? Sitting at his computer, Professor Smith does not take his eyes off his computer screen, but continues working on it as he impatiently says, Come in, come in. Still working on the computer he asks, Wha t seems to be the problem? The student slumps into a chair and looks down on the ground. Professor Smith glances over at the student, sighs, crosses his arms over his chest and remains silent. The student says, My long term relationship just fell to piec es and I am so upset over it that I really dont know what to do. Staring at the student, Professor Smith finally says, I dont know how to help you. Professor Smith checks to see whether his files on the computer are saved. To be sure, he hits the s ave button again. Then he turns back to the student and remarks, I can see this will take more time than we have right now. Lets make an appointment. Scenario Autocratic Male Professor in Personal Situation

PAGE 180

168 Appendix D (Continued) The stu dent knocks on Professor Smiths door. Do you have a minute? Sitting at his computer, Professor Smith gets up from his desk, nods an encouraging hello to the student and says, Come in, come in. Professor Smith sits in a chair opposite the student and asks, What seems to be the problem? The student slumps into a chair and looks down on the ground. Professor Smith leans forward toward the student, folds his arms in his lap, and patiently waits for the student. The student says, My long term relationsh ip just fell to pieces and I am so upset over it that I really dont know what to do. Looking compassionately at the student, Professor Smith gently replies, I dont know how to help you? Professor Smith remains seated across from the student while he lets some time pass. He finally says in a comforting voice, I can see this will take more time than we have right now. Lets make an appointment. Scenario Democratic Male Professor in Personal Situation

PAGE 181

169 Appendix D (Continued) INDEPENDENT MEASURES How would you describe the professor? likable __ ___ ___ ___ ___ not likeable soft ___ ___ ___ ___ ___ tough competent ___ ___ ___ ___ ___ incompetent trustworthy ___ ___ ___ ___ ___ untrustworthy responsible ___ ___ ___ ___ ___ irresponsible effective ___ __ ___ ___ ___ ineffective critical ___ ___ ___ ___ ___ not likeable unprepared ___ ___ ___ ___ ___ prepared aggressive ___ ___ ___ ___ ___ not aggressive well qualified ___ ___ ___ ___ ___ not at all qualified persistent ___ ___ ___ ___ ___ gives up easily objective ___ ___ ___ ___ ___ subjective hardworking ___ ___ ___ ___ ___ lazy considerate ___ ___ ___ ___ ___ inconsiderate dominant ___ ___ ___ ___ ___ submissive not helpful ___ ___ ___ ___ ___ helpful cooperat ive ___ ___ ___ ___ ___ not cooperative powerful ___ ___ ___ ___ ___ powerless timid ___ ___ ___ ___ ___ forceful independent ___ ___ ___ ___ ___ dependent fair ___ ___ ___ ___ ___ not fair popular ___ ___ ___ ___ ___ unpopula r capable ___ ___ ___ ___ ___ not capable influential ___ ___ ___ ___ ___ not influential In your opinion the student in this scenario is Male Female

PAGE 182

170 Appendix D (Continued) Strongly Agree Agree Uncertain Disagree Strongly Disagree 1. The professor is handling the situation well. SA A U D SD 2. The professor has potential for future promotions. SA A U D SD 3. The professor = s reaction to the student is appropriate. SA A U D SD 4. The professor has great communication ski lls. SA A U D SD 5. Students respect this professor. SA A U D SD 6. This professor is well liked by students. SA A U D SD Gender _____M _____F Age ________________________ ______ School Standing _____Freshman _____Sophomore _____Junior _____Senior _____Graduate _____Other (specify) Major: __________________________________________ ( Please write in ) Race/Ethnicity: _____American Indian _____Asian/Pacific Islander ____ _Black (African American) _____ Hispanic _____White (Non Hispanic) _____Other (please specify):_____ Directions: Below are statements describing the professor. Please indicate the extent of your agreement or disagreement by circling your response in the box which best corresponds to your beliefs. Please check the following information about yourself.

PAGE 183

171 Appendix E. Survey Introduction Letter Monday, August 06, 2001 My name is Michela LaRocca, and I am a Doctoral Candidate in the Department of Meas urement and Research within the College of Education at the University of South Florida. I have successfully defended my dissertation proposal, and have received Institutional Review Board (IRB) approval to conduct an experiment designed to examine how ma le and female students perceive leadership traits in male and female professors. The actual experiment consists of versions of a written scenario, which is academic in nature. Leader in this case is defined in the role of the professor. Versions were crea ted in order to vary professor gender, professor behavior, and the situation in which an interaction is described between the professor and one of the students in the class. Extensive pilot testing was conducted to verify the manipulations of the variables as well as the outcome measures. Student participation is voluntary. Pilot studies indicated that most students were able to follow the instructions on the front page of the booklet without incidents. The least disruptive method was to distribute the bo oklets before class. Students will likely take about 8 10 minutes reading the scenario, and filling out the questions. I should be able to have results available for instructors interested by the summer of 2002. This research is very important for under standing how males and females are perceived in leadership roles. Thank you so much for your consideration, and if you have any questions, please feel free to contact me via phone: (813) 792 8501, or via email: michela1@tampabay.rr.com

PAGE 184

172 About the Author Michela A. LaRocca earned a Bachelor of Arts degree in Psychology from University of Kentucky in 1986, and a Master of Arts in Clinical Psychology from Western Kentucky University in 1989. While e arning the PhD. in measurement and research, Ms LaRocca worked as a statistical consultant to other researchers. Ms. LaRocca co authored three publications (she was first author on one publication). Her research focuses on the development of attitude scal es. In addition, Ms. LaRocca examines theoretical aspects of gender roles as they pertain to current gender issues.


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 2200445Ka 4500
controlfield tag 006 m d
007 cr bn
008 031007s2003 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000103
035
(OCoLC)52281590
9
AJL4046
b SE
040
FHM
c FHM
049
FHME
090
LB3051
1 100
LaRocca, Michela A.
0 245
Perception of leadership qualities in higher education
h [electronic resource] :
impact of professor gender, professor leader style, situation, and participant gender /
by Michela A. LaRocca.
260
[Tampa, Fla.] :
University of South Florida,
2003.
502
Thesis (Ph.D.)--University of South Florida, 2003.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 184 pages.
520
ABSTRACT: This experimental study used eight written vignettes to analyze the effects of professor gender, professor leadership style (democratic/autocratic), and type of situation (task/personal) and participant gender on evaluations of professors competence, likeability and masculinity characteristics. Undergraduates from the College of Arts and Science (N=932; Males=464, Females=467), and the College of Education (N=722; Males=140, Females=582) were used. Results indicated that research participants rated democratic professors significantly more competent, likeable, and more feminine than autocratic professors. Contrary to expectations derived from gender spill-over and gender congruency theories, male participants did not rate female professors more negatively than their male counterparts when they acted autocratically in a personal situation (i.e., gender incongruent manner.) Exploratory results revealed trends that are discussed along with theoretical and practical implications.
590
Co-adviser: Dedrick, Robert
650
Leadership
x Psychological aspects.
Sex differences
Social aspects.
653
leadership dimension.
task/personal.
democratic/autocratic.
theoretical implications.
gender stereotypes.
690
Dissertations, Academic
z USF
Measurement and Evaluation
Doctoral.
773
t USF Electronic Theses and Dissertations.
949
FTS
SFERS
ETD
LB3051 (ONLINE)
sv 5/19/03
4 856
u http://digital.lib.usf.edu/?e14.103