Citation
Seeing Clients as They Really Are: Accurate Client Assessment Despite Therapists' Gender Stereotypes

Material Information

Title:
Seeing Clients as They Really Are: Accurate Client Assessment Despite Therapists' Gender Stereotypes
Creator:
PERRIN, PAUL ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Affective symptoms ( jstor )
Emotional expression ( jstor )
Emotionality ( jstor )
Gender roles ( jstor )
Men ( jstor )
Modeling ( jstor )
Psychological counseling ( jstor )
Stereotypes ( jstor )
Voiced sounds ( jstor )
Women ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Paul Perrin. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
11/30/2007
Resource Identifier:
659860295 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

1 SEEING CLIENTS AS THEY REALLY ARE: ACCURATE CLIENT ASSESSMENT DESPI TE THERAPISTS’ GENDER STEREOTYPES By PAUL PERRIN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Paul Perrin

PAGE 3

3 To my family

PAGE 4

4 TABLE OF CONTENTS page LIST OF TABLES................................................................................................................. ..........6 LIST OF FIGURES................................................................................................................ .........7 ABSTRACT....................................................................................................................... ..............8 CHAPTER 1 INTRODUCTION................................................................................................................... .9 Men, Women, Emotionality, and Culture.................................................................................9 Men, Women, Emotionality, and Science................................................................................9 2 LITERATURE REVIEW.......................................................................................................12 Problems with Stereotypes.....................................................................................................12 Stereotypes and Psychotherapy..............................................................................................12 Current Study.................................................................................................................. ........14 3 MATERIALS AND METHODS...........................................................................................15 Participants................................................................................................................... ..........15 Materials...................................................................................................................... ...........16 Therapy Session 6-min Audio Recordings......................................................................16 Observer Alexithymia Scale............................................................................................18 Therapist Attitude Questionnaire – Short Form..............................................................20 Concerns with Questionnaires Administered on the Internet..........................................21 Procedure...................................................................................................................... ..........21 Data Analysis.................................................................................................................. ........23 4 RESULTS........................................................................................................................ .......25 Data Distribution.............................................................................................................. ......25 Overall Regression............................................................................................................. .....26 Transformation................................................................................................................. ......26 Gender......................................................................................................................... ............27 Beliefs about Emotion Scores................................................................................................ 27 Observer Alexithymia Scale Sub-scale Analyses...................................................................28 Co-varying for Manipulation Check Results , Gender, and Sexual Orientation.....................28 Model 2........................................................................................................................ ....29 Model 3........................................................................................................................ ....29 Model 4........................................................................................................................ ....30 Scale Psychometrics fo r the Current Sample..........................................................................30

PAGE 5

5 5 DISCUSSION..................................................................................................................... ....31 Hypothesis 1: Females Would Be Rate d as More Emotional than Males........................... 31 Hypothesis 2: Stereotype Level Wo uld Predict Client Assessment.......................................32 Subscale Analyses.............................................................................................................. ....33 Limitations and Future Directions..........................................................................................34 APPENDIX A THIRTY-THREE ITEM OBSERV ER ALEXITHYMIA SCALE........................................36 B EIGHT ITEM BELIEFS ABOUT EMOTIONS SCALE.......................................................37 LIST OF REFERENCES...............................................................................................................38 BIOGRAPHICAL SKETCH.........................................................................................................40

PAGE 6

6 LIST OF TABLES Table page 3-1 2 (Gender) x 2 (transformation) gactorial design..............................................................16 3-2 Scaling factors used for each transformation.....................................................................17

PAGE 7

7 LIST OF FIGURES Figure page 4-1 Distribution of Observer Alexithymia Scal e ratings with superimposed cormal curve....25

PAGE 8

8 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science SEEING CLIENTS AS THEY REALLY ARE: ACCURATE CLIENT ASSESSMENT DESPI TE THERAPISTS’ GENDER STEREOTYPES By Paul Perrin May 2007 Chair: Martin Heesacker Major: Psychology The purpose of this study was to examine wh ether psychotherapists’ gender stereotypes bias their assessments of client emotionality. Therapist participants ( n = 248) listened to one of four randomly-assigned audio reco rdings of a contrived therapy session with a client having either a male voice, a female voice, the original male voice electronicall y altered to sound like a female, or the original female voice altered to sound like a male. Participants assessed the client using the Observer Alexithymia Scale (OAS), an observer rating of client emotionality, and completed the Beliefs About Emotions (BAE ) scale, a measure of gender stereotype endorsement, with regard to their own beliefs. It was hypothesized that therapists in both female voice conditions would rate the cl ient as more emotional (less al exithymic) than would therapists in the corresponding male voice conditions. Further, the degree of therapist gender stereotype endorsement (BAE score) was hypothesized be a significant predictor of client emotionality ratings (OAS score). Though therapists strongly endorsed gender stereotypes, neither hypothesis was supported, and their stereotype s did not bias client assessme nt. Implications, limitations, and future directions are discussed.

PAGE 9

9 CHAPTER 1 INTRODUCTION Men, Women, Emotionality, and Culture Gender stereotypes about emotionality perm eate popular culture, holding that men and women are very different in how they e xpress and experience em otions. A well-known illustration of these stereotypes emerges in the book, Men Are from Mars, Women Are from Venus , in which John Gray writes, “Not only do men and women communica te differently but they think, feel, perceive, react, respond, love, need and appreciate differently. . . . The truth of these principles is self-evide nt and can be validated by your own experience as well as by common sense” (Gray, 2004, p. xxx). A critical eye can quickly discern fl aws in this reasoning, but unfortunately many people endorse these belief s. Just like Gray’s assertions, most popular gender stereotypes are based not on empirical evidence but on heuris tics and stereotyping. Accordingly, a stereotype is a standardized me ntal representation that members of a group hold in common, often times representing an oversimplif ied opinion, prejudiced attitude, or uncritical judgment (Merriam-Webster’s Collegiate Dictiona ry, 1998). Even though there has been a great deal of change with women taking more dominant societal roles over the past 30 years, there has been little or no change in the prevalence of gender stereotyping (Leuptow, Garovich-Szabo, & Leuptow, 2001). Many people still conceptuali ze men as hypo-emotional and women as hyperemotional. Men, Women, Emotionality, and Science Moving from popular culture to science, research has shown that women report experiencing emotions of greater intens ity than men do (Grossman & Wood, 1993), judge emotions through facial expressions more accu rately than men do (Hall & Matsumoto, 2004), and rate their distress from interpersonal prob lems as more intense than men do (Birditt &

PAGE 10

10 Fingerman, 2003). Men, on the other hand, tend to engage in problem-focused coping, reducing the intensity of their emotions (Vingerhoets & Van Heck, 1990), and tend to withdraw and restrict affect when talking with their relations hip partners about emoti onally difficult topics (Vogel, Wester, Heesacker, & Madon, 2003). In terms of psychophysiology, women and men display different skin conductance responding wh en viewing emotional films (Kring & Gordon, 1998), and women exhibit more extreme electro myograph physiological responding than men do in emotionally charged situations (Grossman & Wood, 1993). On the surface, research appears to ha ve produced convincing evidence of gender differences in emotionality, but many studies of ten yield contradictory results, which complicate an accurate understanding of gender-based emoti onal differences. Furthermore, most studies involve subjective self-report measures, which ma y suggest that these differences emerge only when participants attempt to present themselves concordantly with their genders (Ickes et al., 2000). This line of thought is the framework fo r LaFrance and Banaji’s (1992) assertion that gender differences in emotions appear only in studies in which (1) an indirect measure of emotions is used, (2) others can perceive the se lf-reported emotion, (3) ther e is an interpersonal context for the emotions, and (4) broad rather than specific emotions are under scrutiny. These conditions imply that conformity to internalized gender roles, rather than actual gender-based differences, underlie observed diffe rences in emotional expression. Though some studies show gender differences in emotionality (wheth er the differences are innate or externally-imposed), research al so indicates that men a nd women are much more alike than different in emotional expression. Me ta-analyses have revealed that men and women exhibit similar emotional behaviors over 98% of the time (Wilkins & Anderson, 1991; Canary & Hause, 1993). Similarly, other studies using se lf-report measures have found that men and

PAGE 11

11 women actually differ little in their ability to express emotions (Mallinckrodt, King, & Koble, 1998). Fischer et al. (1993) have asserted that when gender-ba sed differences in emotional expression do appear, they largel y agree with the stereotypes pr evalent in popular culture; this provides further evidence that rather than emana ting intrinsically, these differences could be due to cultural pressures on males and females to conform to stereotypical gender roles.

PAGE 12

12 CHAPTER 2 LITERATURE REVIEW Problems with Stereotypes Gender stereotypes have become so strong a he uristic in popular cult ure that individuals tend to view men and women almost as two different species, or from two different planets. This heuristic is socially damaging such that individuals apply pr e-molded stereotype-concordant schemata to their expectations for male a nd female behavior, cr eating self-handicapping cognitive frameworks. In effect, these gender roles confine men to emotional impotence and women to life on an emotional ro ller coaster. This heuristic becomes even more problematic when it surfaces in areas that demand the utmo st precision in personality understanding, like psychotherapy. An accurate and unbiased understandi ng of emotions is so fundamental to the practice of psychotherapy that any deficit in a practitioner’s knowledge in this area severely limits his or her ability to effectivel y treat clients (Hees acker et al., 1999). Stereotypes and Psychotherapy The first study to address gender ster eotyping in psychotherapy was Broverman, Broverman, Clarkson, Rosenkrantz, & Vogel (1 970), in which clinical psychologists, psychiatrists, and social worker s filled out mental health ques tionnaires with instructions to complete them for 1 of 3 conditions: an ideal, mature, healthy, socially -competent adult person, man, or woman. No differences emerged between me ntal health scores for a mentally healthy adult and a mentally healthy man, but the mental health score for a mentally healthy woman was significantly lower than the other two. Furthermore, the personality characteristics attributed to a mentally healthy woman were less socially desira ble than those attributed to a man or adult. These included characteristics like being s ubservient, having uncontrollable emotions, and disliking math and science. The findings imply that a double-standard of mental health exists for

PAGE 13

13 men and women and that the general standard of mental health applies only to men, whereas women are perceived as less mental ly healthy. It follows that to be mentally healthy, a woman must adjust to societal norms for her gender, ev en though those norms are less socially desirable. Further, clinicians endorse these gender stereoty pes, and as a result help to perpetuate them (Broverman et al, 1970). More recent research has carried on the torch lit by Broverman et al. (1970), examining psychotherapists’ gender-based cl inical bias. Heesacker et al. (1999) created the Beliefs About Emotions (BAE) scale to examin e the degree to which psychotherapists stereotype men as hypoemotional and women as hyper-emotional. Compared to scores from a group of participants who were instructed to fill out the scale as if th ey believed there were no gender differences in emotionality, therapists’ scores indicated a st rong prevalence of gender stereotypes. This research has shown that therapists are not immune from the same gender-based emotional stereotypes that popular cultu re as a whole endorses. Taking research on the presence of therapists’ gender-based stereotyping a step further, the immediate question becomes, “What do these gender stereotypes mean in terms of working with individual clients, in terms of client a ssessment?” Research addressing this question is difficult to perform and, as a result, sparse. The current study tackles this question, and examines whether therapists’ gender stereotypes predict their assessments of client emotionality. This type of research is vital to counseling psychol ogy because therapist self-awareness of gender stereotyping can help clarify th eir perceptions of individual clie nts, reducing the degree of gender stereotyping that therapis ts bring into sessions. Therap ists need to help break the handicapping gender roles that they endorse, and as a result help to perpetuate. The implications of this research could encourage therapists to assess a more comprehensive emotional picture

PAGE 14

14 (Wester, Vogel, Pressly, & Heesacker, 2002) a nd to examine the rela tionship between actual client emotionality level and demand characteristic s that therapists bring into sessions (LaFrance & Banaji, 1992), which perhaps influence the wa ys clients cope with powerful emotions. Current Study In the current study, therapist participants listened to one of four randomly-assigned audio recordings of a scripted re -enactment of a therapy session from a client with either a male voice, a female voice, the original male voice electr onically altered to sound like a female, or the original female voice altered to sound like a male. The overall design is a 2 (Male Voice or Female Voice) x 2 (Gender Transformation or No Gender Transformation) between-subjects factorial design (Table 1-1). Par ticipants will assess the client using the Observer Alexithymia Scale (OAS), an observer rating of client emotiona lity, and will fill out wi th respect to their own beliefs the Beliefs About Emotions (BAE) scale, a measure of the strength of therapists’ gender stereotypes. The followi ng hypotheses are proposed: HYPOTHESIS ONE. Both female clients will receive hi gher ratings of emotionality (lower OAS scores) than their counterpart male clients. Because therapists endorse the stereotype that men are less emotional than women are (Heesacker et al., 1999), therapists’ emotionality assessments will reflect this bias. HYPOTHESIS TWO. The more strongly therapists endor se gender stereotypes, the more likely they will rate the male client as hypo-emotional and the female client as hyperemotional. Specifically, therapist BAE scores (gender stereotype level) will positively correlate with their OAS ratings (emotionality assessments) of male clients and negatively correlate with their OAS ra tings of female clients.

PAGE 15

15 CHAPTER 3 MATERIALS AND METHODS Participants Participants were recruited from an online email database of 14,064 professional psychotherapists that was drawn from a variety of professiona l organizations and publiclyavailable email lists. Potential participants received an email asking them to participate in a 20min study assessing client emotiona l expression in a 6-min audio recording of a therapy session. From their home computers partic ipants clicked on an internet link in the email which directed them to the study’s webpage. They did not receive compensati on for participating. Out of the 14,064 therapists who received an email solicitation, 282 participated. Five participants were excluded because they failed to respond to at least 90% of the items. Of the remaining 277 participants, 16 more were excluded because they closed and then re-opened their internet browsers during the study, making it im possible to link their responses with the conditions to which they had been originally assigned. Nine others we re excluded because a manipulation check at the end of the study indicated that th ey thought the study involved assigning different therapists to either a male or female client; however, only four of these participants indicated that they thought gender cues of the client’s voice had been altered (one of the four because he or she listened to two diffe rent clips by going through the study twice). Four more participants were excluded because the same manipulation check indicated that they thought the client might have been in a homosexua l relationship, despite two clear statements of a heterosexual marriage at the onset of the study. This series of exclusions resulted in a final sample of 248. Most participants responded to every demographic question, providing the following sample profile: 134 females and 112 males; 217 heterosexuals, 15 homosexuals, 8 bisexuals; 231

PAGE 16

16 Caucasian-Americans, 4 African-Americans, 2 Hi spanics, 2 Asians, and 8 people of other or multiracial ethnicities. There were 238 participants from the U.S.A., 8 from Canada, and 2 who marked “Other.” Psychotherapist part icipants had an average age of 47.87 ( SD = 11.45), average number of years in clinical work of 15.76 ( SD = 11.03), and an average current client caseload per week of 16.87 (SD = 12.66) client s. All participants except one had advanced degrees in their fields; 209 had either a Ph.D. or Psy.D. These data suggest that the vast ma jority of participants were seasoned psychotherapists, and only a few were interns or graduate students. Materials Therapy Session 6-min Audio Recordings. Participants listened to one of four 6-min audi o recordings of a script ed therapy session in which both the client and therapist were actors. A professional film direct or was hired first to audition and select actors based on their skills at delivering a realistic re-enactment of a therapy session, and second to direct the actors during a recording sessi on. Once selected, one male and one female actor both acted the part of the clie nt, and their voices were recorded separately as they read aloud the same script. A third actor, a male, played the part of the therapist. When participants heard the audio recording, the client had either a male voice, a female voice, the original male voice electronically altered to soun d like a female, or the original female voice altered to sound like a male (Table 3-1). Table 3-1. 2 (Gender) x 2 (tra nsformation) factorial design Gender No Transformation Transformation Male Original Male Male Transformed to Female Female Original Female Female Transformed to Male The transformation process was meant to ensure that the emotional leve l and voice fluctuations remained constant across transformations, and th at the only variable modified was gender.

PAGE 17

17 The script from the audio tapes was origina lly taken from a real therapy session of a client expressing feelings about a recent divorce. The scripts for all recordings were identical, and all gender-context words and potentially identifying inform ation were removed. The cues from the male therapist were held constant acros s all four conditions by editing in a separate recording of the therapist’s voice. The spouse was given the name “Erin,” which could be interpreted either as the female “Erin” or the male “Aaron.” The actors’ conversations were record ed using a condenser microphone (AT3031; Audiotechnica, Inc.) on a digi tal tape recorder (Sony, DTCZE700). All recordings were performed in a sound attenuating booth to minimi ze extraneous noise. Th e speech signals were then re-sampled at a sampling frequency of 22.05 kHz and with 16-bit quantization. The digitized samples were edited us ing Audition 1.0 (Adobe Systems, Inc.) to separate the two speakers in the conversation. The clients’ speech samples were then anal yzed to extract their fundamental frequency ( f0 ) and the formant frequencies. The f0 and the formants were then scaled to transform the perceived gender of the speakers. The male voice was transformed to the female voice by scaling the f0 and formants higher, whereas the female-tomale transformation required these parameters to be scaled lower. These changes arise from differences in the length of vocal folds and the vocal tract between the two gende rs. The scaling factors used fo r each of these transformations (Table 3-2) were selected th rough trial and error to achieve the best sounding voice quality. Table 3-2. Scaling factors used for each transformation Transformation Original Fundamental Frequency Fundamental Frequency Scaling Factor Formant Scaling Factor Female-to-Male 170 Hz. 0.75 0.80 Male-to-Female 135 Hz. 1.70 1.20

PAGE 18

18 Since the scaling factor for f0 and formant frequencies were c onstant throughout the utterance, it did not affect the overall prosody for the utterances. All analyses and transformations were pe rformed using the “Speech Transformation and Representation based on Adaptive Interpolati on of Weighted Spectrogram” (STRAIGHT; Kawahara, Masuda-Katsuse, & de Cheveign, 1999). STRAIGHT is a high-quality channel VOCODER that is implemented in MATLAB (Mathw orks, Inc.). It allows separation of the speech signal into its source and filter characteri stics. The source and filter characteristics can then be independently manipulated and resynthe sized, thereby enabling specific changes in the speech output. The analysis of the source and fi lter characteristics can be done with high time resolution, which allows for a high quality re synthesis. The transformed speech samples representing the client were then added back with the natural speech of the therapist. This was achieved using Audition (Adobe Systems, Inc.) a nd restored the complete conversation between the patient and the clinician. This transformation process is the musical equivalent of changing key when playing a song on a piano, though the overall progression of notes and song itself remain constant. The original voices were also digitized to mimic th e slight artificiality of the transformed voices. Observer Alexithymia Scale (OAS) Participants filled out the OAS (Haviland et al., 2000), an internally consistent (coefficient = 0.88 and 0.89) 33-item observer measure of a person’s inability to express emotion (alexithymia). The 33 statements from th e OAS appear in Appendix A. The creators of this scale employ a hierarchi cal second order model to expl ain the general concept of alexithymia. The five first-order factors of this model—Distant, Uninsightful, Somatizing, Humorless, and Rigid—combine to produce the second-order construct, alexithymia. The OAS

PAGE 19

19 includes five sub-scales to measure each fact or. The current study adopts the Haviland et al. (2000) construct of alexithymia because it is the most thorough and empirically-validated construct conceptualized wh en using observer ratings. Haviland et al. (2000) administ ered the OAS to a large number of undergraduates, asking them to fill it out for someone they knew well. No differences were found in the ratings given by male and female raters, but surp risingly, women received higher (more alexithymic) ratings than men on the total scale and on the Uninsightful, Somatizing, and Humorless subscales. Men, on the other hand, received higher ra tings on the Distant subscale. These findings do not conform to the typical gender stereotype that men are more out of touch with their fe elings than women are. The most obvious explanation is that when participants were as ked to rate someone they knew well, they selected the most in teresting people in terms of em otional expression, or those who conform least to the re spective gender stereotypes. Other st udies (Mallinckrodt, King, & Coble, 1998), however, have shown no differences in al exithymia between men and women using the Toronto Alexithymia Scale (Bagby, Taylor, & Parker, 1994), a self-report measure of alexithymia. In the current study, different partic ipants rated the same voice, with only gender manipulated as the independent variable, so the presence of stereotypica l gender-based emotional attributions is hypothesized. The OAS has been shown (Haviland, Warren, Riggs, & Nitch, 2002) to have a moderateto-high correlation ( r = 0.69) with another observer measure of alexithymia, the modified Beth Israel Hospital Psychosomatic Questionnaire (BIQ; Bagby, Taylor, & Parker, 1994). In the OAS, participants are asked to rate 33 statements about the person they are assessing by using the following 4-point Likert scale: 0 = never, not at all like the person; 1 = sometimes, a little like the person; 2 = usually, very much like the person ; 3 = all the time, completely like the person. A

PAGE 20

20 higher overall score indicates that the person receiving the ratings is more alexithymic (less emotionally expressive). This scale was chosen over the BIQ because the OAS is more thorough, better validated, and allows sub-scale assessment. Beliefs About Emotions (BAE) Participants filled out the Beliefs About Em otions (BAE) scale (Heesacker et al., 1999), an internally consistent (coefficient = 0.71 and 0.76) measure of the degree to which therapists endorse emotion-based gender stereotypes. The BA E appears in Appendix B. This scale has been shown (Heesacker et al., 1999) to ha ve a small-to-moderate correlation ( r = .47) with two subscales on the Gender Role Conflict Scale (GRCS -I; O’Neil et al., 1986) , a measure of men’s reactions to the unrealistic gende r roles they face. No differences have been found between the scores of male and female participants (Heesack er et al., 1999). In the BAE, participants are asked to rate 8 statements about emotionality and gender using the follo wing Likert scale: 1 = strongly agree; 2 = agree; 3 = sl ightly agree; 4 = slightly disa gree; 5 = disagree; 6 = strongly disagree. Scores range from 8 to 48, and higher sc ores indicate the pres ence of stronger gender stereotypes. Therapist Attitude Questionna ire – Short Form (TAQ – SF) Participants filled out the sixteen-item Th erapist Attitude Questionnaire – Short Form (TAQ – SF; DiGiusseppe, R., & Li nscott, 1993), which assesses th e degree to which therapists endorse rational versus constructivist attitudes. Th is scale appears in Appendix C. Participants filled out the TAQ – SF after the OAS and before the BAE so that it would buffer the two other scales. The TAQ – SF was chosen because it assesse s attitudes that have no relevance to gender stereotypes, though it includes it ems about the appropriateness of emotionality in therapy. Therefore, therapists would have found a seem ingly natural and logica l progression through the three scales. Since the BAE is a slightly face va lid measure, if it were to appear immediately

PAGE 21

21 after the OAS, there is a stronge r probability that therapists w ould realize that the relationship between their gender stereotypes and client as sessment is under scrutiny. In the TAQ – SF, participants are asked to rate 16 statements accord ing to the following Likert scale: 1 = strongly disagree; 2 = moderately disagree; 3 = neither agree nor disagree; 4 = moderately agree; 5 = strongly agree. A statistical analysis of th e TAQ – SF will not be included in the current analyses. Concerns with Questionnaires Ad ministered on the Internet The internet-based nature of this study presents two major concerns, particularly since it is one of the first studies in which therapists assess a client over the web. The first concern is that an internet sample might differ demographica lly or psychologically from a sample in a traditional laboratory-based study. There is indeed some reassuring truth to this concern, as internet samples have been shown to be more diverse than traditional samples in terms of gender, ethnicity, socioeconomic status , geographic region, and age (G osling, Vazire, Shrivastava, & John, 2004). However, internet samples are not anymore maladjusted, socially isolated, depressed, or unmotivated than samples from tr aditional recruitment met hods (Gosling, Vazire, Shrivastava, & John, 2004). A second concern is that since the thre e questionnaires used in this study were not specifically desi gned for internet sampling, ther e might be differences between these and the traditional paperand-pencil format. Accordingly, research has shown that answers and responses from questionnaires administered over the internet do not differ significantly from those on their paper-and-pencil c ounterparts (Ritter, Lorig, Laur ent, & Matthews, 2004; Gosling, Vazire, Shrivastava, & John, 2004). Procedure When participants entered the study’s website , they first read the informed consent, where they were told that the pu rpose of the study is to “evaluat e an audiotape of therapy, and to

PAGE 22

22 explore therapists’ reactions, per ceptions, attitudes and other factor s in relation to that therapy sample.” They then clicked on a “NEXT” butt on, indicating that they understood the informed consent. Participants were randomly assigned to hear a 6-min audio record ing of a contrived therapy session in one of four conditions: a clie nt with a normal female voice, a normal male voice, the female voice modified electronically to sound like a male, or the male voice modified electronically to sound like a female. Participants in the male or female client conditions accordingly read, “This 6-minute audio clip is from a psychotherapy session in which a client talks about going through a divor ce with his (her) wife (husba nd). The client's voice has been disguised electronically to preser ve his (her) confidentiality.” After finishing listening to the clip, particip ants were asked to fr eely type for no more than 2 minutes about their overa ll thoughts on the client’s level of emotional expression. This exercise primed participants’ gender stereo types and produced qualitative data about the participants’ thoughts while proceeding through the experiment. Participants filled out the OAS in which they assessed the client’s overall emotionality levels, the TAQ – SF which examined therapis ts’ rational and constructivist attitudes and buffered the OAS and BAE, and the BAE which assessed the strength of therapists’ gender stereotypes. The data presented in the current study are a subset of the data from a larger project examining relationships between therapist self-c are and therapeutic orie ntation. Several scales followed the BAE measuring these other construc ts, though that data are not presented in the current analyses because they have no known concep tual or empirical relationship to gender bias in clinical assessments.

PAGE 23

23 Participants then gave their demographic information, such as gender, ethnicity, level of education, and psychological profession. Part icipants received three manipulation check questions, which were (1) To make sure you co mpleted the assessment accurately, what was the client's gender? (2) Specifically, what did you th ink was the study's purpose, and at what point did you decide on this purpose? (3) Did the cl ient's electronically disguised voice still sound realistic? Participants were asked to provide dichotomous an swers for Questions 1 and 3, but asked to provide free responses via a text box for Question 2. Participants pressed a button at the bottom of the page labeled “SUBMIT” which when pressed, took them to a debriefing page. Part icipants learned that the therapy session was contrived and that the voice m odifications were used both to persuade participants of the session’s authenticity and to ex amine how modifications in voi ce cues and intonation affect client assessment. Participants we re asked not to talk to anyone el se about the contrived nature of the therapy session so that othe r potential participants were not cued to the study’s use of a contrived psychotherapy session. Data Analysis The primary statistical analysis was a linear multiple regression with three predictors: the dichotomous variable, voice gender (0 = fema le, 1 = male); the dichotomous variable, transformation (0 = no transformation, 1 = tran sformation); and the continuous variable, BAE score. These were used to predict the continuous criterion variable, OAS ratings. Further, the three predictors were also used in the same mann er to predict each of th e OAS’s five sub-scales. The purpose of these analyses was to examine the degree to which gender stereotypes account for the variability in client emotionality ra tings. Therapist gender, sexual orientation, and manipulation check results were examined as possibly significant covariates.

PAGE 24

24 Dummy coding voice gender and voice transforma tion created a 2 x 2 analysis matrix (Table 3-1), essentially the same analysis as a 2 (Gender) x 2 (Transformation) Analysis of Variance (ANOVA), comparing me an differences among the four conditions on OAS ratings. The inclusion of participants’ BAE scores as a predictor variable in the same analysis allowed an examination of therapists’ measured gender st ereotype level in the prediction of client emotionality ratings.

PAGE 25

25 CHAPTER 4 RESULTS Data Distribution A non-significant Levene’s test for homogeneity of variances showed that the variances for Observer Alexithymia Scale (OAS) ra tings were equal across all 4 conditions, F (3, 244) = .84, p = .475. On the same OAS data, a Kolmogorov-Sm irnov test gave a test statistic of .065, p = .014. Similarly, the Shapiro-Wilk gave a test statistic of .983, p = .005. Because both p -values were less than .05, the overall distribution depart s significantly from normality. However, this finding is most likely an artifact of the large n used in this sample, as visual inspection a graph showing the OAS distribution (Figur e 4-1) indicates that the dist ribution of OAS ratings appears approximately normal. 90.00 80.00 70.00 60.00 50.00 40.00 30.00 20.00 OAS Ratings 30 20 10 0 Frequency Figure 4-1. Distribution of OAS Ratings with Superimposed Normal Curve.

PAGE 26

26 The overall standard deviati on of OAS ratings is 12.50, the sk ewness coefficient is .179 ( SE = .155), and the kurtosis coefficient is -.559, ( SE = .308). The skewness coefficient is fairly small, indicating little skew, a nd the kurtosis coefficient is negative, indicating a very sl ightly flattened distribution, compared with the normal curve. Taken together, th ese findings suggest that these data are approximately normal and therefore suitable for analysis via multiple regression and ANOVAs. Overall Regression To examine the effects of client gender and accordingly therapist ge nder stereotype level on therapists’ emotionality assessments, a multiple regression analysis was performed with the following predictor variables: voice gender (femal e or male), transformation (transformation or no transformation), and Beliefs About Emotions (BAE) score (continuous); and with the OAS overall score serving as the criterion scale. The overall model was not a significant predictor of OAS ratings, F (3, 244) = .90, p = .440, R2 = .011. The following analyses examine each of the three predictor variables in the overall regression. Transformation In the overall multiple regression, tran sformation was not a significant individual predictor of OAS ratings, t (244) = .82, p = .414. To clarify this non-si gnificant result, first a oneway ANOVA was performed, showing that the mean OAS scores for the regular male ( M = 48.89, SD = 12.79) and electronic male ( M = 48.63, SD = 13.83) conditions were not significantly different, F (1, 125) = .013, p = .911, d = .020. A second ANOVA found similar non-significant differences between the OAS scores of the regular female ( M = 47.92, SD = 11.70) and the electronic female ( M = 50.89, SD = 11.59), F (1,119) = 1.97, p = .164, d = .257.

PAGE 27

27 Gender In the overall multiple regression, client ge nder was not a significan t individual predictor of OAS ratings, t (244) = -.40, p = .689. Because the two previous ANOVAs examining transformation did not find signifi cant differences between the el ectronic male and the regular male, or between the electronic female and the regular female, the following ANOVA collapses across transformation to compare both male c onditions to both female conditions. A one-way ANOVA showed that there were no significant differences in OAS ratings between male clients ( M = 48.76, SD = 13.26) and female clients ( M = 49.41, SD = 11.69), F (1, 246) = .170, p = .681, d = .053. Beliefs about Emotion Scores In the overall multiple regression, therapis t BAE score was not a significant individual predictor of OAS ra tings, t(244) = 1.36, p = .174. A one sample t -test was performed to compare therapists’ BAE scores to a score of 8, indicati ng the absence of gender stereotypes. The overall scale ranges from 8 (absence of stereotypes) to 48 (presence of strong stereotypes). This t -test showed that therapists’ BAE scores ( M = 27.05, SD = 5.60) were indicative of significant gender stereotyping, t (247) = 53.58, p < .001, d = 3.40. An exploratory ANOVA examined BAE scores with respect to therapist gender, finding that female therapists ( M = 27.74, SD = 5.62) endorsed stronger gender stereotypes th an male therapists did ( M = 26.17, SD = 5.52), F (1, 244) = 4.47, p = .034, d = .272. Because males and females differed in BAE score, participant gender was included as a covariate in th e overall regression. Though the t -statistic for BAE score increased to t (241) = 1.65, p = .101, BAE score was still not a signifi cant individual predictor of OAS ratings. It was hypothesized that the more strongly th erapists endorse gende r stereotypes, the more likely they would rate the male client as hypo-emotional and the female client as hyper-

PAGE 28

28 emotional. Therapists’ BAE scores (gende r stereotype level) were hypothesized to positively correlate with their OAS ratings (emotionality assessments) of male clients and negatively correlate with their OAS ratings of female c lients. Accordingly, one regression was performed individually for the male client condition a nd one for the female client condition. Each regression contained BAE score as the predicto r variable and OAS ratings as the criterion variable. Neither the regression fo r the female client condition, F (1, 120) = .12, p = .729, R2 = .001, nor the male client condition, F (1, 126) = 2.54, p = .113, R2 = .020, achieved significance. Observer Alexithymia Scale Sub-scale Analyses For each OAS sub-scale (Distant, Uninsightfu l, Somatizing, Humorless, and Rigid), a Bonferroni-corrected multiple regression was performed in the same manner as the overall regression with the following predictors: voice gender (femal e or male), transformation (transformation or no transformation), and BA E score (continuous); and with the criterion variable OAS sub-scale ratings (continuous). All five regressions produced non-significant F s, both for the overall models and for each indivi dual predictor within the overall models. Co-varying for Manipulation Check Resu lts, Gender, and Sexual Orientation To examine possible significant covariates , a linear regression on OAS ratings was performed that included 4 models: (1) predicto rs: voice gender, transf ormation, and BAE score (Model 1 was the same as the overall origin al model); (2) predictors: voice gender, transformation, BAE score, and manipulation ch eck results (0 = pass manipulation check, 1 = fail manipulation check); (3) predictors: voice ge nder, transformation, BAE score, manipulation check results, and participant gender; and (4) predictors: voic e gender, transformation, BAE score, manipulation check results, participant ge nder, and participant se xual orientation (0 = heterosexual, 1 = homosexual, bisexual, or other) . These three covariates were chosen because they had a theoretical basis for affecting gend er stereotypes and emo tionality assessment.

PAGE 29

29 Model 2 Participants were asked at the end of the st udy, “Did the client's el ectronically disguised voice still sound realistic?” They indicated “Yes” or “No.” Howeve r, based on participants’ free responses, many therapists seemed to interpret this question to ask whether the client and therapist sounded like actors. Because of this interpretati on, a new manipulation check was computed in that if participants answered “Yes” to the original question, or if participants made any reference to actors, a script, or phonine ss in their free response sections, they were considered to have failed the manipulation ch eck. With the inclusion of this computed manipulation check as a covariate in Model 2, th e overall model approached but failed to reach statistical significance, F (4, 236) = 2.14, p = .077, and this change was a significant improvement from Model 1, F (1, 236) = 5.06, p = .025, R2 = .035. In Model 2, manipulation check results was a significant individual predictor of OAS ratings, t (236) = 2.25, p = .025. To follow up on this finding, an ANOVA showed that participants who failed the manipulation check a nd thought the session was ar tificial gave clients higher OAS ratings ( M = 48.08, SD = 12.55) than participants w ho did not fail the manipulation check ( M = 51.88, SD = 11.95), F (1, 245) = 4.68, p = .031, d = .308. An exploratory regression was performed in which all pa rticipants who failed the manipulation check were excluded, resulting in an n of 178. The predictors in this regre ssion were voice gender, transformation, and BAE score, and the criterion variable was OAS ratings. The overall model was not a significant predictor of OAS ratings, F (3, 174) = .67, p = .571, and neither were each of the three individual predictors. Model 3 When participant gender was added as a c ovariate to the predictors in Model 2, the overall model approached significance, F (5, 235) = 2.06, p = .071, R2 = .042, but this change

PAGE 30

30 was not a significant improvement from Model 2, F (1, 235) = 1.71, p = .192. None of the individual predictors besides manipulat ion check results were significant. Model 4 When participant sexual orientation was added as a covariate to the predictors in Model 3, the overall model approached significance, F (6, 234) = 2.11, p = .053, R2 = .051, but this change was not a significant improvement from Model 3, F (1, 234) = 2.32, p = .129. Again, none of the individual predicto rs besides manipulation check results were significant. Scale Psychometrics for the Current Sample From the current sample, Cronbach’s s were calculated at .91 for the OAS and .83 for the BAE, showing excellent re liability for both measures.

PAGE 31

31 CHAPTER 5 DISCUSSION It was initially hypothesized (Hypot hesis 1) that both female clients would receive higher ratings of emotionality [lower Observer Alexithymia Scale (OAS) scores] than their counterpart male clients, because therapists endorse the ster eotype that women are more emotional than men (Heesacker et al., 1999). Similarly, it was also predicted (Hypothesis 2) that therapists’ Beliefs About Emotion (BAE) scores (gende r stereotype level) would be a significant predictor of OAS ratings (client emotionality assessments) indivi dually for each client gender. Specifically, therapists’ BAE scores would positively correlate with their OAS ratings of male clients and negatively correlate with their OAS ratings of female clients. Neither of these two hypotheses was supported in the current study. Hypothesis 1: Females Would Be Ra ted as More Emotional than Males Surprisingly, when the level of client emotionality was held constant across gender, therapists rated both the male and female clients as equally emotional. In this study, therapists were dealing with specific client s, and as a result, they were ab le to set aside their stereotypes and see the clients as in dividuals, rather than only as member s of one gender group or the other. Because the Cohen’s d of the ANOVA examining differenc es in OAS ratings between male clients and female clients was so small (.053), the e ffect of client gender in this sample can be characterized as nonexistent. Failures to support the alternate hypotheses ca n occur for two reasons: (1) the lack of a statistically significant hypothe sized relationship, or (2) fl aws in the study testing the hypothesized relationship. A reasonabl e place to begin the analysis of potential study flaws that could have obscured an effect is with the reliability of the depe ndent measure, in this case the OAS. The potential concern is that perhaps the OAS was not sufficiently reliable to detect an

PAGE 32

32 effect. With a Cronbach’s of .91 in the current sample, the OAS was remarkably reliable and therefore most likely would have detected an effect if it were present. This was similar to and slightly higher than those re ported by Haviland et al. (2000; = .88 and .89). A second possible source for this lack of effect could be problems in the voice transformations; perhaps differences other than gender were created during the voice transformations. This possibility, too, seems unlik ely. Male and female conditions were the same length, had the same therapist phrases in exac tly the same places, had the same dialogue, and most importantly had the same emotional content. In other words, the respective male and female conditions were exactly alike except for the cruc ial variable, gender. Th e gender transformation process was believable to the point that only four out of the orig inal 282 particip ants indicated that they thought gender cues of the client’s voice had been altered. One of these four participants said this because he or she listened to two different cl ips, and another one of the four participants thought the gender cu es might have been modified for the original female voice (though they had not been). In other words, only 2 participants correctly identified that the gender cues of the client’s voice had been altere d. Less than 2% of participants questioned the believability of this transforma tion process, and those participan ts were excluded from analyses. So believability differences most likely cannot a ccount for the lack of statistical significance. Hypothesis 2: Stereotype Level Would Predict Client Assessment The results of therapists’ BA E scores show that overall they endorsed strong gender stereotypes about emotionality, namely that women are more emotional than men. However, BAE scores were not a significant predictor of OAS ratings. In other words, though therapists endorsed strong gender-based emotional stereot ypes, these stereotypes did not bias their assessments of client emotionality in this study. This finding also supports the earlier assertion that therapists were able to set aside their gender stereotypes and see client s as individuals, not as

PAGE 33

33 only a man or a woman. A Cronbach’s in the current sample of .83 for the BAE showed remarkably high reliability for an 8item measure, even higher than the s reported at .71 and .76 in Heesacker et al. (1999). Furthe r, the OAS had a mean of 49.08 ( SD = 12.50), while the BAE had a mean of 27.05 ( SD = 5.60). Thus, both measures had su fficient variability to perform a regression of BAE on OAS, and if there were an effect of gender ster eotyping, it would have come out significant, especially with a sample size of 248 and high reliability for both scales. The significant finding that female therapists endorse stronger gender stereotypes than male therapists do might be due to the large sample size because the Cohen’s d of .272 can be labeled “small” according to Cohen’s (1988) standards. Therefore, though the fi nding was statistically significant, its practical significance is dubious. Subscale Analyses Haviland et al. (2000) found in their graduate and undergraduate samples that women received higher OAS ratings than men on the to tal score and on the Uninsightful, Somatizing, and Humorless subscales, while men received higher OAS ratings on the Distant subscale. The current study did not find any differences betw een male and female clients on emotionality ratings for any subscale. One key difference is th at Haviland et al. (2000) asked participants to rate someone (other than themselves) whom they knew well, while participants in the current study were asked to rate a specific client. A plausible explanation is that participants in Haviland et al. (2000) rated the most inte resting people they c ould think of, or those who conformed least to their respective gender stereotypes. Hence, wo men were rated as less emotional than men on 3 of the 5 subscales. In the current paradigm, variability between pe ople being assessed was eliminated, as the only difference between them was gender. Just like the overall OAS ratings, therapists’ subscale ratings of clients were fr ee of bias from their gender-based emotional stereotypes.

PAGE 34

34 Limitations and Future Directions The biggest limitation to the current paradigm is its use of actors instead of a real client and therapist. Though the research ers made every attempt to make the session as realistic as possible by using a script of a r eal therapy session and hiring pr ofessional actors and a director, 70 out of 248 therapists indicated that they thought the session wa s contrived. This finding is not surprising because participants in this study had on average over 15 years experience seeing clients, so an actor’s ability to deliver a real istic performance, no matter how well, is limited. However, many therapists interpre ted the script-like nature of the client’s dialogue to be a manifestation of alexithymia. For instance, one ther apist wrote, “To me, the client has restricted ability to express emotions. In fact, he sounds like an actor read ing a script. He intellectualizes, makes clich comments, has isolated his affect w ith minimizing.” This script-like nature of the session was problematic in the sense that partic ipants who indicated they thought the session was contrived rated clients as less emotional than did participants who did not think it was contrived. A possible explanation for this difference is th at once participants decided the session was scripted, they focused on the actor’s emotional di stance from what he or she was reading (or the artificiality of performance), inte rpreting it just as the therapist above did, as an indicator of alexithymia. However, when these 70 participan ts who failed the manipulation check were excluded from analyses, the overa ll regression of client gender, transformation, and BAE score on OAS ratings was still not signi ficant. Manipulation check failure and the resulting therapist impressions of session artificiality most likely cannot account for absence of gender effects. Taken together, the results from this study s how that therapists’ gender-based emotional stereotypes, though strong, did not bias their assessments of client emotionality. These data are consistent with the notion that once therapists get to know an i ndividual client, their general gender stereotypes give way to seei ng the client as an individual . Some therapists in this study

PAGE 35

35 directly voiced the fact that they endorse overall gender-based emotional stereotypes, but they nonetheless saw the clients for w ho they really were. For exampl e, one therapist wrote, “The client seems to be more expressive than many men experiencing the stress that he is experiencing. He is trying to deal with his em otions thoughtfully. He is optimistic and forward looking.” Another wrote, “Above average compared to the male patients I usually work with. He appears to numb out with some frequency, and a lternatively to externaliz e blame. He is very sensitive to the hurts he feels are inflicted on hi m by others.” Clearly, th ese two therapists think that men on the whole have a hard time expressi ng emotions but that this client is good at expressing emotions, despite the fact that he is a male. To increase the mundane realism of this study, other studies should be undertaken that use real clients with a similar method. In the lite rature, there is a gap between the type of experimental manipulation employed in this study and the real pr ocess of counse ling, especially in terms of examining gender ster eotyping. This study rep licated previous findings that therapists endorse the stereotype that women are more emotional than men, so it is important to continue to examine whether or not this stereotype plays out in working with clients. It is positive news for the profession that the stereotype stopped short of biasing client assessment in this study, but there are other areas in the ther apeutic relationship where it can play out such as the clienttherapist interaction. Clearly, more research in this area is need ed, and this study highlights that need.

PAGE 36

36 APPENDIX A THIRTY-THREE ITEM OBSERV ER ALEXITHYMIA SCALE 1. Is a warm person.* 2. Has compassion.* 3. Is good at relationships.* 4. Is sensitive to other people.* 5. Likes to be close to people.* 6. Likes to have close friends.* 7. Likes to explore his or her feelings.* 8. Likes to touch or be touched.* 9. Is flexible.* 10. Is imaginative; creative.* 11. Falls apart when th ings are really tough. 12. Becomes confused when th ings are very stressful. 13. Becomes frustrated in the face of uncertainty. 14. Has strong emotions that he or she cannot explain. 15. Has trouble finding the right words to describe his or her feelings. 16. Seems to lack a sense of purpose. 17. Knows himself or herself well.* 18. Understands his or her needs very well.* 19. Worries too much about his or her health. 20. Talks a lot about physical pain or discomfort. 21. Spends a lot of time worrying about his or her body. 22. Has physical problems that are hard to treat. 23. Has physical reactions to stress (headaches, sweating, stomach problems, muscle pain). 24. Has a good sense of humor.* 25. Tells jokes and makes funny remarks.* 26. Is playful.* 27. Is unexciting; boring. 28. Is colorless; uninteresting. 29. Is too self-controlled. 30. Must “go by the book.” 31. Is stiff; rigid. 32. Sees things only as black or white. 33. Puts off enjoying the good things in life, even when it is not necessary to do so. Note : Items marked with an “*” are reverse scored.

PAGE 37

37 APPENDIX B EIGHT ITEM BELIEFS ABOUT EMOTIONS (BAE) SCALE 1. When it comes to emotion, men and women are quite different. 2. Men don't express their emotions very much. 3. Women have more awareness th an men of their own emotions. 4. Women are better at expre ssing their emotions than men. 5. Men are afraid of their feelings. 6. Men don't connect their emotions to sex as much as women do. 7. Female counseling clients usually don't need as much work as men on expressing their emotions. 8. Men rely on intellectualization more than wo men do to cope with threatening feelings.

PAGE 38

38 REFERENCES Bagby, R. M., Taylor, G. J., & Parker, J. D. A. (1994). The twenty-item Toronto alexithymia scale, II: convergent, discrimi nant, and concurrent validity. Journal of Psychosomatic Research, 38, 33. Birditt, K. S., & Fingerman, K. L. (2003). Age a nd gender differences in adults’ descriptions of emotional reactions to interpersonal problems. Journal of Gerontology: Psychological Sciences, 58B, 237. Broverman, I. K., Broverman, D. M., Clarkson, F. E., Rosenkrantz, P. S., & Vogel, S. R. (1970). Sex-role stereotypes and clinical judgments of mental health. Journal of Consulting and Clinical Psychology, 34, 1. Canary, D. J., & Hause, K. S. (1993). Is ther e any reason to research sex differences in communication? Communication Quarterly, 41, 129. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. DiGiusseppe, R., & Linscott, J. (1993). Philoso phical differences among cognitive behavioral therapists: Rationalism, constructivism, or both? Journal of Cognitive Psychotherapy, 7, 117. Fischer, P. C., Smith, R. J., Leonard, E., Fuqua, D. R., Campbell, J. L., & Masters, M. A. (1993). Sex differences on affective dimensions: Continuing examination. Journal of Counseling and Development, 71, 440. Gosling, S. D., Vazire, S., Shrivastava, S., & John, O. P. (2004). Should we trust web-based studies? A comparative analysis of six pr econceptions about internet questionnaires. American Psychologist, 59, 93. Gray, J. (2004). Men are from Mars, women are from Venus . New York: Harper Collins. Grossman, M., & Wood, W. (1993) . Sex differences in intensity of emotional experience: A social role interpretation. Journal of Personality and Social Psychology, 65, 1010. Haviland, M. G., Warren, W. L., & Riggs, M. L. (2000). An observer scale to measure alexithymia. Psychosomatics, 41, 385. Haviland, M. G., Warren, W. L., Riggs, M. L., & N itch, S. R. (2002). Concurrent validity of two observer-rated alexithymia measures. Psychosomatics, 43, 472. Heesacker, M., Wester, S. R., Vogel, D. L., Wentzel, J. T., Mejia-Millan, C. M., & Goodholm, Jr., C. R. (1999). Gender-bas ed emotional stereotyping. Journal of Counseling Psychology, 46, 483.

PAGE 39

39 Kawahara H., Masuda-Katsuse I., de Cheveign, A. (1999). Restructuring speech representations using a pitch-adaptive time-frequency smoot hing and an instantaneous-frequency-based F0 extraction: Possible role of a repetitive structure in sounds. Speech Comm., 27 , 187– 207. Kring, A. M., & Gordon, A. H. (1998). Sex differe nces in emotion: expression, experience, and physiology. Journal of Personality and Social Psychology, 74, 686. LaFrance, M., & Banaji, M. (1992). Toward a reconsideration of the gender-emotion relationship. In M. S. Clark (Ed.), Review of personality and social psychology (Vol. 14, pp.178). Newbury Park, CA: Sage. Leuptow, L. B., Garovich-Szabo, L., & Leuptow , M. B. (2001). Social change and the persistence of sex typing: 1974. Social Forces, 80, 1-35. Mallinckrodt, B., King, J. L., & Coble, H. M. (1998). Family dysfunction, alexithymia, and client attachment to therapist. Journal of Counseling Psychology, 45, 497. O'Neil, J. M., Helms, B., Gable, R., David, L ., & Wrightsman, L. (1986). Gender role conflict scale: College men's fear of femininity. Sex Roles, 14, 335. Ritter, P., Lorig, K., Laurent, D ., & Matthews, K. (2004). Internet versus mailed questionnaires: A randomized comparison. Journal of Internet Medical Research, 6, No Pagination Specified. “Stereotype.” Merriam-Webster’s Collegiate Dictionary. 10th ed. 1998. Vingergoets, A. J., & Van Heck, G. L. (1990). Gender, coping and psychosomatic symptoms. Psychological Medicine, 20, 125. Vogel, D. L., Wester, S. R., Heesacker, M., & Madon, S. (2003). Confirming gender stereotypes: A social role perspective. Sex Roles, 48, 519. Wester, S. R., Vogel, D. L., Pressly, P. K., & Heesacker, M. (2002). Sex di fferences in emotion: A critical review of the literature and implications for counseling psychology. The Counseling Psychologist, 30 ( 4), 630. Wilkins, B. M., & Anderson, P.A. (1991). Gender differences and similarities in management communication: A meta-analysis. Management Communication Quarterly, 5, 6.

PAGE 40

40 BIOGRAPHICAL SKETCH Paul Perrin was born on April 24, 1983 in Green brae, CA. He moved to Tampa, Florida when he was eleven years old, where he atte nded Jesuit High School. He graduated from the University of Florida in 2005, where he received his B.S. in psychology and B.A. in English. He completed his Master of Science Degree in Psychology at the Univ ersity of Florida and is now pursuing his Doctorate of Philosophy in Counseling Psychology at the Univer sity of Florida.