Evaluations of Prejudice and Stereotype Research

MISSING IMAGE

Material Information

Title:
Evaluations of Prejudice and Stereotype Research How Do People View the Validity of Psychological Science and Scientists?
Physical Description:
1 online resource (103 p.)
Language:
english
Creator:
Schember, Tatiana O
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Psychology
Committee Chair:
Webster, Gregory Daniel
Committee Members:
Moradi, Banafsheh
Shepperd, James A
Algina, James J

Subjects

Subjects / Keywords:
authority -- perception -- prejudice -- stereotype
Psychology -- Dissertations, Academic -- UF
Genre:
Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
Public perceptions of psychology’s validity as a science has direct implications for funding and dissemination of knowledge, yet surprisingly little research has examined public perception of scientific research, let alone the specific features of psychological scientists themselves that affect the perceived validity of psychological science. The current studies examine whether personal characteristics of stereotyping and prejudice researchers—such as gender, ethnicity, and university affiliation—influence the perceived quality of prejudice research and the psychologist’s credibility as a scientist. Additionally, Study 2 examines participants’ opinions on what constitutes scientific expertise. Study 1 was a 2 x 2 x 3 between-person experiment examining the effects of researcher gender,researcher ethnicity, and university prestige on participants’ perceptions of research quality and scientist credibility. Results showed no effects of researcher gender or race, but showed a main effect of university prestige.Study 2 was conceptually similar to Study 1, addressing some of the limitations(weak manipulations, international sample, participant fatigue, overly complex design). Study 2 was a between-person experiment examining the effect of researcher gender on participants’ perception of research quality and scientist credibility. Results showed no significant main effects of gender, but showed some significant gender x covariate interactions; participants in the male researcher condition rated the research differently depending on their level of sexism and their preferences for firsthand experience versus detachment in a scientist.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Tatiana O Schember.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: Webster, Gregory Daniel.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2014-08-31

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2013
System ID:
UFE0045776:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 EVALUATIONS OF PREJUDICE AND STEREOTYPE RESEARCH: HOW DO PEOPLE VIEW THE VALIDITY OF PSYCHOLOGICAL SCIENCE AND SCIENTISTS? By TATIANA OROZCO SCHEMBER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF F LORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Tatiana Orozco Schember

PAGE 3

3 To John

PAGE 4

4 ACKNOWLEDGMENTS First, I would like to thank Dr. Gregory Webster for inviting me to the University of Florida to pursue my graduate studies and for being an extraordinary mentor I could not ask for a better advisor. I would also like to thank the other members of my superviso ry committee, Dr. James Shepperd, Dr. Bonnie Moradi; and Dr. James Algina; their help and advice has been immeasurable. Special thanks go to my family and close friends, especially to my mom, Robert, Ellie, and Rachel, for all their support and for believing in me. Finally, I th ank my husband, John, for all his help, love, understanding, and encouragement.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 Attitudes, Social Influence, and Authority ................................ ................................ 16 Overview of the Present Research ................................ ................................ ......... 21 2 STUDY 1 ................................ ................................ ................................ ................. 23 Method ................................ ................................ ................................ .................... 23 Participants ................................ ................................ ................................ ....... 23 Design and Variables ................................ ................................ ....................... 2 4 Procedure ................................ ................................ ................................ ......... 24 Dependent Measures and Demographics ................................ ........................ 25 Predictions ................................ ................................ ................................ .............. 25 Results ................................ ................................ ................................ .................... 26 Discussion ................................ ................................ ................................ .............. 27 3 STUDY 2 ................................ ................................ ................................ ................. 37 Method ................................ ................................ ................................ .................... 38 Participants ................................ ................................ ................................ ....... 38 Design and Variables ................................ ................................ ....................... 38 Procedure ................................ ................................ ................................ ......... 39 Dependent measures and manipulation check ................................ .......... 40 Individual difference measures ................................ ................................ .. 41 Predictions ................................ ................................ ................................ .............. 42 Predictions for Independent Variables ................................ .............................. 42 Predictions for Individual Difference Measures ................................ ................ 42 Results ................................ ................................ ................................ .................... 44 Data Preparation and Preliminary Analyses ................................ ..................... 45 Main Analyses ................................ ................................ ................................ .. 47 Dependent variable: evaluations of research quality ................................ .. 49 Dependent variable: funding and future research ................................ ...... 52 Depende nt variable: researcher credibility ................................ ................. 56 Discussion ................................ ................................ ................................ .............. 57

PAGE 6

6 Researcher gender effects ................................ ................................ ............... 57 Expertise covariate effects ................................ ................................ ............... 58 Benevolent sexism covariate effects ................................ ................................ 59 Hostile sexism covariate effects ................................ ................................ ....... 60 Participant gender covariate effects ................................ ................................ 61 4 OVERALL DISCUSSION ................................ ................................ ........................ 84 Limi tations and Future Directions ................................ ................................ ........... 86 Implications and Conclusions ................................ ................................ ................. 89 APPENDIX A EVALUATIONS OF RESEARCH AND RESEARCHER CREDIBILITY S CALE ...... 92 B MANIPULATION CHECK, ATTENTION CHECKS, & QUALITY CHECKS (STUDY 2) ................................ ................................ ................................ .............. 95 C WHAT CONSTITUTES AN EXPERT (WCE) ................................ .......................... 96 D AMBIVALENT SEXISM INVENTORY (ASI; GLICK & FISKE, 1996) ...................... 97 E DEMOGRAPHICS ................................ ................................ ................................ .. 99 REFERENCES ................................ ................................ ................................ ............ 101 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 103

PAGE 7

7 LIST OF TABLES Table page 2 1 Contrast Cod ing for Multiple Regression Predictors (Study 1) ............................ 30 2 2 Correlations for Dependent Variables and Descriptive Statistics (Study 1) ........ 31 2 3 Cell Means (Study 1) ................................ ................................ .......................... 32 2 4 Regression Evaluations of Research (Study 1) ................................ ............... 33 2 5 Regression Funding & Future Researc h (Study 1) ................................ .......... 34 2 6 Regression Researcher Credibility (Study 1) ................................ ................... 35 3 1 Squared Multiple Correlations ................................ ................................ ............ 62 3 2 Goodness of Fit ................................ ................................ ................................ .. 62 3 3 Contrast Coding for Multiple Regression Predictors (Study 2) ............................ 62 3 4 Correlations for Dependent Variables and Covariates; Descriptive Statistics (Study 2) ................................ ................................ ................................ ............. 63 3 5 Cell Means (Study 2) ................................ ................................ .......................... 63 3 6 Regression Evaluations of Research, WCE covariate (Study 2) ..................... 64 3 7 Regression Funding & Future Research, WCE covariate (Study 2) ................ 65 3 8 Regression Researcher Credibility, WCE covariate (Study 2) ......................... 66 3 9 Regression Evaluations of Research, ASI benevolent covariate (Study 2) ..... 67 3 10 Regression Funding & Future Research, ASI benevolent covariate (Study 2) ................................ ................................ ................................ ............. 68 3 11 Regression Researcher Credibility, ASI benevolent covariate (Study 2) ......... 69 3 12 Regression Evaluations of Research, ASI hostile covariate (Study 2) ............ 70 3 13 Regression Funding & Future Research, ASI hostile covariate (Study 2) ....... 71 3 14 Regression Researcher Credibility, ASI hostile covariate (Study 2) ................ 72 3 15 Regression Ev aluations of Research, participant gender (Study 2) ................. 73

PAGE 8

8 3 16 Regression Funding & Future Research, participant gender (Study 2) ........... 74 3 17 Regression Researcher Credibility, participant gender (Study 2) .................... 75

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Summary Article of Ay res, I., & Sie gelman, P. (1995) ................................ ........ 36 3 1 Screenshot: Summary Article of Petersen, J. L., & Hyde J. (2010) ................... 76 3 2 Pictures used in female resear cher, male res earcher, and control conditions .... 77 3 3 Factor structure for the full (8 item) What Constitutes an Expert scale ............... 78 3 4 Factor structure for the reduced (5 item) What Constitutes an Expert scale ...... 79 3 5 Predicted scores on the evaluations of research quality dependent variable for one SD above and one SD bel ow the mean on the What Constitutes an Expert covariate. ................................ ................................ ................................ 80 3 6 Predicted scores on the evaluations of research quality dependent variable for one SD above and one SD below the mean on the hos tile sexism (ASI hostile) covariate. ................................ ................................ ............................... 80 3 7 Predicted scores on the funding & future research dependent variable for one SD above and one SD below the mean on the What Constitutes an Expert co variate. ................................ ................................ ................................ ............ 81 3 8 Predicted scores on the funding and future research dependent variable for one SD above and one SD below the mean on the benevolent sexism (ASI benevolent) covariate. ................................ ................................ ........................ 81 3 9 Predicted scores on the funding and future research dependent variable for one SD above and one SD below the mean on the benevolent sexism (ASI benevolent) covariate. ................................ ................................ ........................ 82 3 10 Predicted scores on the funding and future research dependent variable for one SD above and SD below the mean on the hostile sexism (ASI hostile) covariate. ................................ ................................ ................................ ............ 82 3 11 Predicted scores on the researcher credibility dependent variable for one SD above and one SD below the mean on the benevolent sexism (ASI benevolent) covariate. ................................ ................................ ........................ 83

PAGE 10

10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EVALUATIONS OF PREJUDICE AND STEREOTYPE RESEARCH: HOW DO PEOPLE VIEW THE VALIDITY OF PSYCHOLOGICAL SCIENCE A ND SCIENTISTS? By Tatiana Orozco Schember August 2013 Chair: Gregory D. Webster Major: Psychology for funding and dissemination of knowledge, yet surprisingly little resea rch has examined public perception of scientific research, let alone the specific features of psychological scientists themselves that affect the perceived validity of psychological science. The current studies examine whether personal characteristics of s tereotyping and prejudice researchers such as gender, ethnicity, and university affiliation at constitutes scientific expertise. Study 1 was a 2 2 3 between person experiment examining the effects of researcher gender, researcher ethnicity, and university prestige on ults showed no effects of researcher gender or race, but showed a main effect of university prestige. Study 2 was conceptually similar to Study 1, addressing some of the limitations (weak manipulations, international sample, participant fatigue, overly com plex design). Study 2 was a between person experiment examining the effect of researcher gender on

PAGE 11

11 significant main effects of gender, but showed some significant gen der covariate interactions; participants in the male researcher condition rated the research differently depending on their level of sexism and their preferences for firsthand experience versus detachment in a scientist.

PAGE 12

12 CHAPTER 1 INTRODUCTION In Augu st of 2010, Chief U.S. District Judge Vaughn Walker ruled that California Proposition 8 the infamous gay marriage ban proposition violated the due process and equal protection clauses of the fourteenth amendment of the U.S. Constitution (Walker, 2010). In term, same sex relationship and therefore could not be an impartial judge (Badash, 2011). Walker, however, told reporters that he never considered stepping down from ruling on the case because his sexual orientation was irrelevant in deciding whether Proposition 8 violated the U.S. Constitution. Further, he asserted that it would be inappropriate for judges to recuse themselves ba sed on ethnicity, national origin, gender, or sexual orientation, because as prob lematic for his suitability to rule on the case. However, people could argue that his membership in the potentially affected group (regardless of any intention to marry his partner) gives him special insight into the case, thereby enhancing his suitability to rule on the case. One could argue that the insights of partnered gay, lesbian, and bisexual people are necessary and enhance the decision making process; firsthand experience of being denied a right could make one better at recognizing the issue as one of civil liberties. As was the case with Judge Walker and the Proposition 8 ruling, people generally recognize the importance of authorities but do not always agree with an ary as

PAGE 13

13 characteristics of the audience, or the communications between authority and audience (Strong, 1968). These general ideas can apply more specifically to lay perc eptions of psychological scientists and their research. Although people may recognize that scientists are authorities in their respective fields, they may sometimes disagree with ristics of a scientist can influence the perception of that scientist and his or her research, such as characteristics of a lay person can influence the perceptions of scientis ts and their research, such as personal opinions regarding what type of person constitutes a scientific authority for the given field. Although many psychological scientists may prefer to focus their efforts on producing rigorous scientific research, the perceptions of this research and of the psychologists who conduct it are of paramount importance. The perception of psychologists as competent scientists has direct implications for funding and dissemination of knowledge (Lilienfeld, 2012). People may dist inguish between the goals of basic and applied science, but it is probably the case that even basic psychological scientists desire, through their research, to solve problems outside of the otyping, and discrimination most surely hope that their research adds to a knowledge base that will help reduce these problems. A positive general impression of these psychologists and their research is important for the dissemination and application of t heir research. Unfortunately, the

PAGE 14

14 current view of psychology is largely negative (Lilienfeld, 2012). Lilienfeld (2012) police questionable practices in the field. Never theless, some of this negativity is likely due to the way psychologists have portrayed or perhaps more accurately, neglected to try to portray themselves and their research to the general public (Lilienfeld, 2012). However, with all their accumulated knowl edge of human cognition and behavior, psychologists have the ability to more accurately portray psychological science to the general public and to shift public perception towards a greater match between perception and actual state of scientific rigor. Giv en the current negative perceptions of psychological science and scientists, and the fact that these negative perceptions have such strong implications for funding and dissemination of knowledge (Lilienfield, 2012), it is surprising that psychologists have conducted little research specific to this issue. There has been some research on perceptions of scientific research, however. For example, Wilson et al. (1993) showed that scientists were more likely to overlook methodological flaws in a study when the t opic was considered important. Two scientist samples medical school faculty and research psychologists read brief summaries of methodologically flawed, fictitious studies and then completed questionnaires evaluating the scientific rigor and publishability research). Both scientist samples were more willing to overlook methodological flaws (e.g., causal inferences from correlational data) when the topic of study was perceived as impor tant relative to when it was perceived as less important (e.g., heart disease vs. heartburn).

PAGE 15

15 In a study on perceptions of scientific research and methodology, Munro (2010) found that people will dismiss results of scientific research that are at odds wit h their own beliefs, claiming that the topic of research is not suitable for scientific investigation. short research summary articles that either confirmed or disc onfirmed their belief that homosexuality is a mental illness. Participants whose beliefs were disconfirmed by the summary article reported greater agreement with an item stating that the topic addressed in the summaries (homosexuality) could not be examine d through scientific methods. Participants in the belief disconfirming condition were more likely to report that other, unrelated questions (e.g., regarding the effectiveness of spanking or the existence of clairvoyance) could not be answered using scienti fic methods. Munro Beyond these studies, there is little other research (to my knowledge) that examines perceptions of scientific research, let alone scientific prejudice and stereotyp e research specifically. The purpose of the research presented in this dissertation is to examine some characteristics of both psychological scientists and the general public that affect the perceived quality of the psychological research and the credibili ty of psychological scientists. In particular, this research examines the variables that affect the perception of prejudice, discrimination, and stereotyping research, as well as the psychologists who conduct it. The driving questions of this research are what constitutes an authority in prejudice research, and what characteristics of a prejudice rick, Neuberg,

PAGE 16

16 & Cialdini, 2010), but is an expert in prejudice and stereotyping someone who is a member of a stigmatized group because of that first hand experience? Or is an expert in prejudice someone who is specifically not a member of a stigmatized g roup, and so that may be assumed to accompany membership in a stigmatized group? In the following text, I present a brief and non exhaustive review of research on attitudes persuasion, and authority influence. Afterwards, I will present methods and results from two experiments designed to examine some potential variables affecting the perceived quality of prejudice and stereotyping research and perceived researcher credibil ity. I will then discuss the findings and implications of this research A ttitudes, Social Influence, and Authority Much of the social psychological research on attitudes and persuasion has focused on identifying the variables that lead to attitude formati on and change. Attitudes towards a target stimulus are not always shaped by features of the stimulus itself; rather, attitudes are often shaped by external factors that have nothing to do with the stimulus (Cacioppo & Petty, 1986). In particular, the elabo ration likelihood model (Cacioppo and Petty 1986) purports that recipients of a persuasive message form their attitudes about that message through one of two possible routes: a central route and a peripheral route. When the central route is activated, peop le pay close attention to the as whether the arguments are logical. On the other hand, when the peripheral route is activated, people focus on features that are irrelev ant to the quality of the message, such as speaker characteristics (Cacioppo & Petty, 1986). Attributes such as the physical attractiveness and the perceived expertise of the speaker, for example, affect

PAGE 17

17 the persuasiveness of a message (e.g., Chaiken, 1979 ; Olson & Cal, 1984 ). Consistent problematic based on personal information about Walker himself, despite the fact that the constitutionality of Proposition 8 was at the heart of the case, and that sexual orientation is irrelevant to determining constitutionality. In addition to examining information processing in attitude formation focusing on the cognition and behavior of the perceiver or target of influence psychologist s have also studied authority influence as an effective means of attitude formation and change, focusing on variables specific to the authority figure or agent of influence rather than the perceiver or target. One reason people may defer to authority figur es is that authorities have the power to punish or reward behavior. Interestingly however, people still defer to someone who they perceive as an authority even though the perceived authority lacks the power to punish or reward. People may defer to authorit ies because doing so can, but will not always, lead to quick and accurate decisions (Kenrick, Neuberg, & Cialdini, 2002). Appeals to authority are prevalent in advertising and have consistently proven to be a powerful tool of persuasion (Jung & Kellaris, 2 006 ; Kenrick, Neuberg, & Cialdini, 2010). A uthority influence is so potent that the mere appearance of authority which can be easily faked by titles, clothes, and trappings is sufficient to pro duce its effects (Kenrick, Neuberg, & Cialdini, 2010; Sagarin, Cialdini, Rice, & Serna, 2002 ). In a study on resistance to persuasion (Sagarin et al., 2002), participants rated magazine ads exp erts). Before rating the ads, participants randomly

PAGE 18

18 assigned to the experimental condition first read a packet of information about the use of legitimate and illegitimate authority in advertising, instructions on how to distinguish between legitimate and i llegitimate authorities, and example ads. Participants in the control condition read a packet of information about the use of color and tone in advertisements, with accompanying example ads. Participants in the control condition rated the ads with illegiti mate authority as less unduly manipulative and more persuasive, relative to participants in the experimental condition. In a study examining the effects of reputation (a proxy for expertise or authority), college students were asked to evaluate poetry (Ar onson, Turner, & Carlsmith, 1963). Afterwards, participants read an essay that described the use of form in poetry and used the same stanzas just evaluated by participants as either good or bad examples. This essay was attributed to either T.S. Elliot or t o a student at a small college. Participants then re evaluated the poetry. For participants who showed a wide stanzas i.e., participants who showed attitude change peop le in the high expertise (T.S. Elliot) condition showed greater attitude change in their second evaluations than did participants in the low expertise (college student) condition. The personal characteristics of someone who is delivering a message are gene rally considered a peripheral feature according to the elaboration likelihood model; Judge Walker himself essentially made this point when he argued that his sexual s ome people view personal characteristics as central and relevant to determining

PAGE 19

19 op inions on what constitutes a legitimate authority figure, and further, that the perceived legitimacy of authority figures can influence the perceived legitimacy of their decisions. But what constitutes a legitimate authority? Jung and Kellaris (2006) poin t out the existence of multiple authority types and acknowledge that there is a range of factors that influence the effectiveness of an au thority appeal. They note that celebrity endorser appeals and true authority appeals (appeals from superiors in academ ia, the workplace, or government) both tend to fall under the broad umbrella term of (Jung & Kellaris, 2006), though some researchers have begun to distinguish between authority types (e.g., Sagarin et al., 2002). According to socia l psychologists such as Kenrick, Neuberg, and Cialdini (2010), an authority is someone who is considered expert and trustworthy (Kenrick, Neuberg, & Cialdini, 2010). Counseling psychologists have also stressed the importance of expertise and trustworthines s Strong (1968) reviewed social psychological research on communicator variables that affect opinion change and recommended application of this research to the field of counseling, likening counselors to agents of persuasion. Strong asserted that, simila r to a communicator attempting to influence an audience towards some predetermined direction, counselors attempt to influence their clients towards achieving the goals of counseling. Counselors may communicate information or opinions that are discrepant wi Reviewing both social psychological and counseling research on social influence (e.g., Aronson, Turner, and Ca

PAGE 20

20 to achieve his or her counseling goals. A client must believe that the counselor knows what he or she is doi ng (Strong, 1968). A counselor can convey expertise by displaying objective indexes (such as diplomas and certifications) and through behavior (through strategically structuring the client and counselor interview). These displays and behaviors can show the client that the counselor is an expert someone who knows what he or she is doing and who is a source of valid assertions (Strong, 1968). Strong also identified trustworthiness as an important variable affecting counseling success (1968). He defined trustw orthiness as a function of reputation for honesty, social role, sincerity and openness, and perceived lack of motivation for personal gain. Strong suggested that trustworthiness may play a stronger role than expertise, weakening the influence of expertise when trustworthiness is low, or compensating for low expertise when trustworthiness is high (Strong, 1968). A counselor can establish trustworthiness concern for the cli The definition of authority as someone who is both expert and trustworthy does judges, police officers, employers, and professors. Hence, people without power to (Kenrick, Neuberg, & Cialdini, 2010). Thus, even though Ju dge Walker had legal power, it is likely that members of the Protect Marriage group saw him as lacking in

PAGE 21

21 felt justified in protesting against his legal power. The im extend beyond perceptions of the legitimacy of judges and their rulings. The public perception of psychology as science can have a direct impact not only on funding for psychological thereby contribute to positive social change. Psychologists who hope to combat prejudice through their research may be undercut by public perception that they are not legitimate autho rities on prejudice for one reason or another. This begs the question of what constitutes an authority in prejudice research, and what characteristics of a Expertise expert in prejudice and stereotyping someone who is a member of a stigmatized group and who has first hand experience with prejudice and stereotyping, because personal experien ce may grant expertise? Or is an expert in prejudice someone who is specifically not a member of a stigmatized group, and so may be perceived as more accompany membership in a stigmatized group? Overview of the Present Research There are two major goals for the proposed studies. The first goal is to examine how participants perceive the quality of research and the credibili ty of the researchers when they are given certain perso nal characteristics about the researcher (characteristics which portray membership or lack thereof in a group that is stereotyped or discriminated against) The second goal is to examine how participants

PAGE 22

22 own personal characteristics, particularly their per ceptions of what constitutes an expert, affect perceptions of research quality and researcher credibility.

PAGE 23

23 CHAPTER 2 STUDY 1 Study 1 served as a pilot study; its goal was to examine whether people evaluate prejudice research differently depending on whet her or not the research was conducted by a member of a stigmatized group. Participants in this study were randomly assigned to read one of 12 versions of an article summarizing researc h by Ayres and Siegelman (1995), which showed gender and racial discrimi nation in price negotiations in car sales. The content of this article was consistent across conditions, but the primary researcher ethnicity, gender, and university affiliation varied across conditions. Method Participants Participants were 244 people r ecruited (Mturk; http://mturk.com ), a website that allows its user to request and compensate work from other users of the site. Participants eligible to complete this study on Mturk were th ose who had a previous approval rating of at least 95% (i.e., the majority of their prior Mturk work had been compensated and deemed acceptable, rather than rejected and deemed unacceptable, by Mturk requesters) and whose location was not India, Pakistan, or the Philippines. Large differences in minimum wage between these countries and the United States, as well as increasing prevalence of Mturk workers from these countries (India in particular; see Ross, Irani, Silberman, Zaldivar, & Tomlinson, 2010) could lead to over sampling, with workers from these countries completing tasks before participants from other countries have the opportunity to do so.

PAGE 24

24 Of the 244 participants who started this study on Mturk and Qualtrics, 143 completed it and 132 of those part icipants passed all quality checks, resulting in a sample of 132 for analysis (54%). Participants received US$0.26 as compensation. Design and Variables The design of this study was a 2 (researcher ethnicity: African American vs. European American) 2 (re searcher gender: female vs. male) 3 (university prestige: high vs. low vs. no information given) between person experiment. The dependent credibility, and opinions re garding whether future funding and research should be devoted to the research topic (Appendix A ). Procedure Participants completed Study 1 over the Internet via a Qualtrics survey (participants were redirected from the Mturk website after starting the stud y on Mturk). After providing consent to participate in this study, participants were presented with a stimulus article (see Figure 2 1 for an example; red, bolded, and underlined text indicates text that was changed per condition). This article summarized a study examining racial and gender discrimination in car price negotiations (Ayres & Siegelman, 1995). This article was identical across all twelve conditions except for the across conditions to convey both researcher gender and ethnicity. I compiled a list of first and last names from baby name websites, U.S. Census data (1990 & 2000), and suggestions from University of Florida undergraduate students. I then pilot tested thi s list to determine which names were most strongly associated with men, women, African Americans, and European Americans. Using the results of this

PAGE 25

25 pretest, I chose the names most strongly associated with African American men (e.g., Denzel Brown), African American women (e.g., Imani Jackson), European American men (e.g., Charles Collins), and European American women (e.g., Kim Smith) for use in Study 1, with three names per condition for the purposes of stimulus sampling (Wells & Windschitl, 1999). Depende nt Measures and Demographics After reading the Ayres and Siegelman (1995) summary article, participants evaluated the quality of the research itself, the credibility and trustworthiness of the d future study ( Appendix A ). Several of the research quality items from this scale were taken from a willingness to overlook methodological flaws when reading about scien tific research How methodologically sound or flawed would you say the design of this study is? These questions were interspersed with distractor items to reduce suspicion (e.g., participants completed a demograp hics questionnaire (Appendix E ). Predictio ns I made two competing predictions regarding the effects of researcher ethnicity and gender. The first prediction was that participants would view prejudice research as more valid or credible if the researcher was seen as a member of the stigmatized group because such a researcher may be perceived as having first hand knowledge of the phenomenon that he or she is studying. This high personal relevance may also foster

PAGE 26

26 greater researcher investment in his or her work. The competing prediction, however, was t hat participants would view research as more valid or credible if the researcher was not seen as a member of the stigmatized group because such a researcher would be perceived as objective, detached, and free of any identity based bias that may accompany m embership in a group that is often discriminated against. Results The final sample size was 132. Sixty four (48.5%) participants identified as female, 65 (49.2%) as male, and 0 as transgendered. Ages ranged from 18 to 54 years (Mode = 22, M = 29.13, SD = 8 .87). Six (4.5%) participants were living in Africa, 12 (9.1%) in Asia, one (0.7%) in the Caribbean, 56 (42.4%) in Europe, 32 (24.2%) in North America, four (3%) in Oceania, six (4.5%) in South America, and 15 (11.4%) participants provided no location info lower numbers corresponding to more liberal and higher numbers corresponding to more conservative (Appendix E ), the mo de answer was 4 ( M = 3.15 SD = 1.43). Of the 22 U.S. citizens and residents, 12 (54.5%) identified as Democratic, none as Green Party, two (9.1%) as Libertarian, and one (4.5%) as Republican. I ran multiple regression analyses to test for effects of resea rcher ethnicity, researcher gender, and university prestige on the dependent variables. I contrast coded the main effects and all possible two way and three way interactions and entered them as predictors into a multiple regression equation (Table 2 1). Th is multiple regression approach is analogous to performing traditional ANOVAs, but with the added efficiency of using focused, single degree of freedom contrasts to test specific effects of interest (e.g., male vs. female).

PAGE 27

27 All dependent variables were app roximately normally distributed. I averaged the Evaluations of Research and Researcher Credibi lity Scale (Appendix A ) to form three ove rall dependent variables for evaluations of research quality, future funding and research, and researcher credibility ( s = .87, 78, and .85, respectively). Descriptive statistics and correlations among the dependent variables appear in Table 2 2 Cell means for all three dependent variables appear in Table 2 3 Regression coefficients appear in tables 2 4 through 2 6. The main effect of university prestige was the only statistically significant effect in all three regression analyses: b = 0.35, t (120) = 2.83, p < .05, r p = .25 d = 0.52, for evaluations of research quality; b = 0.43, t (120)= 3.26 p < .05, r p = .29, d = 0.60 for future funding and research; and b = 0.23, t (118) = 2.15 p < .05, r p = .19, d = 0.40 for researcher credibility. Interestingly, this effect is in a counter intuitive direction. Participants who believed that the primary researcher was employed by a low prestige university perceived the research summarized in the stimulus article as higher quality and more deserving of future research and funding. They also perceived the researcher him or herself as more credible. Discussion The results yielded evidence for an effect of university prestige on evaluations of future funding and research. Nevertheless, I found no significant evidence for effects of res earcher ethnicity or gender. There are several possible reasons for this lack of effects. It may be there was insufficient power to detect effects that actually exist. It may

PAGE 28

28 also be that the manipulations were not meaningful to participants outside of the U.S. and too subtle even for U.S. participants to produce a detectable effect. It may be that the article was too long and detailed, causing participant fatigue and further weakening the manipulations. It may be that the results of the Ayres and Siegelman (1995) study were expected or easy enough to believe that participants were not inclined to question the research or the person who conducted it, regardless of the univer sity prestige than by personal characteristics of a researcher, such as gender and ethnicity. As for the direction of the university affiliation main effect, it is possible that this effect was due to a failure in stimulus sampling and negative perceptions of elitism for the specific high prestige universities. Although pilot testing showed that the universities used here were considered prestigious, that prestige may not have been a good thing in the minds of participants. At least some participants may ha ve viewed these specific universities as being too far research seriously. Alternatively, some participants may view researchers from these universities as biased liberals with an agenda, and therefore not trustw orthy. These interesting possibilities invite further study. As for the lack of significant effects for the other independent variables, I suspect in particular that the manipulations were too weak. Although I pre tested the names with North American unive rsity students, the sample for Study 1 was international, and so was not limited to either university students or North Americans. Additionally, even if the names did name leaves ma y be fleeting, as well as too weak to overcome any initial assumptions

PAGE 29

29 participants had formed upon reading the article title but before receiving any information about the researchers. In Study 2, I addressed the problem of overly subtle manipulations, an international sample, and participant fatigue. I also more closely examined the concept of researcher credibility. In addition to the problems of low power and weak manipulations, it may be that any effects of researcher ethnicity and gender were responded the opposite way, based on personal opinions regarding expertise. Multiple individual difference measures may predict the direction of the effects of researcher ethnicity and gender, but I suspected that the strongest individual difference was perception of what constitutes a good researcher. Thus, I collected more information on what participants prefer in a researcher impartiality and emotional detachment, or first hand personal experience and relevance.

PAGE 30

30 Table 2 1. Contrast Coding for Multiple Regression Predictors (Study 1) African American European American Male Female Male Female High Low Control High Low Control High Low Control High Low Control Ethnicity 1 1 1 1 1 1 1 1 1 1 1 1 Gender 1 1 1 1 1 1 1 1 1 1 1 1 High vs. Low Prestige 1 1 0 1 1 0 1 1 0 1 1 0 Control vs. Else 1 1 2 1 1 2 1 1 2 1 1 2 Interaction1: ethnicity gender 1 1 1 1 1 1 1 1 1 1 1 1 Interaction2: ethnicity high vs low 1 1 0 1 1 0 1 1 0 1 1 0 Interaction3: ethnicity control vs else 1 1 2 1 1 2 1 1 1 1 1 2 Interaction4: gender high vs low 1 1 0 1 1 0 1 1 0 1 1 0 Interaction5 : gender control vs else 1 1 2 1 1 2 1 1 2 1 1 2 3 way Interaction1: ethnicity gender high vs low 1 1 0 1 1 0 1 1 0 1 1 0 3 way Interaction2: ethnicity gender control vs else 1 1 2 1 1 2 1 1 2 1 1 2

PAGE 31

31 Table 2 2 Correlations for Dependent Variables and Descriptive Statistics (Study 1) Evaluations of Research Future Funding & Research Researcher Credibility Correlations Evaluation of Research Funding & Future Research .67** Researcher Credibility .80** .50 ** Descriptives Mean 4.63 4.48 4.70 SD 1.17 1.25 0.98 ** p < .01. p < .05.

PAGE 32

32 Table 2 3 Cell Means (Study 1) African American European American M SD n M SD n Evaluations of Research Male High prestige university 4.47 1.27 16 4.19 0. 90 13 Low prestige university 5.30 0.77 17 4.67 1.30 7 Control 4.66 1.70 10 4.48 1.11 9 Female High prestige university 4.03 1.19 12 4.24 1.04 12 Low prestige university 4.48 1.35 8 5.26 1.10 9 Control 5.08 0.83 9 4.80 1.13 10 Future Funding & Research Male High prestige university 4.58 1.38 16 4.05 1.12 13 Low prestige university 4.82 0.95 17 4.95 0.89 7 Control 4.77 1.58 10 4.41 1.05 9 Female High prestige university 3.75 1.35 12 3.78 1.43 12 Low prestige university 4.96 1.20 8 4 .89 1.31 9 Control 4.93 1.02 9 4.37 1.25 10 Researcher Credibility Male High prestige university 4.68 0.94 16 4.35 0.85 13 Low prestige university 4.88 0.86 17 4.66 1.12 7 Control 4.63 1.65 10 4.83 0.89 9 Female High prestige university 4.10 1. 02 12 4.63 0.91 12 Low prestige university 4.73 0.78 8 5.30 0.58 9 Control 4.98 0.57 9 4.87 1.14 10

PAGE 33

33 Table 2 4 Regression Evaluations of Research (Study 1) b SE t p r p Main Effects Intercept 4.64 .10 45.12 .00 Ethnicity .03 .10 .31 .7 6 .03 Gender .01 .10 .09 .93 .01 High vs. Low Prestige .35 .12 2.83 .01 .25 Control vs. Else (Prestige) .06 .07 .79 .43 .07 2 way Interactions Interaction1: ethnicity gender .15 .10 1.46 .15 .13 Interaction2: ethnicity high vs low 03 .12 .21 .83 .02 Interaction3: ethnicity control vs else .04 .07 .56 .58 .05 Interaction4: gender high vs low .02 .12 .17 .87 .02 Interaction5: gender control vs else .09 .07 1.19 .24 .11 3 way Interactions 3 way Interaction1: et hnicity gender high vs low .12 .23 .94 .35 .09 3 way Interaction2: ethnicity gender control vs else .09 .07 1.19 .24 .11

PAGE 34

34 Table 2 5 Regression Funding & Future Research (Study 1) b SE t p r p Main Effects Intercept 4.52 .11 40.82 .0 0 Ethnicity .11 .11 1.03 .31 .09 Gender .08 .11 .69 .49 .06 High vs. Low Prestige .43 .13 3.26 .00 .29 Control vs. Else (Prestige) .05 .08 .05 .60 .06 2 way Interactions Interaction1: ethnicity gender .01 .11 .12 .90 .01 Intera ction2: ethnicity high vs low .07 .133 .53 .60 .05 Interaction3: ethnicity control vs else .06 .08 .73 .47 .07 Interaction4: gender high vs low .15 .13 1.11 .27 .10 Interaction5: gender control vs else .05 .08 .66 .51 .06 3 way Interac tions 3 way Interaction1: ethnicity gender high vs low .10 .13 .71 .48 .07 3 way Interaction2: ethnicity gender control vs else .03 .08 .40 .69 .04

PAGE 35

35 Table 2 6 Regression Researcher Credibility (Study 1) b SE t p r p Main Effects Intercept 4.72 .09 53.76 .00 Ethnicity .05 .09 .59 .56 .05 Gender .05 .09 .53 .60 .05 High vs. Low Prestige .23 .11 2.15 .03 .19 Control vs. Else (Prestige) .06 .06 .86 .39 .08 2 way Interactions Interaction1: ethnicity gender .11 .09 1.27 .21 .12 Interaction2: ethnicity high vs low .02 .11 .18 .86 .01 Interaction3: ethnicity control vs else .02 .06 .24 .81 .02 Interaction4: gender high vs low .10 .11 .95 .35 .09 Interaction5: gender control vs else .02 .06 .39 .70 .04 3 way Interactions 3 way Interaction1: ethnicity gender high vs low .01 .11 .08 .94 .01 3 way Interaction2: ethnicity gender control vs else .10 .06 1.50 .14 .14

PAGE 36

36 Figure 2 1. Summary Article of Ayres, I., & Siege l man, P. (1995). Race and gender discrimination in bargaining for a new car. The American Ec onomic Review, 85 (3), 304 321

PAGE 37

37 CHAPTER 3 STUDY 2 Study 2 was conceptually similar to Study 1 but with a key addition; it delved further into examining individual differences in opinions of expertise. In addition, Study 2 was limited to U.S. participants. Different countries may hold different stereotypes and have different minority and stigmatized groups, so the stimuli and questions in Study 1 may not have been a sample to U.S. participants increased the likelihood that the stimuli and questions were meaningful to participants. I also altered the stimuli in Study 2 to make the manipulation stronger Instead of reading a summary of the Ayres and Siegelman (1995) car negotiation study (see Study 1), participants read a summary of a gender differences meta analysis (Petersen & Hyde, 2010). The results of this meta analysis supported the gender similari ties hypothesis, with the majority of gender differences yielding small effect sizes. I suspected that while people in the U.S. are ready to believe that women and racial minorities have historically been discriminated against regarding mechanics or sales, many people probably think of men and women in terms of their differences rather than similarities. Participants may therefore have been more likely to question the research quality and researcher credibility, or at least vary in the extent to which they did so. To further strengthen the manipulation, participants saw a photo of the supposed researcher instead of having to infer race, gender, or other physical characteristics from author names. Participants viewed an ostensible screenshot from a psycholog y

PAGE 38

38 accompanying a few paragraphs summarizing the gender differences meta analysis by Petersen and Hyde (2010). Participants then evaluated the quality of the research and Finally, unlike Study 1, Study 2 included questions to examine what participants believe constitutes a good researcher, and in particular, whether participants believe that personal experience or an emotionally detached perspective is more desirable for a research psychologist. It may be that different opinions about personal experiences versus an emotionally detached perspect ive may moderate effects of researcher characteristics on perceived quality of research or perceived researcher credibility. Method Participants Participants were 360 Mturk workers. Participants eligible to complete this study on Mturk were people who had a previous approval rating of at least 95% and whose location was in the United States. Of the 360 participants who began this study, 268 participants (74%) completed it, passed all quality checks, and for those not in the control condition passed the man received US$1.01 as compensation. Design and Variables The independent variable was researcher gender (female, male, or control). The luations of research quality, perceptions of researcher credibility, and opinions regarding whether funding and future research should be devoted to the research topic (App endix A ).

PAGE 39

39 Participants were randomly assigned to view one of three website screensho ts: one that showed a picture of a man, a woman, or standard symbols for male and female. I used optimal design (McClelland, 1997) and stimulus sampling (Wells & Windschitl, 1999) techniques in creating stimulus materials to manipulate the independent vari able and in assigning participants to conditions. I used four female pictures and four male pictures (Figure 3 2 probability of being assigned to the female researcher, male researcher, and control condition was 0.4, 0.4, and 0.2, resp ectively. Five of the eight researcher photos were volunteered by friends, while three were purchased from http://istockphoto.com Because Mturk participants may have seen many stock photos during the course of their Mturk work and general Web browsing, photos from http://istockphoto.com were selected to reduce the likelihood that Mturk participants would have recognized them from another website. Two of the three stock photos ha d never been purchased, while the other had been purchased only once. All three photos had less than 150 views when purchased. All photos were cropped to show only head and shoulders, and were set against a blue background made to resemble those used in pr ofessional photo studios. Procedure Participants completed Study 2 over the Internet via a Qualtrics survey (participants were redirected from the Mturk website after starting the study on Mturk). After providing consent, participants saw what they believe d to be a screenshot from a Petersen and Hyde (2010) gender differences meta analysis, accompanied by a picture of the ostensible principal investigator (or in the case of the control condition, a picture

PAGE 40

40 of gender symbols; Figure 3 2 ). The screenshot was identical across all conditions except for the picture, which featured either a man, a woman, or a picture of male and female symbols. To increase the likelihood that par ticipants would actually read the research summary on the screenshot as instructed, I imposed a 30 who took the time to read the research summa ry, but would not be available for participants who were just trying to click through the study as fast as possible. After the 30 participants were free to proceed to the dependent m easures, individual difference measures, and demographic questionnaire. Finally, participants were debriefed and given a unique completion code to enter on Mturk for compensation. The debriefing included the correct citation for the Petersen and Hyde (2010 ) meta analysis. Four quality check items were interspersed throughout the questionnaires and questionnaire, all items appeared in random order; the scales and ques tionnaires themselves, however, appeared in the same order. Dependent measures and manipulation check After reading the research summary on the screenshot, participants completed the primary dependent measures of interest, evaluating the quality of the res earch itself, worthiness of funding and future research ( Appendix A ). The items on this scale were

PAGE 41

41 participant suspicion. Next, participants answered the of the psychologist who conducted the gender differe nces research you just read two attention checks (Appendix B ) Individual difference measures After completing the dependent measures and manipulation and attention checks, participants completed individual difference measures. First, partici pants completed questions assessing personal opinions regarding expertise (Appendix C ). An researcher should have a lot of personal experience in his or her chosen field of work. indicate a preference for experts or scientists with firsthand experience and personal relevance in their topic of study, while lower scores indicate a preference for experts or scientists with more emotional detachment from their topic of study. These qu estions formed the primary covariate of interest in the analyses that follow. Participants next completed the Ambivalent Sexism Inventory ( Glick & Fiske, 1997; Appendix D ). The Ambivalent Sexism Inventory is composed of two subscales, benevolent sexism an d hostile sexism, with eleven items for each subscale. According to Glick and Fiske (1997), benevolent and hostile sexism are not mutually exclusive. Hostile sexists tend to have negative feelings toward and hold negative stereotypes about women who reject traditional female roles. Benevolent sexists tend to have positive feelings toward and hold positive stereotypes about women who embrace traditional female roles, such as home makers or mothers. Benevolent sexists are not necessarily hostile to women who reject traditional female roles, unless they are also

PAGE 42

42 Many women have a quality of purity that few men possess. Women are too easily offended. sexism. Finally, participants provided dem ographic information (Appendix E ) before being debriefed and given a unique completion code to enter on Mturk for compensation Predictions Predictions for Ind ependent Variables In Study 1, I made two competing predictions regarding the effect of researcher personal characteristics. Study 1 yielded no significant effects of researcher race or gender, but because these lack of effects may have resulted from probl ems with Study 1, I retained these two competing predictions for Study 2. Participants may view the gender differences meta analysis summary as more valid if they believe that the principal investigator was a woman because they believe that being a woman m ay foster greater investment in the work and allows for first hand knowledge of gender difference stereotypes. Conversely, participants may view the gender differences meta analysis summary as more valid if they believe the principal investigator to be a m an because they believe his lack of membership in a group that has historically been a target of sexism would render him objective, detached, and free of any identity based bias that may accompany membership in a group that has historically been a frequent target of sexism. Predictions for Individual Difference Measures

PAGE 43

43 science and scientists, anoth er major purpose of this study was to examine the effect of individual differences in opinions regarding expertise. I predicted that individual differences in opinions rega rding expertise (Appendix C ) would moderate the effect of researcher gender in this study. Specifically, participants who value personal relevance and firsthand experience in a scientist should view the research and researcher as especially credible if the researcher is a woman (for the same reasons listed above). On the other hand, part icipants who value detachment in a scientist should view the research and researcher as especially credible when the researcher is a man. Glick and Fiske (1996) provided evidence for two related but distinct dimensions of sexism in their Ambivalent Sexis m Inventory (ASI): hostile sexism and benevolent sexism. Hence, I made separate predictions for the hostile and benevolent subscales of the ASI. I predicted that participants who were high in the hostile sexism subscale of the Ambivalent Sexism Inventory w ould view the research and researcher especially of women who reject traditional female roles. Benevolent sexists have positive perceptions of women who embrace traditio nal female roles, but may or may not have negative perceptions of women who reject traditional roles. To the extent that benevolent sexists may also be hostile sexists, it is possible that benevolent sexist participants would view the research and research er especially negatively when the researcher was thought to be a woman (i.e., benevolent sexists may respond like hostile sexists). On the other hand, benevolent sexists may have positive feelings toward and stereotypes about women in general; therefore, i t is possible that benevolent sexists would view the research and researcher as especially credible if they

PAGE 44

44 believed the researcher to be a woman. Therefore, I made no specific predictions about the direction of effect for benevolent sexism. Note, however, that a person can be high in both hostile and benevolent sexism (meaning he or she is high in ambivalent sexism), so it is possible that effects of the hostile and benevolent subscales would be obscured by ambivalent sexism. I therefore made no specific p redictions about the total score (i.e., ambivalent sexism) on the Ambivalent Sexism Inventory. Results Although 360 participants were recruited and paid on Mturk, the final sample size for analysis was 268 (74%). Participants who did not provide unique, m atching completion codes on Mturk and Qualtrics were excluded from the analysis to ensure that all data analyzed came from participants who had been compensated. Participants who did not correctly answer all four quality check items were excluded from data analys e s (Appendix B ). Fin ally, participants who did not correctly answer the allo wed on the manipulation check item, but to allow for the possibility that participants would try to guess a researcher gender even if they were in the control condition (thinking they may have accidentally skipped this information while reading the researc h summary), control condition participants who gave any answer to this question were included in the data analyses. I designed Study 2 so that ea probability of being assigned to the female researcher, male researcher, and control conditio ns were 0.4, 0.4, and 0.2, respectively; with no participation exclusion this would result in cell sizes of 144 for the experimental conditions (both 40% of the total sample size) and 72 for the control

PAGE 45

45 condition (20% of the total sample size). The actual cell sizes were 93 for the female researcher condition (34.7%), 108 for the male researcher condition (40.3%), and 67 for the control condition (25%). Although this deviation is not ideal, the probability of random assignment producing exactly the desired cell sizes in a given study is low due to error variance. The exclusion of participants also contributed to this deviation because of the more stringent manipulation check requirements for the experimental conditions. I made this manipulation check restric tion before analyzing the data. Of the 268 participants, 144 (53.7%) identified as female, 119 (44.4%) as male, three (1%) as transgendered, and two (0.7%) did not provide gender information. 6, M = 34.46, SD = liberal and higher numbers corresponding to more conservative (A ppendix E ), the mode answer was 2, ( M = 3.22 and SD = 1.63). One hundred thirty five (50.4%) participants identified as Democratic, 10 (3.7%) as Green Party, 20 (7.5%) as Libertarian, 41 (Appendix E ). Data Preparation and Preliminary Analyses Researcher Credibility Scale (Appendix A ) to form three overall dependent variables for evaluations of research quality, funding and future research, and researcher credibility ( s = .87, .80, and .87, respectively). Visual inspection of frequency histograms for the overall dependent variables revealed that they were approximately normally distributed.

PAGE 46

46 Items from the What Constitutes an Expert scale ( Appendix C ) yielded a Cronbach of .49. I conducted a confirmatory factor analysis in Amos Graphics factor. Figure 3 3 shows the proposed factor structure and item loadings. Squared multiple correlat ions appear in Table 3 1 None of the items loaded significantly on the expertise factor, though five of the loadings were greater than .40. Three of the eight items on this scale, Items 1, 2, and 7, had especially weak loadings (.03, .13, and .11, respec tively). The fit of this model was poor, 2 (20) = 101.15, p < .05 ( T able 3 2 ). I conducted a second confirmatory factor analysis, dropping items 1, 2, and 7. Figure 3 4 shows the new proposed factor structure and item loadings. Squared multipl e correlations appear in Table 3 1 All items now load ed significantly on the factor and fit indices met goodness of fit criteria ( T able 3 2 ). Because the five item expertise factor model showed acceptable fit, those five items in the What Constitutes an Expert scale were combined to form an overall questions of what constitutes an expert covariate ( WCE ). of .65 and a mean inter item correlation (MIC) of .28. While the alpha for the WCE scale is lower than for the other scales, the WCE scale had the fewest items. Because coefficient alpha is a function of both MIC and number of items (holding MIC constant alpha increases as items increase; Schmitt, 1996), a five item scale with an alpha of .65 and a moderate MIC of .28 is adequate. Indeed, a 10 item scale with the same MIC of .28 would yield an alpha of .80. Visual inspection of the frequency histogram fo r the WCE covariate revealed that it was approximately normally distributed.

PAGE 47

47 sexism as related but separate factors and because of my different predictions for the hostile and benev olent subscales of the Ambivalent Sexism Inventory (ASI), I analyzed the hostile and benevolent items from the ASI separately. Items from the benevolent subscale of the ASI ( Appendix D and were combined to form an overall ben evolent sexism covariate (ASI benevolent). Visual inspection of the frequency histogram for ASI benevolent revealed that it was approximately normally distributed. The 11 i tems from the hostile subscale of the Ambivalent Sexism In v entory yielded a Cronbac of .91 and were combined to form an overall hostile sexism covariate (ASI hostile). Visual inspection of the frequency histogram for ASI hostile revealed that it was approximately normally distributed. Descriptive statistics and correlations among th e dependent variables and covariates appear in Table 3 4 Cell means for the dependent variables appear in Table 3 5 Main Analyses I collapsed across the four female researcher photos for the female researcher condition and collapsed across the four male researcher photos for the male researcher condition. I then used multiple regression analysis to test for effects of researcher gender. A potential problem with this collapsing across photos is that it ignores stimuli nesting in the study design, with part icipants nested within photos, which are in turn nested within broader categories of male and female faces. If individual photos account for variance, then collapsing across photos may result in p values that are spuriously small for any effects involving researcher gender; in other words, collapsing across

PAGE 48

48 photos and analyzing with multiple regression may be an overly liberal analysis method. Analyzing these data with multi level modeling would account for the nesting problem. However, a multi level modeli ng approach would present other problems. For example, while the male and female photos were stimulus sampled so that there were multiple photos nested within each category, the control condition was not stimulus sampled, and so there was only one control image nested with the control category. Additionally, the level 2 sample size in a multi level model would be nine (eight faces, plus one control condition image), and p values based on only nine observations may be too conservative. In any case, I conduc ted a regression analysis for each dependent variable, using photo as the unit of analysis (not collapsing across any photos). There were no gender effects for research quality, funding and future research, or researcher credibility dependent variables; ho wever, because only eight photos were used, these tests were woefully underpowered. Thus, we moved forward with testing participants as the unit of analysis. Using participants as the unit of analysis, I ran multiple regression analyses to test for effect s of researcher gender. The comparisons of (a) male versus female researcher and (b) control versus the average of male and female were contrast coded and entered as predictors into a multiple regression equation (Table 3 3 ). Regression coefficients for th ese effects appear in Tables 3 6 through 3 17 For the purposes of this dissertation, the male versus female researcher contrast is the most important, but the control versus else contrast is necessary for the primary contrast to be interpretable. Further, this contrast accounts for error variance and allows the regression results to be

PAGE 49

49 analogous to results from a traditional AN(C)OVA. This multiple regression approach has the added efficiency of using focused, single degree of freedom tests to assess speci fic effects of interest (e.g., male vs. female). In the first step of each regression analysis, I entered the gender comparisons as predictors into the regression equation. In the second step, I mean centered the covariate (either the What Constitutes an Expert scale, the benevolent subscale of the Ambivalent Sexism Inventory, or the hostile subscale of the Ambivalent Sexism Inventory) and entered it into the regression equation. In the third step, I entered the gender covariate interactions (male vs. fe male covariate, control vs. else covariate) into the regression equation. Dependent variable: evaluations of research quality Main effect and WCE covariate There were no significant main effects of gender on the evaluations of research quality depende nt variable. When I entered the WCE covariate into the model in Step 2, the effect of the covariate was significant, b = 0 .33, t (264 ) = 4.86 p < .05, r p = .29. As WCE scores increased, so did perceived quality of the research presented in the gender differ ences meta analysis summary. Thus, participants who valued personal relevance more than emotional detachment rated the research as higher quality. When I entered the WCE gender interactions into the regression model in Step 3, the covariate effect was ag ain significant ( b = 0.30, t (262 ) = 4.25 p < .05, r p = .25) along with a female vs. male WCE interaction ( b = 0.19, t (262 ) = 2.41 p < .05, r p = .15, d = 0.30; Figure 3 5 ). A simple effects test revealed that for participants who scored relatively low ( by convention, 1 SD below the mean ; see Aiken & West, 1991) on the WCE, those in the

PAGE 50

50 male researcher condition rated the research as lower in quality, b = 0.24, t (262 ) = 2.11 p < .05, r p = .13, d = 0, 26 For participants who scored relatively high on t he WCE, however, ratings of male and female researchers did not differ significantly. In other words, people who valued emotional detachment in an expert rated research ostensibly conducted by a man as lower in quality, while participants who valued firsth and experience in an expert did not differ according to researcher gender in their research quality ratings (see Figure 3 5 ). A simple effects test for the WCE covariate showed that, for participants in the male researcher condition, ratings of the resea rch quality increased along with participant score on the WCE covariate, b = 0.45, t (262 ) = 4.73 p < .05, r p = .15, d = 0.58. This covariate simple effect was not significant for the female researcher condition. In other words, the more participants valued firsthand experience in a researcher, the greater they perceived quality of the research, but only when they believed the research had been conducted by a man. Benevolent sexism covariate. In a separate regression analysis (still using evaluations of res earch quality as the dependent variable), I entered ASI benevolent into the model in Step 2. There were no significant effects, but there was a marginally significant main effect for the covariate with a small effect size, b = 0.14, t (264 ) = 1.81 p = .07 r p = .11. As benevolent sexism increased, perceptions of research quality decreased (though again, this effect was not statistically significant). When I entered the ASI benevolent gender interactions into the regression model in Step 3, results showe d no significant effects. The marginally significant covariate effect from Step 2 decreased in significance and effect size, b = 0.12, t (262 ) = 1.61 p = .11 r p = .10.

PAGE 51

51 Hostile sexism covariate. In a separate regression analysis, I entered ASI hostile i nto the model in Step 2. The hostile sexism covariate effect was significant, b = 0.25, t (264 ) = 3.23 p < .05 r p = .20. As hostile sexism increased, perceptions of research quality decreased. When I entered the ASI hostile gender into the regression model in Step 3, the covariate effect was again significant ( b = 0.22, t (262 ) = 2.88 p < .05 r p = .18).The ASI hostile control vs. else interaction was also significant, b = 0.13, t (262 ) = 2.41 p < .05 r p = .15, d = 0.30 (Figure 3 6 ). A simple effe cts test revealed that, f or participants low on hostile sexism, ratings of research quality were higher in experimental conditions than in the control condition b = 0.18, t (262 ) = 2.57 p < .05 r p = .16, d = 0.32 For participants high on hostile sexi sm, however, ratings of research quality did not differ between experimental and control conditions. In other words, for participants low on hostile sexism, the gender differences research summary was perceived as higher quality when it was accompanied by a picture of the researcher (regardless of researcher gender), compared to when the research summary was accompanied by a picture of the standard gender symbols (when no researcher gender information was given). Participants high in hostile sexism did not show this difference (Figure 3 6 ). A simple effects test for the ASI hostile covariate revealed that, for experimental conditions only, ratings of research quality decreased as hostile sexism increased, b = 0.35, t (262 ) = 4.02 p < .05 r p = .24, d = 0 .50 Participant gender covariate. Although I made no predictions about participant gender, I tested the participant gender main effects as well as the participant gender researcher gender interactions for all three dependent variables. (Participants who

PAGE 52

52 identified as transgendered or who did not provide gender information were excluded from the participant gender regression analyses.) However, when I entered participant gender into Step 2 of a separate regression analysis (still using evaluations of rese arch quality as the dependent variable), the participant gender covariate main effect was not significant. When I entered the participant gender researcher gender interactions into the model in Step 3, there were no significant effects. Dependent variab le: funding and future research Main effect and WCE covariate There were no significant main effects of gender on the funding and future research dependent variable. When I entered the WCE covariate into the model in Step 2, the covariate effect was signi ficant, b = 0.33, t (264 ) = 4.02 p < .05, r p = perceived worthiness of funding and future study. Participants who valued personal relevance more than emotional detachment rated the research as more wor thy of funding and future study, relative to participants who valued emotional detachment. When I entered the WCE gender interactions into the regression model in Step 3, the covariate effect was again significant ( b = 0.30, t (262 ) = 3.56 p < .05, r p = .22) along with a male vs. female WCE interaction ( b = 0.26, t (262 ) = 2.64 p < .05, r p = .16, d = 0.33; Figure 3 7 ). A simple effects test showed that, for participants who scored low on the WCE, ratings for the funding and future research dependent v ariable did not differ between the male and female researcher conditions. For participants who scored high on the WCE, however, those in the male researcher condition rated the research as more worthy of funding and future study, b = 0.35, t (262 ) = 2.77 p < .05, r p = .17, d = 0.34. In other

PAGE 53

53 words, participants who valued emotional detachment in an expert rated the research as equally worthy of funding and future study regardless of whether they thought the principal investigator was a man or woman. However participants who value firsthand experience in an expert rated the research as more worthy of funding and future study if they thought the principal investigator was a man (see Figure 3 7 ). A simple effects test for the WCE covariate revealed that, for participants in the b = 0.47, t (262 ) = 4.04 p < .05, r p = .24, d = 0.50. This simple effect for th e WCE covariate was not significant in the female researcher condition. In other words, the more participants valued firsthand experience, the more they perceived the research as worthy of funding and future study, but only for participants who believed th at the research was conducted by a man. Benevolent sexism covariate. In a separate regression analysis (still using the funding and future research variable as the dependent variable), I entered ASI benevolent into the model in Step 2. None of the effects were significant, but there was a marginally significant covariate main effect with a small effect size, b = 0.18, t (264 ) = 1.99 p = .05, r p = .12. As benevolent sexism increased, participants rated the research as less worthy of funding and future res earch. When I entered the ASI benevolent gender interactions into the regression model in Step 3, there were no significant effects. Three effects, however, were marginally significant: the ASI benevolent covariate main effect, the ASI benevolent male vs. female interaction, and the ASI benevolent control vs. else interaction. The

PAGE 54

54 marginally significant covariate effect from step two decreased in significance and effect size, b = 0.15, t (262 ) = 1.66 p = .10 r p = .10. The ASI benevolent male v s. female interaction was marginally significant, b = 0.18, t (262 ) = 1.68 p = .09 r p = .10, d = 0.21 ( Figure 3 8 ). A marginally significant simple effects test revealed that for participants low in benevolent sexism, research ostensibly conducted by a man was perceived as more worthy of funding and future study, relative to research ostensibly conducted by a woman, b = 0.25, t (262 ) = 1.95 p = .05 r p = .10, d = 0.24. This marginal simple effect of gender was not significant for participants high in benevolent sexism. A simple effects test for the ASI benevolent covariate in the male researcher condition yielded a significant effect, b = 0.46, t (262 ) = 3.09 p < .05 r p = .11, d = s worthiness for funding and future study decreased, but this was only true for participants in the male researcher condition. The simple effect of the ASI benevolent covariate was not significant in the female researcher condition. The ASI benevolent control vs. else interaction was marginally significant, b = 0.13, t (262 ) = 1.86 p = .06 r p = .11, d = 0.23 ( Figure 3 9 ). A simple effects test showed that for participants low in benevolent sexism, ratings for the funding and fu ture study were higher in experimental conditions than in the control condition b = 0.21, t (262 ) = 2.48 p < .05 r p = .15, d = 0.31 This simple effect was not significant for participants high in benevolent sexism. Hostile sexism covariate. In a sep arate regression analysis (still using the funding and future research variable as the dependent variable), I entered ASI hostile

PAGE 55

55 into the model in Step 2; this covariate main effect was significant, b = 0.27, t (264 ) = 2.93 p < .05, r p = .18. As hostil e sexism increased, participants rated the research as less worthy of funding and future research. When I entered the ASI hostile gender interactions into the regression model in Step 3, the covariate effect was again significant, b = 0.21, t (262 ) = 2 .36 p < .05, r p = 14 The ASI hostile control vs. else interaction was also significant, b = 0.19, t (262)= 2.84, p < .05, r p = .17, d = 0.35 (Figure 3 10 ) A simple effects test for this significant interaction revealed that, f or participants low on h ostile sexism, scores for the funding and future research dependent variable were higher in experimental conditions than in the control condition b = 0.26, t (262 ) = 3.11 p < .05, r p = 19, d = 0.38 This simple effect was not significant for particip ants high on hostile sexism. In other words, for participants low on hostile sexism, the gender differences research summary was perceived as more worthy of funding and future study when it was accompanied by a picture of the researcher (regardless of rese archer gender), compared to when the research summary was accompanied by a picture of the standard gender symbols (when no researcher gender information was given). Participants high in hostile sexism did not show this difference (Figure 3 10 ). Participant gender covariate. In a separate regression analysis, I entered participant gender into the model in Step 2. The covariate main effect for participant gender was not significant. When I entered the participant gender researcher gender interactions into t he model in Step 3, there were no significant effects.

PAGE 56

56 Dependent variable: researcher credibility Main effect and WCE covariate There were no significant main effects of gender on the researcher credibility dependent variable. When I entered the WCE covar iate into the model in Step 2, the effect of the covariate emerged as significant, b = 0.17, t (264 ) = 2.66 p < .05, r p = .16. As scores on WCE increased, so did the perceived credibility of the researcher. Participants who valued personal relevance more t han emotional detachment rated the researcher (regardless of researcher gender) as more credible, relative to participants who valued emotional detachment. When I entered the WCE gender interactions into the regression model in Step 3, the covariate eff ect was again significant ( b = 0.14, t (262 ) = 2.03 p < .05, r p = .13). Unlike the results for the research quality and the future funding and research dependent variables, the male vs. female WCE interaction did not emerge as significant for the researc her credibility dependent variable. Benevolent sexism covariate In a separate regression analysis (still using researcher credibility as the dependent variable), I entered ASI benevolent into the model in Step 2. None of the effects were significant. Whe n I entered the ASI benevolent gender interactions into the regression model in Step 3, none of the effects were significant, although the ASI benevolent male vs. female interaction was marginally significant with a small effect size, b = 0.15, t (262 ) = 1.81 p = .07 r p = .11, d = 0.22 (Figure 3 11 ). The simple effect of researcher gender was not significant for participants low in benevolent sexism, nor was it significant for participants high in benevolent sexism. However, a simple effects test f or the ASI benevolent covariate in the male researcher condition was significant, b = 0.25,

PAGE 57

57 t (262 ) = 2.20 p < .05 r p = .14, d = 0.27. As benevolent sexism increased, perceptions of male researcher credibility decreased. This covariate simple effect w as not significant for female researchers. When participants thought the researcher was a man, t sexism scores did not predict perceptions of researcher credibility (Figure 3 11 ). Hostile sexism covariate. In a separate regression analysis (still using researcher credibility as the dependent variable), I entered ASI hostile into the model in Step 2 as a covariate; this effect was significant, b = 0.22, t (264 ) = 3.25 p < .05, r p = .20. As hostile sexism increased, participants rated the researcher as less credible. When the ASI hostile gender interactions were entered into the regression model in Step 3, the covariate effect was again significant, b = 0.21, t (262 ) = 3.05 p < .05, r p = .19). Participant gender covariate. In a separate regression analysis, I entered participant gender into the model in Step 2. The covariate main effect for par ticipant gender was not significant. When I entered the participant gender researcher gender interactions into the model in Step 3, there were no significant effects. Discussion Researcher gender effects The results showed no significant main effects of researcher gender on any of the three dependent variables of interest. Researcher gender alone did not significantly funding and further study, or of the principal invest

PAGE 58

58 specify a direction for the effects, I did predict that gender would have a main effect on all three dependent variables. Results do not support these predictions. Gender did, however, interact with other covariates t o have significant effects on evaluations of research quality and research worthiness of funding and future study. Expertise covariate effects The expertise covariate (WCE) had significant main effects on evaluations of research quality, funding and futur e research, and researcher credibility. The higher a other words, participants who expressed a preference for firsthand experience in a scientist (vs. participants who ex pressed a preference for detachment) had more positive perceptions of both the research and researcher. There were significant gender expertise covariate interactions for the evaluations of research dependent variable and for the funding and future rese arch dependent variable (but not for researcher credibility). For participants in the male researcher condition, higher scores on WCE predicted higher ratings for evaluations of research and for funding and future research. In other words, participants who expressed a preference for firsthand experience in a scientist (vs. participants who expressed a preference for detachment) had more positive perceptions of the research, but only if they thought the researcher was a man. Although I did predict that the W CE covariate would moderate the effect of gender, I predicted that both the research and high and they thought that the researcher was a woman. Results did not support thes e predictions; when WCE was high, the simple effects comparing male and female

PAGE 59

59 researchers were either not significant (for evaluations of research and researcher credibility) or were in the opposite direction as predicted (for funding and future research) Instead, it was in the male researcher condition that increases in positive perceptions were related to WCE increases. Again, these results did not support my predictions. It may be that participants view the gender similarities research as more self rel evant to male researchers than to female researchers; future research should include measures of perceived relevance. Benevolent sexism covariate effects The benevolent sexism covariate ASI Benevolent had significant (or marginally significant) main effect s on the evaluations of research quality dependent variable, and on the funding and future research dependent variable. As benevolent sexism increased, perceptions of the research and the researcher became more negative. In other words, the more positively participants viewed women who embrace traditional female roles, the less positively they perceived the research (not taking into account researcher gender). There were marginally significant researcher gender benevolent sexism covariate interactions fo r the funding and future research and the researcher credibility dependent variables. As benevolent sexism increased, perceptions of the research and researcher became more negative, but only for the male researcher condition. In other words, the more posi tively participants perceived women who embrace traditional female roles, the less positively they perceived the research and researcher, but only if they thought the research had been conducted by a male. I did predict that benevolent sexism would moderat e the effects of gender, but I made competing predictions about

PAGE 60

60 the direction of the effect; I predicted that women and their research would either be perceived especially positively when benevolent sexism was high, or that they would be perceived especial ly negatively when benevolent sexism was high (because benevolent sexism could correspond to hostile sexism). The results did not support either of these predictions; when benevolent sexism was high, the simple effects comparing male and female researchers were not significant for either the funding and future study dependent variable or for the researcher credibility dependent variable. Moreover, means for all three dependent variables in the female condition tended to remain near the midpoint. Conversely results showed that the research and researcher were perceived low and when they believed the researcher to be male The results do show moderation by benevolent sexism, but results were not consis tent with either of the specific possible effects I predicted. Hostile sexism covariate effects The hostile sexism covariate (ASI Hostile) had significant main effects on evaluations of research quality, funding and future research, and researcher credibi lity. As hostile sexism increased, perceptions of both the research and researcher became more negative. In other words, the more negatively participants perceived women who rejected traditional female roles, the more negatively they perceived both the res earch and researcher. There were no significant gender hostile sexism interactions for the evaluations of research, funding and future research, or researcher credibility dependent variables. The effect of researcher gender did not vary depending on lev el of hostile sexism.

PAGE 61

61 Although the hostile sexism covariate did have significant main effects, I predicted that hostile sexism would moderate the effects of researcher gender; these results did not support my predictions. Participant gender covariate effe cts There was no evidence for participant gender main effects on any of the three dependent variables. Further, there was no evidence for participant gender researcher gender interactions. In sum, although researcher gender showed no evidence for the ma in effects I predicted, it did interact with covariates to produce significant interaction effects. And although I did predict covariate moderation of the gender effect, the patterns of the particular independent variable covariate interactions did not m atch my predictions. personal opinions on expertise affect perceptions of prejudice and stereotypes research. There is also some limited evidence that gender interacts with these variables to affect perceptions of prejudice and stereotypes research; however, out of the 12 interactions involving researcher gender, only two were significant and two were marginally significant. As the number of tests increases, so does the probability of Type I errors (false rejection of a null hypothesis); thus, these results for researcher gender should be interpreted with caution.

PAGE 62

62 Table 3 1. Squared Multiple Correlations Model 1: Full WCE Model 2: Reduced WCE WCE item 1 .00 WCE item 2 [R] .02 WCE item 3 .21 .21 WCE item 4 [R] .48 .50 WCE item 5 .18 .19 WCE item 6 [R] .34 .33 WCE item 7 [R] .01 WCE item 8 .30 .30 Table 3 2. Goodness of Fit Model Chi square CFI TLI RMSEA Full WCE 2 (20)=101.15, p <.05 .67 .40 .12 Reduced WCE 2 (5)=6.25, p =.28** .99** .98** .03** *meets historic al criteria for goodness of fit. **meets current criteria for goodness of fit Table 3 3 Contrast Coding for Multiple Regression Predictors (Stud y 2) Female Male Control Female vs. Male 1 1 0 Control vs. Else 1 1 2

PAGE 63

63 Table 3 4 Correlations for Dependent Variables and Covariates; Descriptive Statistics (Study 2) Evaluations of Research Future Funding & Research Researcher Credibility WCE A SI Benevolent ASI Hostile Correlations Evaluation of Research Funding & Future Research .71** Researcher Credibility .76** .49** What Constitutes Expert .28** .23** .16** Ambivalent Sexism Inventory Benevolent .11 .12* .04 .13* Ambivalent Sexism Inventory Hostile .21** .18** .20** .15* .40** Descriptives Mean 4.56 4.19 4.86 3.51 2.73 2.48 SD 1.04 1.28 0.97 0.93 0.85 0.85 ** p < .01. p < .05. Table 3 5 Cell Means (Study 2) Female Male Control M SD n M SD n M SD n Evaluations of Research 4.67 1.04 93 4.56 1.10 108 4.41 0.97 67 Future Funding & Research 4.16 1.30 93 4.35 1.27 108 3.97 1.26 67 Researcher Credibility 4.91 0.97 93 4.85 0.97 108 4.80 0 .98 67

PAGE 64

64 Table 3 6. Regressi on Evaluat ions of Research, WCE c ovariate (Study 2) b SE t p r p Step 1 Intercept 4.55 .07 69.44 .00 Female vs. Male .06 .07 .74 .46 .05 Control vs. Else .07 .05 1.36 .18 .08 Step 2 Intercept 4.54 .06 72.29 .00 Female vs. Male .04 .07 .54 .59 .03 Control vs. Else .08 .05 1.62 .11 .10 WCE (mean centered covariate) .325 .07 4.86 .00 .29 Step 3 Intercept 4.55 .06 72.74 .00 Female vs. Male .04 .07 .53 .00 .03 Control vs. Else .08 .05 1.71 .09 .11 WCE (mean centered covariate) .296 .07 4.25 .00 .25 WCE Female vs. Male .19 .08 2.41 .01 .15 WCE Control vs. Else .04 .05 .69 .49 .04

PAGE 65

65 Table 3 7 Regression Funding & Future Research WCE c ovariate (Study 2) b SE t p r p Step 1 Intercept 4.16 .08 5 2.21 .00 Female vs. Male .09 .09 1.04 .30 .06 Control vs. Else .10 .06 1.62 .11 .10 Step 2 Intercept 4.16 .08 53.64 .00 Female vs. Male .11 .09 1.27 .21 .08 Control vs. Else .11 .06 1.84 .07 .11 WCE (mean centered covariate) .33 .08 4 .02 .00 .24 Step 3 Intercept 4.16 .08 54.17 .00 Female vs. Male .11 .09 1.27 .21 .08 Control vs. Else .12 .06 1.98 .05 .12 WCE (mean centered covariate) .30 .09 3.56 .00 .22 WCE Female vs. Male .26 .10 2.64 .01 .16 WCE Control vs. Else .09 .07 1.38 .17 .09

PAGE 66

66 Table 3 8. Regression Researcher Cre dibility, WCE c ovariate (Study 2) b SE t p r p Step 1 Intercept 4.85 .06 80.21 .00 Female vs. Male .03 .07 .49 .63 .0 3 Control vs. Else .03 .05 .57 .57 .04 Step 2 Intercept 4.85 .06 81.10 .00 Female vs. Male .03 .07 .36 .72 .02 Control vs. Else .03 .05 .69 .49 .04 WCE (mean centered covariate) .17 .06 2.66 .01 .16 Step 3 Intercept 4.86 .06 81 .19 .00 Female vs. Male .02 .07 .33 .74 .02 Control vs. Else .03 .05 .69 .49 .04 WCE (mean centered covariate) .14 .07 2.03 .04 .13 WCE Female vs. Male .12 .08 1.54 .13 .10 WCE Control vs. Else .03 .05 .63 .53 .04

PAGE 67

67 Table 3 9 Regression Evaluatio ns of Research, ASI benevolent c ovariate (Study 2) b SE t p r p Step 1 Intercept 4.55 .07 69.44 .00 Female vs. Male .06 .0 7 .74 .46 .05 Control vs. Else .07 .05 1.36 .18 .08 Step 2 Intercept 4.55 .07 69.74 .00 Female vs. Male .05 .07 .73 .46 .05 Control vs. Else .07 .05 1.37 .17 .08 ASI benevolent (mean centered covariate) .137 .08 1.81 .07 .11 Ste p 3 Intercept 4.55 .07 69.96 .00 Female vs. Male .05 .07 .73 .47 .05 Control vs. Else .07 .05 1.37 .17 .08 ASI benevolent (mean centered covariate) .12 .08 1.61 .11 .10 ASI benevolent Female vs. Male .14 .09 1.60 .11 .10 ASI benevolent Control vs. Else .06 .06 1.04 .30 .06

PAGE 68

68 Table 3 10. Regression Funding & Future Research, ASI benevolent c ovariate (Study 2) b SE t p r p Step 1 Intercept 4.6 .08 52.21 .00 Female v s. Male .09 .09 1.04 .30 .06 Control vs. Else .10 .06 1.62 .11 .10 Step 2 Intercept 4.16 .08 52.50 .00 Female vs. Male .10 .09 1.06 .29 .07 Control vs. Else .10 .06 1.63 .10 .10 ASI benevolent (mean centered covariate) .18 .09 1.99 .05 .12 Step 3 Intercept 4.16 .08 52.94 .00 Female vs. Male .10 .09 1.08 .28 .07 Control vs. Else .10 .06 .1.65 .10 .10 ASI benevolent (mean centered covariate) .15 .09 1.66 .10 .10 ASI benevolent Female vs. Male .18 .11 1.68 .09 .10 ASI benevolent Control vs. Else .13 .07 1.86 .06 .11

PAGE 69

69 Table 3 11. Regression Researcher Credibility, ASI benevolent covariate (Study 2) b SE t p r p Step 1 Intercept 4.85 .06 80.21 .00 Female vs. Male .03 .07 .49 .63 .03 Control vs. Else .03 .05 .57 .57 .04 Step 2 Intercept 4.85 .06 80.12 .00 Female vs. Male .03 .07 .48 .63 .03 Control vs. Else .03 .05 .57 .57 .04 ASI benevolent (mean centered covariate) .04 .07 .60 .55 .04 Step 3 Intercept 4.85 .06 80.68 .00 Female vs. Male .03 .07 .48 .63 .03 Control vs. Else .03 .05 .58 .56 .04 ASI benevolent (mean centered covariate) .03 .07 .33 .74 .02 ASI benevolent Female vs. Male .15 .08 1.81 .07 .11 ASI benevolent Control vs. Else .08 .05 1.52 .13 .09

PAGE 70

70 Table 3 1 2 Regression Evaluations of Research, ASI hostile covariate (Study 2) b SE t p r p Step 1 Intercept 4.55 .07 69.44 .00 Female vs. Ma le .06 .07 .74 .46 .05 Control vs. Else .07 .05 1.36 .18 .08 Step 2 Intercept 4.55 .06 70.77 00 Female vs. Male .05 .07 .66 .51 .04 Control vs. Else .06 .05 1.17 .24 .07 ASI hostile (mean centered covariate) .25 .07 3.32 .00 .20 Step 3 Intercept 4.54 .06 71.00 .00 Female vs. Male .05 .07 .65 .52 .04 Control vs. Else .07 .05 1.35 .18 .08 ASI hostile (mean centered covariate) .22 .08 2.88 .00 .17 ASI hostile Female vs. Male .06 .09 .66 .51 .04 ASI hostile Control vs. Else .13 .06 2.41 .02 .14

PAGE 71

71 Table 3 1 3 Regression Funding & Future Research, ASI hostile covariate (Study 2) b SE t p r p Step 1 Intercept 4.16 .08 52.21 .00 Female vs. Male .09 .09 1.04 .30 .06 Control vs. Else .10 .06 1.62 .11 .10 Step 2 Intercept 4.16 .08 52.98 .00 Female vs. Male .10 .09 1.15 .25 .07 Control vs. Else .09 .06 1.46 .15 .09 ASI hostile (mean centered covariate) .27 .09 2.93 .00 .18 Ste p 3 Intercept 4.15 .08 53.55 .00 Female vs. Male .10 .09 1.15 .25 .07 Control vs. Else .10 .06 1.70 .09 .10 ASI hostile (mean centered covariate) .21 .09 2.36 .02 .14 ASI hostile Female vs. Male .16 .11 1.54 .13 .09 ASI hostile Control vs. Else .19 .07 2.84 .01 .17

PAGE 72

72 Table 3 1 4 Regression Researcher Credibility, ASI hostile covariate (Study 2) b SE t p r p Step 1 Intercept 4.85 .06 80.21 .00 Female vs. Male .03 .07 .49 .63 .03 Control vs. Else .03 .05 .57 .57 .04 Step 2 Intercept 4.86 .06 81.68 .00 Female vs. Male .03 .07 .40 .69 .03 Control vs. Else .02 .05 .38 .70 .02 ASI hostile (mean centered covariate) .22 .07 3.25 .00 .20 Ste p 3 Intercept 4.85 .07 81.19 .00 Female vs. Male .03 .07 .39 .70 .02 Control vs. Else .02 .05 .44 .66 .03 ASI hostile (mean centered covariate) .21 .07 3.05 .00 .19 ASI hostile Female vs. Male .02 .08 .20 .84 .01 ASI hostile Control vs. Else .05 .05 .87 .38 .05

PAGE 73

73 Table 3 1 5 Regression Evaluations of Research, participant gender (Study 2) b SE t p r p Step 1 Intercept 4.54 .07 68.88 .00 Female vs. Male .07 .07 .94 .35 .06 Control vs. Else .06 .05 1.14 .25 .07 Step 2 Intercept 4.54 .07 68.32 00 Female vs. Male .07 .08 .93 .36 .06 Control vs. Else .06 .05 1.15 .25 .07 Participant Gender .01 .07 .09 .93 .01 Step 3 Intercept 4.5 4 .07 67.51 .00 Female vs. Male .07 .08 .93 .35 .06 Control vs. Else .06 .05 1.19 .24 .07 Participant Gender .01 .07 .18 .86 .01 Participant Gender Female vs. Male .00 .08 .03 .98 .00 Participant Gender Control vs. Else .02 .05 .37 .71 .02

PAGE 74

74 Table 3 1 6 Regression Funding & Future Research, participant gender (Study 2) b SE t p r p Step 1 Intercept 4.15 .08 51.78 .00 Female vs. M ale .08 .09 .86 .39 .05 Control vs. Else .09 .06 1.43 .15 .09 Step 2 Intercept 4.17 .08 51.67 .00 Female vs. Male .07 .09 .77 .44 .05 Control vs. Else .08 .06 1.34 .18 .08 Participant Gender .10 .08 1.28 .20 .08 Step 3 Intercept 4 .16 .08 51.12 .00 Female vs. Male .06 .09 .70 .49 .04 Control vs. Else .09 .06 1.44 .15 .09 Participant Gender .09 .08 1.14 .26 .07 Participant Gender Female vs. Male .01 .09 .61 .55 .04 Participant Gender Control vs. Else .04 .0 .61 .54 .04

PAGE 75

75 Table 3 1 7 Regression Researcher Credibility, participant gender (Study 2) b SE t p r p Step 1 Intercept 4.86 .06 79.53 .00 Female vs. Ma le .03 .07 .49 .63 .03 Control vs. Else .02 .05 .48 .63 .03 Step 2 Intercept 4.86 .06 79.04 .00 Female vs. Male .04 .07 .53 .59 .03 Control vs. Else .02 .05 .43 .67 .03 Participant Gender .04 .06 .72 .47 .05 Step 3 Intercept 4.86 .0 78.26 .00 Female vs. Male .03 .07 .46 .65 .03 Control vs. Else .02 .05 .33 .74 .02 Participant Gender .05 .06 .72 .47 .05 Participant Gender Female vs. Male .06 .07 .92 .36 .06 Participant Gender Control vs. Else .02 .05 .34 .73 .02

PAGE 76

76 Figure 3 1. Screenshot: Summary Article of Petersen, J. L., & Hyde, J. (2010). A meta analytic review of research on gender differences in sexuality, 199 3 2007. Psychological Bulletin, 136(1), 21 38. doi:10.1037/a0017504

PAGE 77

77 Figure 3 2. Pictures used in female researcher, male researcher, and control conditions. Participants were randomly assigned to view a screenshot with one of the following images.

PAGE 78

78 Figure 3 3 Factor structure for the full (8 item) What Constitutes an Expert scale (WCE; Appendix C ). [R] indicates reverse scored items. Significant loadings are denoted by *.

PAGE 79

79 Figure 3 4 Fac tor structure for the reduced (5 item) What Constitutes an Expert scale (WCE; Appendix C ). [R] indicates reverse scored items. Significant loadings are denoted by *.

PAGE 80

80 Figure 3 5 Predicted scores on the evaluations of resear ch quality dependent variable for one SD above and one SD below the mean on the What Constitutes an Expert covariate. denotes a significant simple effect of gender. Figure 3 6 Predicted scores on the evaluations of rese arch quality dependent variable for one SD above and one SD below the mean on the hostile sexism (ASI hostile) covariate. denotes a significant simple effect of gender.

PAGE 81

81 Figure 3 7 Predicted scores on the funding & futur e research dependent variable for one SD above and one SD below the mean on the What Constitutes an Expert covariate. denotes a significant simple effect of gender. Figure 3 8 Predicted scores on the funding and future research dependent variable for one SD above and one SD below the mean on the benevolent sexism (ASI benevolent) covariate. denotes a significant simple effect of gender.

PAGE 82

82 Figure 3 9 Predicted scores on the funding and f uture research dependent variable for one SD above and one SD below the mean on the benevolent sexism (ASI benevolent) covariate. denotes a significant simple effect of gender. Figure 3 10 Predicted scores on the fundin g and future research dependent variable for one SD above and SD below the mean on the hostile sexism (ASI hostile) covariate. denotes a significant simple effect of gender.

PAGE 83

83 Figure 3 11 Predicted scores on the researcher credibility dependent variable for one SD above and one SD below the mean on the benevolent sexism (ASI benevolent) covariate.

PAGE 84

84 CHAPTER 4 OVERALL DISCUSSION In Study 1, there was no evidence for main effects or interactions of researcher race or researc her gender. There was evidence for a main effect of university prestige, but it was in a counter intuitive direction, with higher prestige causing more negative perceptions of the research and researcher. In Study 2, there was no evidence for main effects of researcher gender, but researcher gender did interact with the personal opinions on expertise and the benevolent sexism covariates. Despite a lack of significant main effects for the researcher gender independent variable, two interesting sets of resul ts emerged. First, an interesting pattern of significance occurred among the dependent variables. While researcher gender interacted with the expertise covariate for the two dependent variables relating the quality of the research itself (evaluations of re search, funding and future research), gender did not interact with the expertise covariate for the dependent variable relating to the researcher (researcher credibility variable). In fact, there were no significant effects for the researcher credibility va riable at all, only one marginally significant interaction. In other words, participants rated the research itself differently depending on opinions of what constitutes an expert and researcher gender, but they did not rate the researcher differently depen ding on opinions of what constitutes an expert and researcher gender. Either the interaction of gender and opinions on expertise truly did not affect perceptions of the researcher, or participants were simply unwilling to admit to it when it came time to e valuate the researcher him or herself. Future research should seek to distinguish between these two possible explanations.

PAGE 85

85 A second intriguing set of results related to the covariate simple effects observed when breaking down interaction effects by gender. Across dependent variables, means for the female researcher condition showed little variation, while means for the male There are at least two possible explanations for benevolent sexism, hostile sexism, or opinions on what constitutes an expert, participants did not vary in their perceptions of women researchers or of research conducted by women. In other words, sexism and opini ons regarding what constitutes an expert did not affect how participants viewed women and their research (though it did seem to affect how participants viewed men and their research). A second possible explanation is that participants were simply unwilling to admit to perceiving women and their research any differently according to their own opinions of what constitutes an expert or their own level of sexism. I suspect that the latter explanation is true; scores on the dependent variables hovered around sca le midpoints for the female researcher condition, but for the male researcher condition, scores on the dependent variables constitutes an expert. Although many people today may still hold sexist views and stereotypes that are disparaging to women, there may be social norms against female scientists and their research are truly unaffected by se xism and opinions regarding expertise, or if participants are unwilling to admit to being so affected. Future research could compare implicit and explicit attitudes towards women who study gender prejudice or stereotypes research.

PAGE 86

86 Limitations and Future D irections There are several limitations to the research presented here. The first limitation is my choice of multiple regression analysis over multi level modeling. As discussed owever, there would have been problems with a multilevel model as well (see Main analyses in the Results section for Study 2). Additionally, although using multiple photos for the two experimental conditions posed a nesting problem, I believe the stimulus sampling in this study was an improvement over prior studies that did not use stimulus sampling techniques. Using one picture for each condition would have circumvented nesting and obviated the need for a multilevel model. However, failing to stimulus samp le would have posed problems for generalizability and construct validity (Wells & Windschitl, 1999). Using only one female photo and only one male photo would have limited the g only one female photo and one male photo would confound the effect of gender with the effect of those particular faces. It would have been impossible to know whether any observed gender effects were true gender effects or the result of idiosyncratic diff erences in the two targets (one male, one female). Using only one photo per condition could result in Type 1 errors (falsely rejecting a null hypothesis, claiming to find an effect or difference when in fact there is none), Type 2 errors (falsely retaining a null hypothesis, claiming that no effect or difference exists when evidence supports its diametrical error occurs when a null hypothesis is correctly rejected, but the sample direction of the effect is in the wrong direction, differing from the population direction of the effect (Wells & Windschitl, 1999). Wells and Windschitl (1999) showed that failure to

PAGE 87

87 stimulus sample could result in any of these three errors, and t his compromises a have avoided the nesting problem in this study, it would have presented other serious problems, which would have defeated the purpose of conducting a n experiment. Another limitation is that the dependent variables are self report, rather than behavioral. The self report nature of the dependent variables is appropriate given that perceptions of prejudice and stereotyping researchers and their work were the outcome of interest. Nevertheless, to the extent that perceptions affect behavior, it would be ideal for future research to have some sort of behavioral outcomes. Psychologists should want to be perceived positively by others for several reasons that extend beyond perceptions. First, they may want to be viewed positively so that their research based recommendations on how to solve problems are taken seriously. Psychologists may also desire to be perceived positively so that people in power (e.g., grant review panelists, job search committees) will continue to allocate resources and opportunities for research to psychologists (e.g., funding, research jobs). Another limitation to this study is that it examined lay perceptions of prejudice and stereotypi ng research, but not necessarily perceptions from the scientific community. It is possible that some of the participants had careers in science, but it is likely that the majority did not. Scientist perceptions of prejudice and stereotyping research and re searchers are particularly relevant on the job market. To the extent that researcher gender could interact with audience variables (e.g., sexism, opinions regarding scientific expertise) to affect perceptions of research quality and researcher credibility, stereotype and prejudice researchers may miss job opportunities because a

PAGE 88

88 different stereotype and prejudice researcher is seen as more credible, or because researcher and audience factors combine to negatively influence the perceived quality or importanc e of prejudice and stereotyping research. Future research should include participants sampled from the scientific community, including psychologists and non psychologists. Although the research presented here was focused specifically on prejudice and ste reotype research conducted by relevant stigmatized groups, this specific focus is limiting. It may be inappropriate to generalize the specific effects found here to other types of research or other characteristics of scientists. Future studies could examin e other types of prejudice and stereotype research (and relevant personal characteristics), and could also compare perceptions of stereotype and prejudice research to perceptions of other social psychological research (e.g., women who study sexism vs. wome n who study judgment and decision making, racial minorities who study racism vs. racial minorities who study altruism). Finally, the online nature of these studies poses some limitations. To an extent, the online nature of this study is efficient completi ng a study online rather than coming to lab is much more convenient for most participants. Further, because these studies were completed online, they may have had more mundane realism. The circumstances under which participants completed these studies were probably similar to circumstances under which participants read about research at home, listening to music, chatting online with friends, or switching rapidly among websites. However, experimenters have much less environmental control in online experiment s. When participants come to the lab to participate in studies, researchers can control the

PAGE 89

89 environment around them to minimize distraction and to mitigate any potential extraneous influences. Unlike participants who come to the lab, there is no such contr ol over environmental conditions and extraneous influences with most online samples. Implications and Conclusions Although perceptions were the outcome of interest in these studies, the importance of the results extend beyond perceptions to tangible outcom es. Public directions (Lilienfield, 2012). There has been some research on perceptions of scientific research. For example, a study by Wilson et al. (1993) showed t hat scientists were more likely to overlook methodological flaws in a study when the topic was considered important. In a study on perceptions of scientific research and methodology, Munro (2010) found that people will dismiss results of scientific resear ch that are at odds with their own beliefs, claiming that the topic of research is not suitable for scientific investigation. Beyond these studies, however, there is little other research (to my knowledge) that examines lay perceptions of psychological res earch (and none that specifically examines perceptions of prejudice research and researchers). The focus of this dissertation was on psychologists who research topics that are self relevant, so its application to fields beyond psychology is limited but st ill possible. For example, there is nothing inherent about a female or male astronomer, or a scientist and research is questionable because the topic is too self relevant to t he choose self relevant fields. For example, a biologist who has cancer, or whose close family member has cancer, may choose to study cancer treatment. A multilingual

PAGE 90

90 neurologist may choose to study differences in patterns of brain activation between bilingual and monolingual children. A Southeast Asian anthropologist may do field work in Southeast Asia. To the extent that people who study self relevant topics may be pe rceived differently than their colleagues who do not study self relevant topics, and to the extent that this self relevance is readily apparent (a scientist can, if possible, choose to hide these personal characteristics), the research here may be relevant Lay people, job search committee members, grant review panelists, and other third parties may differ in their opinions on what constitutes a scientific expert, and so may view scientists differently depending on the perceived self relevance of those scie These perceptions may in turn affect funding and research opportunities. Perceptions of science and scientist credibility are important to explore scientifically because these perceptions can affect funding and job opportunities for scie example, prejudice researchers presenting data for a job talk or at an academic conference may find that the validity of their work or their credibility as scientists is de constitutes expertise and their perceptions of people who study self relevant prejudice. A fellow graduate student and friend of mine once voiced his concern that people would no middle ha

PAGE 91

91 world problems, i t is import ant that we understand when psychological credibility is undercut by perceptions of bias.

PAGE 92

92 APPENDIX A EVALUATIONS OF RESEARCH AN D RESEARCHER CREDIBILITY SCALE You will now be asked some questions to evaluate the research you just read about. In some of these questions there may be reference to the primary researcher or psychologist who conducted the gender differences and similari ties research. Note that these questions refer to the psychologist whose research was featured on the screen capture you read -NOT the researcher whose name appeared on the informed consent page (the very first page) of this survey. Evaluations of Resea rch Items: 1. How important would you say the topic of this study is?* 1 2 3 4 5 6 7 Not at all important Very important 2. Based on the study description you read, how likely would you be to recommend that a report of this study be accepted by a high quality academic journal? 1 2 3 4 5 6 7 Very unlikely Very likely 3. How methodologically sound or flawed would you say the design of this study is? 1 2 3 4 5 6 7 Very flawed Very sound 4. supported by the data? 1 2 3 4 5 6 7 Not at all supported Very well supported 5. To what extent do you agree with the conclusions drawn by the researchers? 1 2 3 4 5 6 7 Not at all Very much 6. To what extent do you think these results can be trusted? 1 2 3 4 5 6 7 Not at all Very much

PAGE 93

93 7. Do the results of the research you read about reflect the real world? 1 2 3 4 5 6 7 Not at all Very much Future Funding & Research Items: 1. How well does the study you read about sti mulate new areas of research? 1 2 3 4 5 6 7 Not at all Very well 2. More funding should be given to research like the kind summarized in the article. 1 2 3 4 5 6 7 Disagree very strongly Agree very strongly 3. Researchers should invest m ore time into studies like the one summarized in the article. 1 2 3 4 5 6 7 Disagree very strongly Agree very strongly Researcher Credibility Items: 1. How credible do you think the researchers are? 1 2 3 4 5 6 7 Not at all credible Very cr edible 2. Based on the information given, how competent do you find the researchers? 1 2 3 4 5 6 7 Not at all competent Very competent 3. Based on the information given, how biased would you say the researchers are? [R] 1 2 3 4 5 6 7 Not at al l biased Very biased

PAGE 94

94 4. Based on the information given how objective would you say the researchers are? 1 2 3 4 5 6 7 Not at all objective Very objective 5. How qualified do you think the primary researcher is to conduct research like the kind you read about? 1 2 3 4 5 6 7 Not at all qualified Very qualified 6. How likely are other researchers to believe the conclusions made by the researchers of this study? 1 2 3 4 5 6 7 Not at all likely Very likely Note : *indicates item s taken from Wilson et al. (1993).

PAGE 95

95 APPENDIX B MANIPULATION CHECK, ATTENTION CHECKS, & QUALITY CHECKS (STUDY 2) Manipulation Check: 1. What was the gender of the psychologist who conducted the gender differences research you just read about? Female Male No t sure Attention Checks: 2. What ethnicity would you say the primary researcher is? 3. In the article you read, what was the overall conclusion about the size of gender differences? Quality Checks: 4. 5. 6. 7.

PAGE 96

96 APPENDIX C WHAT CONSTITUTES AN EXPERT (WCE) Please indicate your agreement with the following statements using t his scale. 1 2 3 4 5 6 7 Disagree strongly Disagree moderately Disagree a little Neither agree nor disagree Agree a little Agree moderately Agree strongly Personal Experience vs. Emotional Detachment and Objectivity 1. A researcher should have a lot of pe rsonal experience in his or her chosen field of work.* 2. Personal experience is not important in a researcher as long as the researcher has a degree in his or her chosen field of work. [R]* 3. Researchers should choose topics of study that have a lot of relev ance to their personal lives. 4. Researchers who study issues that are personally relevant to them run the risk influencing the results of their studies. [R] 5. Researchers who study issues that are personally relevant to them are no more likely to influence the results of their studies than are researchers who study topics that are not more personally relevant. 6. Researches should choose topics of study for which they can be detached and objective. [R] 7. Researchers who study issues for which they can be detac hed and objective do not run the risk of influencing the results of their studies. [R]* 8. Researchers who study issues for which they can be detached and objective are no less likely to influence the results of their studies than are researchers who study i ssues they are personally invested in. Note: [R] indicates reverse scored items. This set of items is to be scored so that higher scores indicate preferences for personal relevance and firsthand experience in a researcher, whereas lower scores indicate p reference for emotional detachment and objectivity in researcher. *indicates items that were dropped for final analyses due to poor item loadings on a confirmatory factor analysis.

PAGE 97

97 APPENDIX D AMBIVALENT SEXISM INVENTORY (ASI; GLICK & FISKE, 1996) Please mark the response that most accurately represents your views. 1 2 3 4 5 Disagree Strongly Agree Strongly 1. No matter how accomplished he is, a man is not truly complete as a person unless he has the love of a woman. 2. Many women are actually seeking special favors, such as hiring policies that favor them over men, under the guise of asking for "equality." 3. In a disaster, women ought not necessarily to be rescued before men. [R] 4. Most women interpret innocent remarks or acts as being sexist. 5. Women are too easily offended. 6. People are often truly happy in life without being romantically involved with a member of the other sex. [R] 7. Feminists are not seeking for women to have more power than men. [R] 8. Many women have a qu ality of purity that few men possess. 9. Women should be cherished and protected by men. 10. Most women fail to appreciate fully all that men do for them. 11. Women seek to gain power by getting control over men. 12. Every man ought to have a woman whom he adores. 13. M en are complete without women. [R] 14. Women exaggerate problems they have at work. 15. Once a woman gets a man to commit to her, she usually tries to put him on a tight leash. 16. When women lose to men in a fair competition, they typically complain about being d iscriminated against.

PAGE 98

98 17. A good woman should be set on a pedestal by her man. 18. There are actually very few women who get a kick out of teasing men by seeming sexually available and then refusing male advances. [R] 19. Women, compared to men, tend to have a sup erior moral sensibility. 20. Men should be willing to sacrifice their own well being in order to provide financially for the women in their lives. 21. Feminists are making entirely reasonable demands of men. [R] 22. Women, as compared to men, tend to have a more r efined sense of culture and good taste. Note: [R] indicates reverse scored items. Benevolent subscale: items 1, 3, 6, 8, 9, 12, 13, 17, 19, 20, and 22. Hostile subscale: items 2, 4, 5, 7, 10, 11, 14, 15, 16, 18, and 21.

PAGE 99

99 APPENDIX E DEMOGRAPHICS Please answer the following questions. You may skip any question that you do not feel comfortable answering, but please remember that all responses are confidential and will not be linked back to you. 1. How old are you (in years)? 2. What is your gender? Female Ma le Transgender Prefer not to answer 3. Which of the following religions best describes you? Buddhist Catholic Fundamentalist/Evangelical Christian Hindu Jewish Muslim Protestant (Methodist, Lutheran, Episcopalian, etc.) Other No Religious Affiliation 4. If you 5. On a continuum from liberal to moderate to conservative, how would you describe your political beliefs? 1 2 3 4 5 6 7 Liberal (or left wing) Moderate Conservative (or right wing) 6. Which of the following political parties do you identify with most? Democrat Green Party Libertarian Republican Other/No Party affiliation 7. How much money does your family make (i.e., income before taxes) in an average year? (Please estimate if unknown.)

PAGE 100

100 8. How much money do you make (i.e., income before taxes) in an average year? (Please estimate if unknown. Negative income is fine.) 9. How much debt do you have? (Please estimate if unknown) 10. What is the highest level of education you have completed? Less than High Sc hool High School/GED Some College 2 year College Degree 4 year College Degree Doctoral Degree Professional Degree (JD, MD) 11. In which educational categories do you have children? Please select all that apply. Less than 3 years old Nursery s chool or preschool Kindergarten Elementary: grades 1 4 Elementary: grades 5 9 High school: grades 9 12 College, undergraduate Graduate or professional school 12. Where do you currently live (city, state, country)? 13. Where d id you grow up (city, state, country)? 14. Which of the following ethnic groups do you consider yourself a member of? You can check multiple groups. African Asian Caucasian/White Hispanic/Latino(a) Pacific Islander Other 15. ic group, please specify.

PAGE 101

101 REFERENCES Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions Thousand Oaks, CA US: Sage Publications, Inc. Aronson, E., Turner, J. A., & Carlsmith, J. (1963). Communicator credibilit y and communication discrepancy as determinants of opinion change. The Journal Of Abnormal And Social Psychology 67 (1), 31 36. doi:10.1037/h0045513 Asch, S. E. (1956). Studies of independence and conformity: I. A minority of one against a unanimous majori ty. Psychological Monographs: General And Applied, 70(9), 1 70. doi:10.1037/h0093718 Ayres, I., & Siegelman, P. (1995). Race and gender discrimination in bargaining for a new car. The American Economic Review, 85 (3), 304 321. Badash, D. (2011, April 26). Prop 8: Anti complete text. Retrieved from http://thenew civilrightsmovement.com/prop 8 anti gay lawyers motion to vacate walker ruling complete text/discrimination/2011/04/26/19289 Cacioppo, J.T., & Petty, R.E. (1986). Communication and persuasion: Central and peripheral routes to attitude change. New York: Sp ringer Verlag. Chaiken, S. (1979). Communicator physical attractiveness and persuasion. Journal of Personality and Social Psychology, 37, 1387 1397. Glick, P., & Fiske, S. T. (1996). The Ambivalent Sexism Inventory: Differentiating hostile and benevolent sexism. Journal Of Personality And Social Psychology, 70(3), 491 512. doi:10.1037/0022 3514.70.3.491 Glick, P., & Fiske, S. T. (1997). Hostile and benevolent sexism: Measuring ambivalent sexist attitudes toward women. Psychology Of Women Quarterly, 21(1), 119 135. doi:10.1111/j.1471 6402.1997.tb00104.x Jung, J., & Kellaris, J. J. (2006). Responsiveness to authority appeals among young French and American consumers. Journal Of Business Research, 59(6), 735 744. doi:10.1016/j.jbusres.2006.01.011 Kenrick, D. T ., Neuberg, S. L., & Cialdini, R. B. (2010). Social psychology: Goals in interaction (5th ed.). Boston, MA: Allyn & Bacon. Levine, D. (2011, April 6). Gay judge never thought to drop marriage case. Retrieved from http://www.reuters.com/article/2011/04/06/us gaymarriage judge idUSTRE7356TA20110406

PAGE 102

102 Lilienfeld, S. O. (2012). Public skepticism of psychology: Why many people perceive the study of human behavior a s unscientific. American Psychologist, 67(2), 111 129. doi:10.1037/a0023963 McClelland, G. H. (1997). Optimal design in psychological research. Psychological Methods, 2(1), 3 19. doi:10.1037/1082 989X.2.1.3 Munro, G. D. (2010). The scientific impotence exc use: Discounting belief threatening scientific abstracts. Journal Of Applied Social Psychology, 40(3), 579 600. doi:10.1111/j.1559 1816.2010.00588.x Olson, J.M., & Cal, A.V. (1984). Source credibility, attitudes, and the recall of past behaviours. European Journal of Social Psychology, 94, 757 776. Petersen, J. L., & Hyde, J. (2010). A meta analytic review of research on gender differences in sexuality, 1993 2007. Psychological Bulletin, 136(1), 21 38. doi:10.1037/a0017504 Ross, J., Irani, L., Silberman, M. Zaldivar, A., & Tomlinson, B. (2010, April). Who are the crowdworkers?: shifting demographics in mechanical turk. In Proceedings of the 28th of the international conference extended abstracts on Human factors in computing systems (pp. 2863 2872). ACM. Sa garin, B. J., Cialdini, R. B., Rice, W. E., & Serna, S. B. (2002). Dispelling the illusion of invulnerability: The motivations and mechanisms of resistance to persuasion. Journal Of Personality And Social Psychology, 83(3), 526 541. doi:10.1037/0022 3514.8 3.3.526 Strong, S. R. (1968). Counseling: An Interpersonal Influence Process. Journal Of Counseling Psychology, 15(3), 215 224. doi:10.1037/h0020229 Walker, V.R. (2010, August). Order No C 09 2292 VRW. Retrieved from https://ecf.cand.uscourts.gov/cand/09cv2292/files/09cv2292 ORDER.pdf Wells, G. L., & Windschitl, P. D. (1999). Stimulus sampling and social psychological experimentation. Personality And Social Psychology Bulle tin, 25(9), 1115 1125. doi:10.1177/01461672992512005 Wilson, T. D., DePaulo, B. M., Mook, D. G., & Klaaren, K. J. (1993). Scientists' evaluations of research: The biasing effects of the importance of the topic. Psychological Science, 4(5), 322 325. doi:10 1111/j.1467 9280.1993.tb00572.

PAGE 103

103 BIOGRAPHICAL SKETCH Tatiana Orozco Schember was born in Mesa, Arizona in 1986. She graduated with highest honors from Hamilton High School in 2004 and enrolled at Arizona State University in August of 2004. She graduated cu m laude with a Bachelor of Science in Psychology from Arizona State University in May of 2008. She enrolled in the graduate program for social psychology at the University of Florida in August of 2008. She earned her M.S. in December 2010; h sis research focused on reducing prejudice using a sociofunctional, threat based approach. She earned her Ph.D. from the University of F lorida in August 2013 ; her doctoral dissertation research incorporated social influence research and perspectives to exa mine lay perceptions of prejudice and stereotype research.