(.52 T
1.... 5 - . rgug
weir; .rt. 1 .lluhh :. {fit};
lath": .K: $0.51.:
54.”. .W. J. .u”
a... r noun... 4?? 1.: .51....1.
was”??? .5 .f a ... é .
0

n .129

03 x): ;
1-11.“. III:

.. rah?
. .l .
L. . {ihfisa:19...v§:i... 1..
.. 5...." .5... v {1.1.}..-
1 .....n.:...:. .51..

V . Rhys...
.198... .

‘1‘:
§o’;;v... . ‘
:.

V
3‘. a $9 .
.. .. 2.... 1:32.
kin... w... 4.... :11 {a
03.!!- !“ b !.lL.—..O.IG
3.2.3.. 3.1:. 1...... v
V , 0x21. :1 4‘529153
‘ I‘IIL4 I 35...; c.
3-;— g‘s". t n.
. A s 2‘. ”N‘s!
!
..... .. ,
., 4f 5
i

iovna

1.1351511! .

1-5 11.20 I

.5 2‘3.
its)!”
‘1:-

.Snz$nxnu|f&.

_ 11:.

n » .ex.
5:). t 6!!

3:.
ur ..

n.3, ,
.2 .4
‘1‘]: z.

 

 

 

:

”, lifl). 515:). .n , .

...; ,. .3... a . n .. .

5.4.51.5 ,. “.2 “$13.33:
3305.3.

 

 

:7." I run A my
I A Michigan State
University

 

 

This is to certify that the
dissertation entitled

A PILOT STUDY TO DEVELOP THE GAY INDIVIDUAL
IDENTITY — SYNTHESIS (GIISS) AND THE GAY GROUP
MEMBERSHIP — SYNTHESIS (GGMSS) SCALES FOR GAY
MEN BASED ON MCCARN AND FASSINGER'S (1996)
SEXUAL MINORITY IDENTITY FORMATION MODEL

presented by
ARIEL l. AGOSTO-CEPEDA

has been accepted towards fulfillment
of the requirements for the

_h_P.D degree In Counselin Ps cholo

///4/

/ Major Professor’ 5 Signature
,A/{A/ 4% gay
Date

 

MSU is an Affirmative Action/Equal Opportunity Employer

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KIProj/Aod-Pres/ClRC/Datebue.indd

 

A PILOT STUDY TO DEVELOP THE GAY INDIVIDUAL IDENTITY — SYNTHESIS
(GIISS) AND THE GAY GROUP MEMBERSHIP — SYNTHESIS (GGMSS) SCALES
FOR GAY MEN BASED ON MCCARN AND FASSINGER’S (1996) SEXUAL
MINORITY IDENTITY FORMATION MODEL
By

Ariel I. Agosto-Cepeda

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of

DOCTOR OF PHILOSPOHY
Counseling Psychology

2009

UMI Number: 3435274

All rights reserved

INFORMATION TO ALL USERS
The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if material had to be removed,
a note will indicate the deletion.

UMI
A
Dissertation Publishing
UMI 3435274
Copyright 2010 by ProQuest LLC.

All rights reserved. This edition of the work is protected against
unauthorized copying under Title 17, United States Code.

Pro uest’

 

ProQuest LLC
789 East Eisenhower Parkway
PO. Box 1346
Ann Arbor, MI 48106-1346

ABSTRACT
A PILOT STUDY To DEVELOP THE GAY INDIVIDUAL IDENTITY — SYNTHESIS
(GIISS) AND THE GAY GROUP MEMBERSHIP — SYNTHESIS (GGMSS) SCALES
FOR GAY MEN BASED ON MCCARN AND FASSINGER’S (1996) SEXUAL
MINORITY IDENTITY FORMATION MODEL
By

Ariel I. Agosto-Cepeda

The LGBT community comprises 10% to 15% of the overall population. Mental
health and research fields have held predominantly negative attitudes toward same-sex
feelings and behaviors during the first 70 years of the 20th century. Since the 70’s up until
nowadays research findings have emerged suggesting a positive relationship between an
individual’s acceptance of same-sex feelings and behaviors and psychological
adjustment. As a result of the shift in attitudes towards gayness several developmental
models of gay identity have emerged. The purpose of this study was to empirically test
McCarn and Fassinger’s Sexual Minority Identity Formation Model through piloting two
scales: Gay Individual Identity — Synthesis Scale (GIISS) and the Gay Group

Membership - Synthesis Scale (GGMSS).

Participants consisted of 94 gay men recruited through e-mails sent to listservs of gay
men through different associations and groups in the United States and advertisements of
the study posted in LGBT chat rooms. In this web-based study, participants were asked to
complete instruments measuring gay identity, self-concept, ethnic identity and
psychological well being. It was postulated that gay identity was a bidimensional
construct comprising individual gay identity and gay group membership. Significant

relationships were expected between gay identity, ethnic identity and psychological well

being. No significant relationship was expected between gay identity and self-concept.
The findings did not support the bidimensionality of gay identity as a construct.
Participants did not exhibit any significant changes in self-concept based on their level of
gay identity. The level of gay identity did not influence the development of an ethnic
identity among the sample. Participants with higher levels of gay identity development
appeared to experience lower levels of psychological well being. Implications of the
findings, limitations of the study as well as future directions for research and practice are

discussed.

Copyrighted by

Ariel I. Agosto-Cepeda

2010

DEDICATION

In loving memory of my mother, Cannen A. Cepeda, who instilled in me the
hunger for knowledge, the importance of persistence, and the value of high moral

standards, for showing me what really matters in life, for her infinite love and support.

I dedicate this project to my father, Angel L. Agosto, who showed me what it
means to have a perfect work ethic without neglecting his parenting responsibilities, for

being present in every major step of my life, for his undying love and support.

ACKNOWLEDMENTS

I want to express my deepest gratitude to each member of my dissertation
committee. Thank you Dr. Nancy Crewe for stepping in at the end to help complete this
project, without your help and support I would have never completed my degree. Thank
you Dr. Gloria Smith, my Program Advisor, for all your support and wisdom, you were a
sympathetic ear whenever I needed one. Thank you Dr. Robert Fabiano, for being a
caring supervisor and a wonderful mentor, my experience at PAR allowed me to gain
confidence in my assessment skills. Thank you Dr. Lee June, for being accessible

regardless of your busy schedule, your comments were always very helpful.

Thank you Dr. Victor Bonilla, for providing excellent consultation on research

design matters.

Thank you Ms. Alba Brugueras, for your immense help with the statistical

analyses and the tutoring!

Special thanks to the Lontz family and the Wallach family, for your love and

support, for making me feel at home in East Lansing, MI and Chicago, IL.

To all my friends, for making my life better.

To my family for making my life interesting and chaotic, for providing me with

wonderful stories to tell. I love you deeply.

vi

TABLE OF CONTENTS

List of Tables .......................................................................................... x
List of Figures ....................................................................................... xiii
Introduction ............................................................................................ 1
Chapter I — Literature Review ........................................................................ 5
Brief Historical Background of Same-Sex Behavior:
From Homosexuality to Gayness ........................................................... 5
Developmental Models of Gay Identity ................................................... 7
Cass’s (1979) homosexual identity formation model ........................... 8
Empirical support for the Cass homosexuality identity
formation model .................................................................... 15
Limitations of previous models of gay identity development .............. 17
Mch and F assinger’s (1996) sexual minority
identity formation model .......................................................... 18
Empirical support for the McCarn and Fassinger’s (1996)
sexual minority identity formation model ...................................... 23
Model Specification ........................................................................ 24
Gay identity and self-concept .................................................... 25
Gay identity and psychological well-being ..................................... 26
Gay identity and racial identity .................................................. 28
Research Questions and Hypotheses ..................................................... 30
Chapter II - Methodology .......................................................................... 33
Instrumentation .............................................................................. 33
Universe of potential indicators .................................................. 33
Content sampling and substantive sampling ................................. 37
Item Development .......................................................................... 41
Expert review revision ............................................................ 42
Data Collection ............................................................................. 46
Sample .............................................................................. 46
Procedures ......................................................................... 47
Instruments ......................................................................... 49
Data Analyses ............................................................................... 54
Chapter III — Results ................................................................................ 62
Demographic Data .................................................................... ‘ ...... 62
Age .................................................................................. 62
Sexual orientation ................................................................. 62
Language ........................................................................... 62
Ethnicity ............................................................................ 63
Religion ............................................................................ 63
Education .......................................................................... 63

vii

Occupation ......................................................................... 63

Partnership status ................................................................. 64
Dating ................................................................................ 65
Disclosure of sexual orientation .................................................. 65

Research Question #1: Are the items reliable and precise
enough to grant further development of the items for research

and clinical uses? ............................................................................ 66
GIISS and GGMSS .............................................................. 67
GIQ ................................................................................. 68
Six-F actor Self-Concept Measure ............................................... 69
Scales of Psychological Well-Being ............................................ 70
Multigroup Ethnic Identity Measure ........................................... 71

Research Question #2: Is gay identity a bidimensional
construct that comprises individual identity formation

and group membership formation? ...................................................... 71
Exploratory factor analyses series with 20 items .............................. 74
Exploratory factor analyses series with
Miller & Fassinger’s (1996) 10 Items .......................................... 86
Exploratory factor analyses series with
10 items developed by researcher .............................................. 95
Unidimensionality vs. bidimensionality of gay identity ................... 102

Research Question #3: Do the responses to the items
correlate with other constructs in a manner that is consistent

with the external model? ................................................................. 104
First Step .......................................................................... 105
Second Step ..................................................................... 106
Third Step ........................................................................ 111
Fourth Step ...................................................................... 116
Fifth Step ......................................................................... 121
Sixth Step ........................................................................ 125
Seventh Step ..................................................................... 129
Eighth Step ...................................................................... 133

Research Question #4: Do the correlation between the
responses to the items and other constructs exhibit

significant differences based on ethnicity? ............................................ 137
Chapter IV — Discussion .......................................................................... 139
lrnplications ............................................................................... 144
Limitations ................................................................................ 145
Future Directions .......................................................................... 146
Appendices ......................................................................................... 1 48
Appendix A: Gay Individual Identity Scale - Synthesis and
Gay Group Membership Scale - Synthesis ............................................. 148
Appendix B: Informed Consent ........................................................ 152

viii

References ........................................................................................... l 54

LIST OF TABLES

Table I. Potential Indicators of Gay Identity ..................................................... 38
Table 11. Relative Weight by Indicator in GIISS and GGMSS ................................. 40
Table III. Number of Items in GIISS and GGMSS ............................................... 42
Table IV. Expert Review Revisions of GIISS and GGMSS ..................................... 44
Table V. GIISS and GGMSS Item Placement Per Indicator ................................... 45
Table VI. Measurement Instruments Per Latent Variable ...................................... 54
Table VII. Distribution of Occupations by Major Occupational Group ..................... 64
Table VIII. Percentage of Missing Cases Per Instrument ....................................... 67
Table IX. Reliability of GIISS and GGMSS ...................................................... 68
Table X. Reliability of GIQ ......................................................................... 68
Table XI. Reliability of Six-F actor Self-Concept Measure ..................................... 69
Table XII. Reliability of the Scales of Psychological Well—Being ............................. 70
Table XIII. Reliability of Multigroup Ethnic Identity Measure ................................ 71

Table XIV. Factor Loadings for Exploratory Factor Analyses of GIISS
and GGMSS 20 items, Non-Imputed and Imputed Data ............................... 75

Table XV. Factor Loadings for 2-F actor Exploratory Factor Analyses
of GIISS and GGMSS 20 Items, Non-Imputed and Imputed Data .................... 77

Table XVI. Factor Loadings for Exploratory Factor Analyses With
Varimax Rotation of GIISS and GGMSS 20 items ...................................... 79

Table XVII. Factor Loadings for Exploratory Factor Analyses With
Oblimin Rotation of GIISS and GGMSS 20 items ...................................... 81

Table XVIII. Factor Loadings for 2—Factor Exploratory Analyses With
Varimax Rotation of GIISS and GGMSS 20 items ..................................... 83

Table XIX. Factor Loadings for 2-F actor Exploratory Analyses With
Oblimin Rotation of GIISS and GGMSS 20 items ...................................... 85

Table XX. Factor Loadings for Exploratory Factor Analyses of
McCarn & Miller 's (1995) 1 0 items from GIISS and GGMSS ........................ 87

Table XXI. Factor Loadings for 2-F actor Exploratory Factor Analyses of
McCarn &. Miller ’s (I 995) 10 items fi'om GIISS and GGMSS ........................ 88

Table XXII. Factor Loadings for Exploratory Analyses With
Varimax Rotation of McCarn & Miller ’s (1995) 10 items from
GIISS and GGMSS ........................................................................... 90

Table XXIII. Factor Loadings for Exploratory Analyses With
Oblimin Rotation of McCarn & Miller ’s (1995) I 0 items from
GIISS and GGMSS ........................................................................... 91

Table XXIV. Factor Loadings for 2-F actor Exploratory Analyses With
Varimax Rotation of McCarn & Miller ’s (1995) 10 items fiom
GIISS and GGMSS ......................................................................... 93

Table XXV. Factor Loadings for 2-F actor Exploratory Analyses With
Oblimin Rotation of McCarn & Miller ’s (1995) 10 items fiom
GIISS and GGMSS .......................................................................... 94

Table XXVI. Factor Loadings for Exploratory Factor Analysis of
Agosto ’s (2004) I 0 items from GIISS and GGMSS .................................... 96

Table XXVII. Factor Loadings for 2-F actor Exploratory Factor Analyses
of Agosto ’s (2004) I 0 items from GIISS and GGMSS ................................. 97

Table XXVIII. Factor Loadings for Exploratory Factor Analysis With
Varimax Rotation of Agosto ’s (2004) I 0 items from GIISS and GGMSS. . . . ..98

Table XXIX. Factor Loadings for Exploratory Factor Analysis With
Oblimin Rotation of Agosto ’s (2004) I 0 items from GIISS and GGMSS. . . . ..99

Table XXX. Factor Loadings for 2-F actor Exploratory Factor Analyses
With Varimax Rotation of Agosto ’s (2004) 10 items fiom
GIISS and GGMSS ........................................................................ 100

Table XXXI. Factor Loadings for 2-F actor Exploratory Factor Analyses
With Oblimin Rotation of Agosto ’s (2004) 10 items from

GIISS and GGMSS ........................................................................ 102
Table XXXII. Parameter estimates of modified model I ...................................... 109
Table XXXIII. Parameter estimates of modified model 11 .................................... 1 14

xi

Table XXXIV. Parameter estimates of modified model 111 ................................... 119

Table XXXV. Parameter estimates of modified model I V .................................... 124
Table XXXVI. Parameter estimates of modified model V .................................... 128
Table XXXVII. Parameter estimates of modified model VI .................................. 132
Table XXXVIII. Parameter estimates of modified model VII ................................ 136

Table XXXIX. ANO VA of Ethnic Identity With Gay Identity
As Dependent Variable ................................................................... 138

xii

LIST OF FIGURES

Figure 1. Developmental Stages of Cass’s Homosexual Identity Formation Model ........ 9
Figure 2. McCarn and F assinger’s Sexual Minority Identity Formation Model ............ 20
Figure 3. External Model ........................................................................... 32
Figure 4. Internal Model of the Gay Individual Identity -— Synthesis and

the Gay Group Membership — Synthesis Scales ........................................ 36
Figure 5. Full Model: Gay Identity, Self Concept, Ethnic Identity and

Psychological Well Being .................................................................. 57
Figure 6. Modified Model I ....................................................................... 108
Figure 7. Modified Model I ....................................................................... 113
Figure 8. Modified Model 111 .................................................................... 118
Figure 9. Modified Model IV ..................................................................... 123
Figure 10. Modified Model V .................................................................... 127
Figure I I . Modified Model V1 ................................................................... 131
Figure 12. Modified Model VII .................................................................. 135

xiii

INTRODUCTION

The purpose of this study was to empirically test a part of McCarn and
F assinger’s Sexual Minority Identity Formation Model. More specifically, this study
preliminarily tested the Gay Individual Identity — Synthesis (GIISS) and the Gay Group
Membership — Synthesis (GMSS) scales, each one containing 10 items purported to
measure the Synthesis phases in the Individual Identity Formation and Group
Membership processes depicted by McCarn and F assinger’s (1996) model. This is the
first phase of a larger study, the purpose of which is to develop the Gay Identity
Inventory (GH). This measure will provide information about the current phase of gay
identity development of gay men of all ages and different ethnicities who are proficient in
the English language and live in urban areas of the United States, according to McCarn
and F assinger’s model.

The lesbian, gay, bisexual and transgender (LGBT) community comprises 10% to
15% percent of the overall population (Fassinger, 1991). Same-sex behaviors and
relationships have been documented since ancient history (Scasta, 1998). However, the
last century witnessed a major shift in the societal views of gayness. Mental health and
research fields held predominantly negative attitudes toward same-sex feelings and
behaviors during the first 70 years of the twentieth century (Fassinger, 1991). As a result,
until the 1970’s, most of the psychological research and treatment focused on
homosexuality which was presumed as pathological (Garnets & Kimmel, 1993). Most of
the treatments at that time dealt with how to “converse” or “reverse” same-sex orientation
to heterosexuality (Jordan & Deluty, 1995). During the 1970’s, research findings from

studies with non-clinical samples suggested that lesbian and gay men who accept their

same-sex feelings and behaviors were more adjusted psychologically than those who try
to deny or repress their sexuality (Jordan & Deluty, 1995). Consequently, in 1973 the
American Psychiatric Association (APA) removed homosexuality as a personality
disorder from the Diagnostic and Statistical Manual, Second Edition (DSM-II). Later on,
in 1975, the American Psychological Association not only supported this decision but
also encouraged its membership to play a proactive role in removing the stigma
associated with same-sex behaviors and feelings (Fassinger, 1991; Gamets & Kimmel,
1993). Accordingly, homosexuality is currently viewed by many as one possible result of
normal development in individuals instead of a pathological condition that requires
treatment. In fact, recent findings have suggested that promoting a healthy gay identity
development in individuals leads to beneficial outcomes (Beals, Peplau & Gable, 2009;
Bieschke, Eberz, Bard & Croteau, 1998; Lark & Croteau, 1998, Phillips and Fisher,
1998; Moe, Dupuy & Laux, 2009).

As a result of this shift in attitudes toward gayness, many theorists have proposed
several models to describe the gay identity development process (Cass, 1979; Chapman
& Brannock, 1987; Coleman, 1982; McCarn & F assinger, 1996; Minton & McDonald,
1984; Sophie, 1985, 1986; Troiden, 1989). These models are particularly important
because they help the therapist to understand the specific challenges faced by this
population. However, empirical evidence to support the validity of these models is, at
best, inconclusive. To this date, only four published studies (Brady & Busse, 1994; Cass,
1984; F assinger and Miller, 1996; Halpin & Allen, 2004) have attempted to validate two
of these models (Cass, 1979; McCarn & Fassinger, 1996). Of these models, McCarn and

Fassinger’s (1996) is the most recent one. This gay identity developmental model has two

characteristics that separate it from the rest. First, it was developed considering the role of
racial and ethnic diversity in the development of gay identity. Second, rather than
conceptualizing gay identity development as a linear process consisting of fixed stages
with particular developmental tasks, the model establishes the difference between
individual gay identity formation and group membership as two reciprocal but distinct
recyclable processes. Although it has not been explicitly established by the authors, this
conceptualization suggests that gay identity might be a bidimensional construct.
Consequently, this study constitutes the first phase of the development of an instrument
to measure gay identity as a multidimensional construct considering the racial and ethnic
diversity of the members of the LGBT community in urban areas of the United States.
According to Millman and Greene’s (1993) classification scheme, testing the
synthesis phases of McCarn and Fassinger’s (1996) model through twenty items falls
within two levels of the type of inference (individual and group system) and two levels of
domain of inference (curricular and criterion). On an individual level, an easily
administered measure of gay identity development would facilitate therapists’ work in the
identification of developmental issues and needs of LGBT clients. This assessment can
be considered as a type of diagnosis (curricular inference). Moreover, it would assist
therapists in selecting efficient therapeutic interventions to foster the development of a
healthy gay identity (criterion). On a systemic level, this measure would enable further
research to test the validity of the internal and external models on which the measure is
based. Accordingly, the instrument will help the therapist to make decisions on both the
individual and systemic levels. On an individual level, the therapist can determine what

course of action is most appropriate for the client based on the information gathered

through the instrument. This course of action includes specific therapeutic interventions
as well as referrals to other therapeutic and health-related services. On a systemic level, a
valid measure of gay identity development will be helpful in identifying the relationship
between gay identity development process and other relevant psychological and health-
related constructs. For example, it is suspected that the level of gay identity development
moderates the relationship between safe-sex self-efficacy and the actual practice of safer
sex strategies in LGBT individuals. However, a lack of adequate measures has hindered
the efforts of supporting this claim with valid empirical evidence. Conclusive evidence
about this issue may impact the nature of the guidelines HIV/AIDS prevention programs

can use to target the LGBT community.

CHAPTER I
Literature Review

Brief Historical Background of Same-Sex Behavior: From Homosexuality to
Gayness

The construct of sexual orientation is a “unique historical phenomenon of the
nineteenth and twentieth centuries” (Fassinger and Miller, 1996, p.54). Accordingly, gay
identity is considered a social construction rooted in a historical, political and
sociocultural reality. Before Judaic law canonized rules against same-sex behavior
between men, other ancient cultures seemed to be more tolerant of this behavior. For
example, in both the Greek era and the Early Roman Empire, transgenerational same-sex
relationships were within the norm (Scasta, 1998). As Christianity became the state
religion in the Roman Empire in the third century AD, the tolerance of same-sex
behavior began to wane. Based on the assumption that nature was the manifestation of
one of the divine truths, the theologian Augustine (354-430 AD) and, later, St. Thomas
Aquinas (1225-1274), concluded that same-sex behaviors were “unnatural” as they were
not observed in Nature. Since all behaviors against nature were considered against God,
same-sex behaviors were codified as immoral and sinful activities (Scasta, 1998). This
view of same-sex behavior as a moral issue held its supremacy for the next six centuries.

It was not until the early 1800s that some academicians proposed an alternative
view about the etiology of same-sex behavior. Richard Krafi-Ebing, Karl Ulrichs,
Magnus Hirschfeld and Havelock Ellis advocated for organic causes of what was called
at the time “homosexuality” (Scasta, 1998). In Psychopathia Sexualis (1886), Kraft-

Ebbing argued that homosexuality was the cause of perversion while Ellis, in Sexual

Inversion (1897), argued that it was inborn. Ulrichs proposed that same-sex behavior was
the product of heredity, specifically a feminized brain, whereas Hirschfeld argued that it
was the result of glandular secretions. However, with the exception of Hirschfeld, these
academicians talked about same-sex behavior from a pathological perspective within the
disease model.

The pathological view of homosexuality was reinforced by the psychoanalytic
school, particularly the American analysts. Contrary to common belief, Freud’s views on
homosexuality were not negative. He argued that everybody is bisexual in nature and
thought that homosexuals could not change their orientation because “the condition was
based on the pleasure principle rather than on the distress of neurosis” (Scasta, 1998, p.
9). Furthermore he did not think that a homosexual identity was an impediment for
psychoanalytic training. It was the school of American analysts that rejected the idea of
bisexuality in 1940 and adopted the view of homosexuality as “thwarted heterosexuality”
(Scasta, 1998, p. 10). The psychoanalytic literature of Irving Bieber, Charles Socarides
and Joseph Nicolosi promoted the pathological view of homosexuality. Furthermore, in
the 19708 both Socarides and Nicolosi advocated against the removal of homosexuality
from the Diagnostic Statistical Manual, Second Edition.

Concurrent with the psychoanalytical literature on homosexuality, other
researchers were taking different perspectives on this issue. For example, in his study
Sexual Behavior in the Human Male (1948), Alfred Kinsey concluded that same-sex
behavior was more frequent than what the general public thought. Ford and Beach (1951)
found that 49 of the 76 societies they examined sanctioned same-sex behavior and

considered it normal. Maybe the most important contribution came from Evelyn Hooker.

In 1954 she conducted a double blind study with thirty gay men matched with 30
heterosexual men. Using three different projective tests, the blind evaluators were not
able to distinguish the two groups, concluding that the pathological findings of the
projective testing were not more severe in homosexual men than in heterosexual men
(Hooker, 1957). Based on this study, Judd Marmor stated that pathologizing
homosexuality was a way of justifying social discrimination against gays and society’s
intervention into the private lives of individuals (Scasta, 1998). While he was the
president-elect of the American Psychiatry Association, homosexuality was removed
from the DSM-II and replaced with Ego Dystonic Homosexuality. In 1986, as the current
president-elect of the APA, Paul Fink recommended the removal of the Ego Dystonic
Homosexuality diagnosis from the DSM-III entirely. Currently, instead of homosexual,
the term gay is more widely used and implies not only same—sex behavior but also the
capacity to establish intimacy within the context of a romantic relationship. Actually, one
of the consequences of the Shifi in the views of same-sex behavior is the idea that being
gay is more a matter of essence and identity than just behavior. As a result, the last forty
years have witnessed the emergence of many developmental models to explain how an
individual becomes gay.
Developmental Models of Gay Identity

With the removal of homosexuality as’a mental disorder from the DSM-II in
1973, empirical research and conceptual articles have taken new directions (Cass, 1989).
Both research and conceptual writings on homosexuality have been moving from an
emphasis on etiology and treatment programs for “psychological adjustment” to an

emphasis on describing the process of gay and lesbian identity development. Gay

identity development (or “coming out”) is the process by which the individual accepts
that “one’s primary or predominant sexual orientation is gay in the context of a
heterosexist and homophobic society” (F assinger and Miller, 1996, p.54). In other words,
it is the process by which the individual self-labels as gay (Cass, 1973). Developing a
positive gay identity, also known as “coming out”, is a lengthy developmental process
with considerable variation depending on several factors, such as gender, race, ethnicity,
social class, age and religion (Fassinger, 1991). Over the past 25 years, several models of
gay/lesbian identity development have been proposed. One of the most cited models is
Cass’s (1979) Homosexual Identity Formation Model.

Cass’s (1979) homosexual identity formation model.

Cass’s (I979) Homosexual Identity Formation Model relies on two general
assumptions. First, identity is acquired through a developmental process. Second, the
stability of and the change in behavior is the outcome of the interaction between the
individuals and their environments. More specifically, this model rests on the
interpersonal incongruency theory. The basic unit of this model is the interpersonal
matrix which consists of three elements: individual’s own perception of the self (self-
concept), individual’s perception of his/her own behavior as an outcome of the
characteristics of the self, and the individual’s perception of how other people see those
characteristics. Accordingly, the gay identity develops from one stage to the other by
means of achieving consistency between the three components of the matrix. Cass
(1984) stresses that identity, self-concept, and perceptions are cognitive constructs. Still,

her gay developmental model considers affective and behavioral components as well.

Cass (1979) views the gay identity formation as a process in which the individual
changes his/her interpersonal matrix from one that is congruently defined as non-gay to

one that is defined as gay. Cass proposes six stages for gay identity formation (see

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1).
Identity Synthesis
/
Identity Pride
Identity Acceptance
/
Identity Tolerance
/
Identity Comparison
/
Identity Confusion

 

 

 

Figure 1. Developmental Stages of Cass’s Homosexual Identity Formation Model

Identity confusion.

The first stage is identity confusion. The onset of this stage occurs when the
individual labels his/her own behaviors as “possibly homosexual”. This shifts the
previously congruent interpersonal matrix to a state of incongruency. To resolve this
confusion, the individual uses one of three approaches. If the gay behavior is perceived
both as correct and acceptable, the person begins to change his/her perceptions about the
self to make them more congruent with the new behavior. If the gay behavior is
perceived as correct but undesirable, the person tries to restore the original interpersonal

matrix by avoiding and denying all gay behavior and avoiding all sources of gay

information. Finally, if the gay behavior is perceived as both incorrect and undesirable,
the person redefines this behavior as non-gay. Of all these approaches, the first one is the
only one that forces the person into the next stage. The other two promote identity
foreclosure. Because of this particular incongruity, this stage is characterized by
confusion and turmoil as an outcome of questioning the previous perceptions regarding
the sexual identity. “I doubt that I am homosexual, but still am confused about who I am
sexually” and “I don’t act like most homosexuals do, so I doubt that I’m homosexual” are
assertions that can describe this phase (Brady & Busse, 1994).

Identity comparison.

The second stage is identity comparison. The turmoil and confusion of the first
stage have reduced considerably because the person has accepted that his/her identity is
gay. Still, with this acceptance of the self as gay, comes along a greater incongruence
between the self and the other two components of the interpersonal matrix. As a result,
the person experiences a sense of alienation. Basically, the person realizes that all the
behaviors and rules related to heterosexuality are not applicable to him/her anymore. Yet
these behaviors and rules have not been replaced by other congruent behaviors and rules.
In this stage there are four approaches to dealing with the feelings of alienation. One of
these approaches consists of devaluing the importance of others in order to lessen the
impact of this component and, thus, reducing the incongruency. At the same time, the
individual continues to present a public image of non-gay behaviors in order to avoid
others’ negative evaluation of his/her gay self (passing strategy). According to Cass
(1979), there are four ways of passing: avoiding threatening situations, controlling

personal information, deliberately presenting a non-gay image and role distancing

10

(detachment from gay situations). Individuals that adopt this approach usually act
positively on the fact that they are different and perceive the related self and behavioral
components as desirable. Accordingly, this approach helps the person to move to the
next stage.

The second approach occurs when the individual accepts the meaning of his/her
behavior as gay but finds it undesirable as a self-image. Accordingly, the person will try
to change his/her perceptions about the self without trying to change the related
behaviors. In this particular case there are four possible strategies: restructuring the
meaning of the behavior as gay only in relationship to a particular person (special case
strategy), perceiving the self as both gay and non-gay (ambisexual strategy), accepting
the gay self-image as temporary (temporary identity strategy), and accepting the gay self-
image but refusing any responsibility for it (personal innocence strategy).

The third approach consists of accepting his gay self and the meaning of behavior
as gay but, because of fear of social alienation, sees this behavior as undesirable.
Accordingly, the person implements strategies to change the behavior: inhibition of all
gay and non-gay behaviors and reducing the impact of others by social distance. Finally,
the fourth approach occurs when the person perceives both the gay self and behavior
components of the interpersonal matrix as undesirable and tries to change them both.
Cass states that this is done by the inhibition of all gay behaviors, the devaluation of
being gay and the positive view of heterosexuality. This promotes a greater sense of
alienation that, in some extreme cases, could end up with suicide. These approaches

stagnate the developmental process and promote identity foreclosure. “1 probably am

11

sexually attracted to men and women” and “I don’t feel as if I’m heterosexual or
homosexual” are statements that can describe this stage (Brady & Busse, 1994)

Identity tolerance.

During this stage the person experiences more commitment to his/her gay self and
behaviors. In order to deal with an increasing sense of alienation, the person seeks out
other gays and the gay culture. Still, for the person this is viewed as something that
needs to be done instead of something that he/she really wants to do. Cass (1979)
explains that the individual tolerates rather than accepts his self and behaviors. These
contacts with other gays can have positive or negative effects in the identity development.
Then again, to understand these effects a distinction needs to be made between those who
perceive both his/her gay self and behaviors as desirable components of their
interpersonal matrixes and those who perceive the behaviors as desirable but not the gay
self. The last group may still engage in using the strategies from the previous stage in
order to make the gay-self component of the interpersonal matrix more desirable.
However, rewarding contacts with gays that accept both his gay self and behaviors may
challenge this perspective, moving the person toward a reevaluation of previously held
negative perceptions of the gay self. This may lead to more commitment to a gay self.
On the other hand, unrewarding contacts with other gays may lead to devaluation of the
gay culture. In turn, this leads to reduction of contact with other gays and inhibition of all
gay behaviors and further identity foreclosure. Through “mixing with the gay culture”
(p.231) both groups have the opportunity to be in contact with some positive features as,
for example, meeting a partner, a role model who represents gayness as acceptable, learn

effective coping techniques to manage the gay identity, and practice of social skills

12

within a support group. If this is the case, by the end of this stage, commitment to a gay
self-image has increased. “”I’m probably homosexual, even though I maintain a
heterosexual image in both my personal and public life” are statements that describe this
stage (Brady & Busse, 1994).

Identity acceptance.

The main feature of this stage is the increasing contact with other gays. These
contact become more frequent because the person accepts rather than tolerates his/her gay
self-image. By this point the individual has satisfactorily answered the questions about
his/her essence (Who I am?) and belongingness. The main task that confronts the
individual in this stage has to do with the restructuring of the interpersonal environment
as a result of an increasing contact with the gay culture. At this point, the type of group
the person relates with will influence his/her progress through the remaining stages. Cass
(1979) establishes a difference between two types of groups: those that legitimize
gayness privately and publicly and those that only legitimize gayness privately. The
acceptance of the full legitimization accentuates the incongruency between how the
person sees his/her gay self and how he/she thinks the others see it. The attempt to
resolve this incongruency moves the person to the next stage. As for the partial
legitimization of gayness (privately), it creates no tension because the emphasis is on
keeping with the existing interpersonal matrix. The individual handles the situation by
trying to “fit in” both with gay and non-gay established institutions. The person may use
three strategies to deal with this incongruency: passing, limiting the contact with non-gay
people that challenges his/her gay self-image and selective disclosure of his/her self-

image to people that will keep it a secret. As a result, the matrix remains unchanged and

13

foreclosure occurs. This situation implies a low incongruency and a positively
formulated gay identity that can be satisfactory for some gays. “I live a homosexual
lifestyle at home, while at work/school I do not want others to know about my lifestyle”
and “Even though I am definitely homosexual, I have not told my family” are statements
that describe this stage.

Identity pride.

This stage is marked by the incongruency between the person’s own positive gay
self-image and society’s rejection of this concept. To deal with this incongruency, the
person uses strategies to devalue the importance of others to his/her self-image and
revalue gay others more positively. For these strategies to work, the individual
dichotomizes the world into gays and non-gays. The individual attributes gay people
credibility and sees them as significant, while seeing non-gays as lacking credibility and
significance. The individual immerses in the gay culture and mixes with gay groups that
share his/her same philosophy. Actually, the individual sees these groups as the only
source of emotional fulfillment, as opposed to the non-gay structures, which remind
him/her of non-gay values that do not fit the gay self-image. This conflict creates
feelings of anger that are dealt with through activism. Disclosure is the main strategy
during this stage. Cass (1979) points out that disclosure has two positive effects: it
creates more situations where the person’s gay identity is known and lends support to the
gay self-image and aligns person’s public identity with his/her private identity. If the
reaction to this disclosure is negative, the person reaffirms his current interpersonal
matrix and remains in the stage. On the other hand, if the person perceives positive

reactions from non-gay people it creates an incongruency that moves the person to the

14

next stage. “I have little desire to be around most heterosexuals” and “I am not about to
stay hidden as a gay for anyone” are statements that describe this stage.

Identity synthesis.

With the realization that the dichotomy of gay/non-gay world does not hold true,
the person’s incongruency is reduced. As a result, the individual’s personal and public
identities are synthesized into one self-image that receives support from the individual’s
interpersonal environment. In this stage, gayness is no more the only identity but just one
aspect of the self. “I generally feel comfortably being the only gay person in a group of
heterosexuals” and “1 am openly gay and fully integrated into heterosexual society” are
statements that describe this stage.

Empirical support for the Cass homosexuality identity formation model.

To test the validity of her model, Cass (1984) hypothesized that where individuals
were allocated into homosexual identity formation stages, those at a particular stage
would acknowledge the profile of that stage as the one that best describes their current
functioning. Cass (1984) also hypothesized that“ the degree of similarity or
correspondence between an individual’s current functioning and the various stage profiles
would decrease progressively as a firnction of the distance of all other stages (as proposed
by the model) from that to which the individual belonged” (p. 153). To test these
hypotheses, Cass (1984) developed two instruments: the Stage Allocation Measure and
the Homosexual Identity Questionnaire. The Stage Allocation Measure contains one-
paragraph descriptions of every stage in the Homosexual Formation Identity Model. This
measure enables the individual to allocate himself/herself in one of the stages by asking

him/her to choose the description that best fits his current functioning. The Homosexual

15

Formation Questionnaire is composed of 210 multiple response items. Cass (1984)
predicted how individuals at each stage of homosexual identity formation might respond
on each item. Accordingly, predicted scores for each stage were grouped together to
form six separate scoring keys. Cass’s (1984) results indicate a distinction between four
stages, instead of six. The lack of distinction between stages was most obvious between
stages 1 and 2 and stages 5 and 6. However, a discriminant analysis indicated that the six
stage groups could be adequately distinguished. To explain this inconsistency, Cass
(1984) argues that the scoring keys were unable to measure adequately the differences
between groups. Moreover, the method of stage allocation and the possibility of poor
item construction, unclear predictions and incorrect theorizing might have affected the
results (Cass, 1984).

Brady and Busse (1994) developed another questionnaire based on the Cass
Homosexual Identity Formation Model (1979). Their questionnaire contains 45 true-or-
false items that are easy to score for the purpose of identifying a subject’s stage of
homosexual identity formation. In their study, 225 male subjects answered the
questionnaire. The authors recruited this sample from a variety of sources: counseling
and discussion groups for gay men, gay professional organizations, beaches, parks and
gyms frequented by gay men and psychotherapy practice with predominantly gay male
clientele. The questionnaire contained three items used as validity checks.

The findings support two of the three hypotheses. Positive significant
. relationships were found between stage of homosexual identity formation and
psychological well-being and stage of homosexual identity formation and homosexual

adjustment. No significant relationship was found between homosexual identity

16

formation and background characteristics (age, education, income, religiosity, political
values). Regarding the validity of the model, the findings suggested a two-stage model.
The first stage consists of stages 1,2 and 3 while stage two consists of the remaining three
stages (i.e. 3, 4 and 5).

Halpin and Allen (2004) examined the relationship between gay identity
development and psychosocial well-being using the Gay Identity Questionnaire (Brady &
Busse, 1994). One of the aims of this study was to “explore an alternative approach to
gay identity development, investigating a more dimensional concept of developmental
stages” (p. 113-114). Halpin and Allen addressed the lack of subjects in the Cass’s early
developmental stages by recruiting a bigger sample of men who reported being attracted
to other men (n=425) through Internet. They grouped the participants based on the
highest score of the GIQ subscales and computed the mean age in each one of these
groups. The researchers argued that the progressive increase in the mean age from the
first developmental stage of identity confusion through the last one (identity synthesis) is
evidence of the validity of Cass’s model.

Limitations of previous models of gay identity development.

Cass’s and all the previous models have been criticized because of four major
flaws. McCarn and F assinger (1996) stress the fact that these models do not differentiate
between the development of an articulated gay identity within an articulated self and the
development of minority group membership. All the models confound these two
processes, using the first stages to describe the gay identity development and the later
stages to address issues related to group membership. An outcome of this confounding is

the assumption that public disclosure is always necessary to fully develop a healthy gay

17

identity. However this view of lack of disclosure as a developmental arrest fails to
consider the diversity of social realities gay individuals face. It is known that cultural
factors will determine the extent to which public disclosure and politicization of the gay
identity occurs (Fassinger, 1991). Moreover, in some cultures and countries disclosure is
not even an alternative, which does not necessarily mean that there is no room for
developing a healthy gay identity. Grov, Bimbi, Nanin and Parsons (2006) have noted the
importance of paying attention to the diversity of identity development among the LGBT
population as it relates to the person’s racial and ethnic background.

Another flaw of previous models has to do with its linearity. Most of these
models rest on the assumption that the development of a healthy gay identity occurs
within the context of a linear progression. This implies that developmental maturity
depends on an immutable homoerotic identification that fails to recognize the possibility
of alternative options, e. g. bisexuality (Fassinger, 1998; McCarn & Fassinger, 1996).
This is incongruent with the widely accepted notion of sexuality and sexual identity as a
flexible construct rather than a set of rigid and exclusive categories.

Finally, although these models claim to be relevant to both gay men and lesbians,
most of them fail to acknowledge the fact that there are differences between gay and
lesbian identity development. McCarn and F assinger (1996) point out that gender
differences regarding the expression of sexuality and masculine and feminine roles may
represent differences in terms of how gay men and lesbians acquire their sexual identities.

McCarn and F assinger’s (1996) sexual minority identity formation model.

In an attempt to address these three limitations, McCarn and F assinger (1996)

have proposed a model of gay identity development. Contrary to previous models (Cass,

18

1979; Troiden, 1989), the Sexual Minority Identity Formation Model (McCarn and
Fassinger, 1996) conceptualizes individual identity formation and the group membership
formation as two different processes that consist of four recyclable phases: Awareness,

Exploration, Deepening/Commitment and Internalization/Synthesis (see Figure 2 p.20).

19

 

  

$25.83“

 

 

 

 

:ozfiofixm

   

 

 

 

 

 

 

Benz—Ebony
\ wEfiEooQ

 

.Oll
.-

 

  

 

 

     
     

 

   

ammofiam
\ nogEanE—

 

 

 

Esp—3&0: 95.5

Ego—2 accustom >553 bro—:2 338m whom—53m wan F8022 .N PEEK

 
       

.....i_l mmo=083<

coup—2%....—

 

 

 

 

 

 

/

 

 

 

 

 

 

 

15'

 

28:25:80
\ wficoqooo

  

 

  

 

 

   

 

flmofiEnm
\ 833:2:85

  

 

 

 

 

£232 :6

 

$.33 .3222:

 
   

20

Awareness.

During the awareness phase the individual experiences same-sex feelings and
thoughts without self—labeling as gay. Individually, this phase starts with a general feeling
of being different. This difference is perceived by means of same-sex feelings and
thoughts that are quite different from the heterosexual norm. Still, the experience of these
feelings and thoughts may not lead to self-labeling as gay or lesbian. In terms of group
membership identity, the realization that nongay identity is not universal marks the onset
on this phase. This initial awareness causes confusion because of the new realization of
living under non-gay assumptions that do not fit the gay feelings and behaviors. Still, at
this phase there is no awareness of the oppression entailed in this difference.

Exploration.

The exploration phase involves active questioning of these feelings and behaviors
and exploration of sexual feelings and behaviors. During this phase, the individual
develops positive attitudes toward the gay community but does not self-identify with the
group. As far as the individual identity is concerned, this phase involves active
questioning of the feelings and behaviors first noticed in the prior phase. Although it may
involve exploration of sexual feelings, this does not necessarily translate to exploration of
sexual behavior. This holds particularly true for women. In terms of the group
membership, this phase is characterized by an attempt to define one’s own position in
relation to the gay reference group. The individual considers both his/her attitudes toward
other gays and the possibility of belonging to the group. It is possible during this phase
that an individual develops positive attitudes toward the gay community and yet does not

self-identify with the group.

21

Deepening/Commitment.

The deepening/commitment phase is characterized by . .the crystallization of
some choices about sexuality” (McCarn and F assinger, 1996, p.522). During this phase,
the individual begins to self-label as gay and perceives the world as a dichotomy
consisting of gay and non-gay cultures. Regarding the individual identity formation, the
exploration about the same-sex feelings and thoughts leads to choosing same-sex
relationships, other-sex relationships or both types of relationships. Accordingly, identity
and intimacy become more congruent within the person. The commitment to a specific
sexual identity poses an influence to the group identity process. A gay/lesbian group
identity is assumed with both the awareness of oppression and the consequences of this
choice. Personal involvement with the reference group best represents this commitment.
As an outcome, the individual may experience feelings ranging from excitement and
pride to rage and internal confusion resulting from the dichotomized worldviews, i.e.
gays and nongays.

Internalization/Synthesrs.

Finally, during the intemalization/synthesis phase, the individual has gone
through a process of internal clarification that enables him/her to integrate his/her
gay/lesbian/bisexual sexual identity into the overall self-concept. As a result, the
individual feels comfortable belonging to the gay/lesbian community and being one
across different societal contexts. In terms of individual identity formation, the individual
experiences higher levels of self-acceptance related to his/her feelings of desire and love
for same-sex partners. Regarding the group membership identity, during this phase the

person abandons the dichotomized view of the world. At this point, individuals evaluate

22

people on an individual basis rather than using any stereotype. During this phase, the
individual fully realizes the oppression related to his sexual identity and group
membership. This does not mean that the person politicizes his/her situation. According
to Fassinger (1996) and McCarn and Fassinger (1996), it is possible to start developing
an integrated self—concept without making any disclosure. Still, the current societal rules
promote the disclosure because of its demands on expressing our sexual identities
through the fulfillment of certain roles, e. g. marrying, having children, etc.

Empirical support for the McCarn and F assinger’s (1996) sexual minority
identity formation model.

To obtain preliminary validity of McCarn & F assinger’s (1996) Sexual Minority
Identity Formation, Fassinger and Miller (1996) conducted a Q-sort study with a sample
of 34 gay men from diverse ages, races, educational, occupational and socioeconomic
backgrounds. The Q-sort method consists in asking the sample to sort a series of items in
a specific number of predetermined categories. In this study, the participants were
instructed about the Sexual Minority Identity Formation Model (McCarn & F assinger,
1996) before sorting 40 items into the two parallel processes (individual sexual identity
and group membership identity) and the four phases (awareness, exploration,
deepening/commitment and internalization/synthesis) of the model. Besides answering a
demographic questionnaire, the participants were asked to select the 10 items that best
described them at the moment of the study. The authors proposed three hypotheses. The
first hypothesis was that the participants would be able to distinguish the two branches of
the model. The second and third hypotheses were that the participants would be able to

distinguish the four phases in each one of the branches of the model. These hypotheses

23

were tested by means of the proportion of agreement between participants and the
hypothesized model on the location of items. The results suggest strong support for the
Sexual Minority Identity Formation Model, particularly the distinction between
individual sexual identity and group membership identity. The distinction between phases
two and three in the group membership branch of the model was less clear. To explain
this finding, the authors argue that inadequate wording on some of the items and diversity
in the developmental paths taken by individuals into the gay/lesbian community.
F assinger and Miller (1996) emphasize that an easily administered measure of gay
identity development would facilitate the testing of the model, its reliability and validity
as well as the relationship of gay identity development with other phenomena.
Model Specification

The purpose of the current study is to elaborate on F assinger and Miller’s (1998)
results by testing the phases of Awareness in both the Individual Identity and Group
Membership processes. This partial testing consists in piloting two scales containing
twenty items, including the ten items used by the previous authors. The pilot study
addresses issues of validity based on Messick’s (1995) framework and uses Structural
Equation Modeling (SEM) analyses instead of the frequency analyses used in the Q-sort
study of Fassinger and Miller’s (1998). SEM’s goal is to determine the extent to which
the theoretical model is supported by sample data.” (Schumaker & Lomax, 2004, p. 2).
SEM has grown popularity over the past decades due to three major advantages. First, it
allows complex phenomena to be statistically modeled and tested. Second, it takes
measurement error into consideration while statistically analyzing the data. Third, the

maturity of its techniques increases the capacity to analyze more advanced SEM models.

24

SEM analyses follow a logical sequence of five steps: model specification, model
identification, mode estimation, model testing and model modification. Model
specification is the first step and consists of using relevant theory research and
information to develop a theoretical model. Similarly, Messick refers to the external
validity model as the relationship between the latent variable under study and other
similar (convergent validity) and dissimilar (divergent validity) constructs. The
theoretical model comprises various variables and the expected relationships among
them. Accordingly, the external model of gay identity includes three variables in the
model specification process: self—concept, psychological well-being and racial identity.
Decisions about the inclusion of these constructs in the external model were guided by
published scarce empirical evidence and solid conceptual arguments. Figure 3 (page 3)
represents the relationship of these constructs with gay identity.

Gay identity and self-concept.

Although, they may be used indistinctively, self-concept and identity are not
synonyms. Self-concept refers to “what one thinks one is like” (Troiden, 1984, p. 100).
Self-concept is the set of beliefs about one’s personal qualities and typical behavior
whereas identity refers to a “unified, purposeful aspect of self and hence is only part of
the self-concept” (Frable, 1997). A multidimensional approach defines self-concept as a
multidimensional construct that comprises six areas: likeability, task accomplishment,
power, vulnerability, gifiedness and morality (Stake, 1994). The self-concept contains
many identities the salience of which varies over time and across situations. Identity
refers to “organized sets of characteristics an individual perceives as definitely

representing the self in relation to a social situation (imagined or real)” (Troiden, 1984).

25

Troiden (1984) argues that self-concept is “broader in scope” and includes more
categories than identity, which is more limited and involves categories only relevant to
particular situations. Therefore, an individual may have multiple identities but only one
self-concept. Larson (1985) conducted a study to examine the relationships between
gender, social sex roles and sexual orientation as the three components of sexual identity
and the relationship of these components to the overall self-concept. The sample of 160
subjects answered and returned a research packet containing the Bem Sex Role Inventory
(BSRI) and the Tennessee Self Concept Scale (TSCS). Four groups were compared: gay
men , lesbians, straight men and straight women. The findings suggest an interaction in
the relationship of gender and sexual orientation to self-concept. However, sexual
orientation alone did not translate to significant differences in self-concept. These
findings support Troiden’s arguments regarding the relationship between gay identity and
self-concept. According to both theoretical arguments and the scarce research findings,
no significant relationship is expected between gay identity and self-concept. However,
for the purpose of this study, self-concept was included in the model to test its
relationship with the gay identity measures as a way of establishing divergent validity.

Gay identity and psychological well-being.

Ryff and Keyes (1995) define psychological well-being as positive functioning in
six areas: self-acceptance, positive relations with others, autonomy, environmental
mastery, purpose in life and personal growth. Self-acceptance regards the general attitude
toward the self, acknowledgement and acceptance of good and bad aspects of the self and
feelings about past life. Positive relations with others comprises the nature and number of

relationships with others, caring about the welfare of others, empathy, affection and

26

intimacy. Autonomy is characterized by self-determination and independence, ability to
resist social pressures to think and act in certain ways, behavior auto-regulation and
personal standards to evaluate the self. Environmental mastery is the degree to which an
individual can manage his environment to create contexts that satisfy his needs and
values. Purpose in life is defined by a sense of meaning in past and present life and the
presence of goals for the future. Finally, personal growth is the sense of continued
development, openness to new experiences and a sense of realizing the individual’s full
potential.

Since the late 60’s, research findings on gay identity have revealed not only that
gay individuals can be as psychologically well-adj usted as their heterosexual counterparts
but also that positive gay identity development was associated with psychological well-
being while internalized homophobia is a risk factor for depression, anxiety and suicide
(DeLuca, 1967; Dunkle, 1994; Hooker, 1965; Halpin & Allen, 2004; Igartua, Gill &
Montoro, 2003; Kertzner, Meyer & Frost, 2009; Miranda & Storms, 1989; Sagir, Robins,
Walbran & Gentry, 1970; Szymanski & Gupta, 2009). This relationship can be explained
by means of the role of each one of Ryff and Keye’s (1995) dimensions in both
psychological well-being and the healthy development of a gay identity in an individual.
McCarn and F assiger’s (1996) model explains the process by which the individual self-
labels as gay. Underlying this process is the assumption that, regarding the individual
identity developmental process, the more the individual accepts himself as a gay, the
more developed his gay identity will be. Along the same lines, one of the indicators of
group membership identity development is the nature of social relationships. The more

the individual is able to relate with people of every sexual orientation, the more

27

developed his sense of group identity is, which suggests the presence of positive
relationships with others that nurture his psychological well being. In terms of autonomy,
McCarn and F assinger (1996) describe a gay man in the Internalization/Synthesis phase
as an individual that evaluates people and situations on an individual basis rather than
using common stereotypes, which suggests the ability to be self-determining and
independent. Similarly, being comfortably gay across different social contexts suggests
theiability of making effective use of the surrounding opportunities, which is
characteristic of individuals with high levels of environmental mastery. Regarding
personal growth, McCarn and F assinger’s model implies that with the development of
gay identity comes a higher degree of self-knowledge and an openness to new
experiences that is characteristic of high levels of personal growth. Finally, it is not far-
fetched to think that the internal clarification process essential to the development of a
gay identity may result on higher levels of purpose in life, as it may give sense to past
and present life, as well as provide useful guidance for the future. Therefore, a positive
relationship is expected between gay identity and psychological well-being.

Gay identity and racial identity

Most models of gay identity development rely on the assumption that to be fully
developed, gay men and lesbians have to publicly acknowledge their sexual orientation.
This process, usually known as coming out, assumes the importance of individualism,
independent identity and separation from the family as important requisites of growing up
(Smith 1997). Along the same line, many researchers and theorists have assumed that
coming out is a desired end while suggesting that failure to come out implies resistance,

self-hatred, shame and embarrassment (Grov, Bimbi, Nanin & Parsons, 2006).

28

Consequently, not disclosing the sexual orientation is considered a sign of arrested
development. However, coming out has been described as a “White, Western, middle-
class phenomenon” (Estrada & Rutter, 2006; Smith, 1997) because it does not consider
the fact that other cultures rely more on the importance of family and interdependence as
indicators of healthy development. Factors as race and ethnicity may interact with the
coming-out process (Grov et al., 2006; Rosario, Schrimshaw & Hunter, 2004). In some
cases, disclosure of gay sexual orientation to a family member is a challenge to ethnic
minority families, who are not used to discussing sexuality matters and prefer to assume a
heterosexual orientation. For example, tolerance or acceptance of same-sex feelings has
been noted in many instances among African-American communities, however this
acceptance comes with an implicit agreement of not disclosing or displaying the sexual
orientation (Crawford, Allison, Zamboni & Soto, 2002; J amil, Harper and Fernandez,
2009). As a result, many African-American gay men lead a compartmentalized life,
monitoring what they reveal about themselves within their ethnic community while
seeking out the European-American gay community to express their gay sexual
orientation. In a case discussion, Estrada and Rutter (2006) explained the importance of
family loyalty in Puerto Rican families, the highly valued sense of family respect that
implies a willingness to sacrifice oneself for the welfare of the group and its implication
in the coming out process. For Puerto Rican families with traditional cultural norms
coming out as gay or lesbian is indeed a transgression that may be viewed as a blow to
the family reputation (Estrada & Rutter, 2006; Mohr & Fassinger, 2003). However,
ethnic minority gay LGBT individuals have reported race-based discriminatory treatment

within the latter (Adams & Kimmel, 1997; Greene, 1997; Harper, Jemewall & Zea,

29

2004). Consequently, LGBT individuals from ethnic minorities may experience
conflicting loyalties to two communities that marginalized them for different reasons
(Crawford et al., 2002; Goode-Cross & Good, 2009). The scarce research findings on the
interphase of these two identities suggest that attachment to cultural heritage and identity
as an ethnic minority are most important for emotional health and well being (Crawford
et al., 2002). Accordingly, it is expected to see differences in how individuals from
different ethnic backgrounds decide to disclose their sexual orientation. More
specifically, it is expected that White men will be more inclined to publicly disclose their
sexual orientation and participate more actively in LGBT groups, while gay men of color
identified with their own ethnicity will not necessarily engage on public disclosing of
their sexual orientation. Still, according to McCarn and F assinger’s model, the lack of
public disclosure does not necessarily mean lack of psychological adjustment. Therefore,
this researcher expects a stronger positive relationship between ethnic identity and gay
individual identity formation when compared with the relationship between ethnic
identity and gay group membership among gay men located in the Synthesis phase.
Research Questions and Hypotheses

Although the shift in the attitudes toward gayness have resulted in many
developmental models that attempt to explain gay identity, empirical findings to support
these theoretical models is at best scarce and inconclusive. Most of these models were
developed based on the experience of the Middle class White male and failed to consider
the interaction of multiple identities among people of color. Furthermore, the majority of
these models assume that gay identity is a unidimensional construct that can be

conceptualized as a series of discreet stages arranged in a specific order. The purpose of

30

this study is to pilot two scales purported to measure high gay individual identity and
group membership according to McCarn and F assinger’s Sexual Minority Identity
Formation Model. This study intends to evaluate the validity of these scales according to
Messick’s perspective on validity while using structural equation modeling statistical
analyses. Accordingly, these are the research questions and hypotheses:
1. Are the items reliable and precise enough to grant further development of the
items for research and clinical uses?
2. Is gay identity a bidimensional construct that comprises individual identity
formation and group membership formation?
a. Hypothesis 1: Significant differences will be observed between the scores
of the Gay Individual Identity — Synthesis Scale (GIISS) and Gay Group
Membership — Synthesis Scale (GGMSS).
3. Do the responses to the items correlate with other constructs in a manner that is
consistent with the external model?
a. Hypothesis 1: No significant relationship will be observed between the
scores of the Gay Individual Identity — Synthesis Scale (GIISS) and the
Gay Group Membership - Synthesis Scale (GGMSS) and the scores of the
Six-F actor Self-Concept Scale.
b. Hypothesis 2: A significant positive relationship will be observed between
the responses to the items included in the Gay Individual Identity —

Synthesis Scale and the Psychological Well Being Scales.

31

c. Hypothesis 3: A significant positive relationship will be observed between
the scores of the Gay Group Membership —- Synthesis Scale and the
Psychological Well Being Scales.

d. Hypothesis 4: Differences in the scores on the Gay Group Membership —
Synthesis Scale will be observed based on the scores in the Multigroup
Ethnic Identity Measure.

4. Do the correlations between the responses to the items and other constructs
exhibit significant differences based on ethnicity?

a. Hypothesis 1: A stronger positive relationship will be observed between
the scores of the Individual Identity — Synthesis scale and ethnic identity
when compared with the relationship between the responses to the Group
membership — Synthesis Scale and ethnic identity among gay men of color
located in the Synthesis phase.

Figure 3 represents the external model in which this research study is based.

Gay Identity _ Ethnic
Identity

 

     

Psychological
Well Being

Self-Concept

Figure 3. External Model

32

CHAPTER II
Methodology

The methodology includes the developing process of the Gay Individual Identity
— Synthesis and the Gay Group Membership — Synthesis scales as well as the data
collection procedures and the data analyses conducted to answer the research questions
and test the research hypothesis.

Instrumentation

To develop the Gay Individual Identity - Synthesis and the Gay Group
Membership - Synthesis scales, the first step was the instrumentation. The
instrumentation consists of the identification of the universe of potential indicators and
the process of content and substantive sampling from this universe of indicators.

Universe of potential indicators.

The scarce research on gay identity development has argued that gay identity
development can be observed through a definite set of dimensions. For example, Cass
(1979) proposed 16 dimensions to test her theoretical model of Homosexual Identity
Formation: commitment, disclosure, generality, identity evaluation, group identification,
social interaction, alienation, inconsistency, sexual orientation activity, acculturation,
deference to others, dichotomization, personal control, strategies, personal satisfaction
and professional contact. The main assumption is that cognitive, behavioral and affective
changes in these dimensions reveal the progression of an individual through the stages of
gay identity development. However, it is important to emphasize that most of the gay
identity models assume certain cultural homogeneity among LGBT individuals and fail to

consider the diversity within the community. Therefore, establishing a set of dimensions

33

that are too specific might leave out of the equation a significant part of the LGBT
community.

Another way of approaching this problem is focusing on the context in which gay
identity development occurs. Following other developmental models, we can assume that
gay identity development occurs within the context of the individual relationship with
himself/herself, his/her family, friends, coworkers and society at large. Accordingly, the
individual will develop certain cognitions, attitudes, feelings and behaviors that tap into
his current phase. Furthermore, this approach allows for a more flexible model of
development as opposed to restricted developmental stages. Messick’s (1993) substantive
validity refers to the observed consistencies in the responses based on theoretical process
models of task performance supported by empirical evidence. Accordingly, the current
model proposes that the construct of gay identity can be observed through self-reported
information about behaviors, feelings, cognitions and attitudes toward gays and nongays.
Therefore, the individual exhibits a cluster of feelings, behaviors, cognitions and attitudes
that are related to both the group membership and the individual identity development
processes. This is a departure from McCarn and F assinger’s (1996) model. These authors
proposed a model that separates these two processes. However, they argue that group
identity formation is observable only through attitudes of the individual toward gays and
nongays whereas the individual identity formation is observed through homoerotic desire
in the context of relational identity (i.e. who I choose as a sexual partner). Figure 3
represents the relationship between gay identity, individual identity formation, group

membership identity and their related feelings, behaviors, cognitions and attitudes.

34

To summarize, gay identity development is operationally defined as the process by which
an individual, in the context of the relationship between him and his environment,
acquires a set of cognitions, attitudes, behaviors and feelings that differentiates him from
individuals that hold and display heterosexual, or “straight”, views and behaviors. This

internal model is depicted in Figure 4.

35

mpg—mom mmmofiim I 35.5952 95.5 .30 2: can maofiim I 33:02 3.6365 >90 05 MO .322 REES .V 3wa

 

flogafim

 

 

 

 

 

 

 

 

83:2

 

      
    

28$:on

 

 

AmeUUV

9323—52
9.30 ~30

 

mwéoom

 

 

 

 

wage—Lem

 

 

 

 

 

88:3.

 

 

 

     
       

£253 :5

 

£82:on

 

 

$39
$52

3:22....— aw

 

mwczoom

 

 

 

36

Content sampling and substantive sampling.

The process of content sampling consists of identifying and choosing the contents
(indicators) based on which the items for a particular instrument will be elaborated. Both
the GISS and the GMSS assume that the Individual Identity Synthesis and Group
Membership Synthesis Formation phases of gay identity development are observable
through individual’s same-sex feelings, behaviors, cognitions and attitudes. However,
these feelings, behaviors, cognitions and attitudes are not restricted to the individual’s
romantic/sexual relationships. They also include other types of relationships: family,
fiiendships, community participation, etc. Based on this operational definition, this
researcher elaborated a list of 17 potential indicators of gay identity development (see
Table 1), based on personal experience and the stage descriptions of previous gay identity
developmental models (Cass, 1979; McCarn & Fassinger, 1996; Troiden, 1989). Table 1
contains these indicators. These indicators represent potential areas based on which
specific items about the synthesis phase in Individual Identity and Group Membership

formation could be elaborated.

37

Table 1

Potential Indicators of Gay Identity

 

Potential indicators of gay identity

 

General same-sex cognitions

General same-sex feelings

Same-sex thoughts about specific individuals
Same-sex feelings about specific individuals
Same-sex behaviors (sexual activity)
Same-sex sexual fantasies

Sexual attraction

Dating (Whom do I date?)

Intimacy issues (With whom do I prefer to establish a romantic/committed relationship?)
Self-labeling as gay

Participation in LGBT community life
Friendships with gays

Friendships with nongays

Family relationships

Attitudes toward other gays

Attitudes toward nongays

Attitudes toward the self

 

Financial and time constraints made it impossible to measure all the potential indicators.

Therefore, four indicators were selected: general same-sex cognitions, general same-sex

feelings, same-sex behaviors and attitudes. These indicators were chosen for two reasons.

First, they can encompass the rest of the indicators. Furthermore, they are congruent with
the internal model of the construct of gay identity. Secondly, these domains are
conventionally used to measure psychological constructs like identity and self-concept.
Other constructs (e.g. intelligence, academic performance) may have a greater and more
varied universe of potential indicators. However, gay identity is a construct with

psychological, historical, political and sociocultural dimensions that cannot be measured

38

by other means than collecting information about what LGBT individuals think and feel
about their sexual orientation and how this influences their same-sex-related attitudes and
behaviors.

The next step was to determine the relative weight of each selected indicator in
both Individual Identity Synthesis and Group Membership Synthesis phases. The relative
weight of each indicator guides decisions regarding the number of items per indicator to
be included in a measurement instrument. Usually these weights are expressed in terms of
the percentages of a specific indicator’s items within the total items of the instrument. In
some instances, decisions about the given weight of an indicator may reflect its
importance to the general construct. However, the published literature on gay identity
development does not establish that any indicator should be more important than the
others. Decisions about indicators’ weight seemed to be based on pragmatic issues and
tend to result in an even distribution of number of items across indicators. Regarding the
development of the Gay Individual Identity — Synthesis Scale and the Gay Group
Membership — Synthesis Scale the differences in the number of items per indicator
should not be assumed as a reflection of the relative importance of each indicator. These
differences are rather the result of pragmatic decisions. Each one of these scales includes
5 items tested in F assinger and Miller (1996) Q-sort study and 5 additional items
developed by this researcher to address the lack of representation of certain indicators in
McCarn and Fassinger’s items. The intention was to have ten items per scale. Since none
of the indicators is more important than the others, this researcher decided to give each
one of the four indicators a relative weight of 25%. However, 25% of ten items is 2.5

items, which is not possible. Therefore, this researcher decided that each indicator would

39

have at least a weight of 20%, that is, two items per indicator, taking into consideration
that some of the indicators could be represented by more than two items. Table 2
summarizes the weights of the four indicators in the Synthesis phases of Individual
Identity Formation and Group Membership Formation.

Table 11

Relative Weight by Indicator in GIISS and GGMSS

 

 

. Individual Identity — Group Membership —
Indicator Synthesis Phase Synthesis Phase
Same-sex feelings 30% 20%
Same—sex cognitions 30% 20%
Same-sex attitudes 20% - 30%
Same-sex behaviors 20% 30%

 

The table indicates that the indicators same-sex feelings and same-sex cognitions would
be represented by more items than the indicators same-sex attitudes and same-sex
behaviors in the Individual Identity F ormation-Synthesis phase, while the opposite would
be true for the Group Membership Formation — Synthesis phase.

The next step was to establish the limits of the universe of admissible evidence,
which consists in the type of observations that the items will report. The universe of
admissible evidence for gay identity is very restricted. Two general options are self-
report information (i.e. the subjects reports how he/ she feels, what he/ she thinks, what are
his/her attitudes and what he/she does) and direct observation by the researcher of
behaviors that may reflect subject’s feelings, cognitions and attitudes. Therefore, the only
admissible evidence in this case is self-report information. As it was stated earlier, self-

reported same-sex feelings, cognitions, attitudes and behaviors will describe the

40

Synthesis phases in both Gay Individual Identity and Gay Group Membership formation
processes.
Item Development

After determining the universe of potential indicators, content sampling and
substantive sampling, the next step was to create the blueprint of the GIISS and GGMSS.
To elaborate this blueprint the researcher used a matrix consisting of four rows for the
selected indicators (same-sex feelings, same-sex, cognitions, same-sex-attitudes and
same-sex behaviors) and two columns for the Gay Individual Identity Synthesis phase
and the Gay Group Membership Synthesis Phase. The blue print for the
Internalization/Synthesis phase of the model resulted in a matrix of four dimensions (i.e.
feelings, cognitions, attitudes and behaviors) per two processes (i.e. Individual Identity
Formation and Group Membership Development). This researcher placed the items tested
by Fassinger and Miller (1996) in the blueprint matrices. These items come fi'om a 97-
item pool initially developed for testing the complete model on a sample of lesbians. Two
pilot studies were conducted using female expert raters to reduce and refine the item
pool. The final item pool included 48 items. For their study with a male sample,
Fassinger and Miller used the same 97-item pool. The items were reworded for
appropriateness for gay male sample. Five male raters (mental health professionals with
expertise in LGBT issues) sorted the items in two piles representing the individual and
the group membership formation processes. Then the raters were asked to sort each pile
in four groups representing the four developmental phases (i.e. awareness, exploration,
deepening/commitment and intemalization/synthesis). Frequencies were tabulated for

placement of items in each one of the eight groups. The results were analyzed based on

41

inter-rater agreement. The final item pool included 40 items that were tested with the gay
male sample. Ten of these items were included in the current study to test the
Internalization/Synthesis phase of the model and were placed in the matrix of the GIISS
and GGMSS.

Once these items were placed in the blueprint matrices, this researcher wrote
additional items to fill in the empty cells of the indicator matrix for a total of 20 items.
According to the given weights of the indicators, this researcher determined the number
of items per indicator based on the total of ten items per scale (GIISS and GGMSS).
Table 3 presents the number of items per indicator in both the GIISS and the GGMSS.
Table 111

Number of Items in GIISS and GGMSS

 

 

 

. Gay Individual Identity — Gay Group Membership —
t . .
Indrca or Synthesrs Scale Synthesrs Scale
Same-sex Feelings 3 2
Same-sex Cogrritions 3 2
Same-sex Attitudes 2 3
Same-sex Behaviors 2 3
Total 10 10

 

To establish content and substantive validity, the written items were distributed in a way
that each indicator in each scale was represented by, at least, two items (20%).

Expert review revision.

Once the items were elaborated and placed in the blueprint, three mental health
professionals with expertise in LGBT issues reviewed these items for grammatical and
content appropriateness. They were provided with the blueprint of the GIISS and the
GGMSS as well as with a detailed description of the Gay Sexual Minority Formation

42

Model (McCarn & Fassinger, 1996). The experts were asked to identify not only
grammatical errors but also offensive language and determine if the items reflected the
phases they were purported to describe.Two items were revised to avoid grammatical
errors. One item was revised to discard offensive language because it could be read as if
sexual orientation were not important at all. Two items were deleted and substituted by

two new items to avoid psychological jargon. Table 4 summarizes these revisions.

43

Table IV

Expert Review Revisions of GIISS and GGMSS

 

 

 

 

 

 

 

 

 

Deleted Items
Scale Indicator Items
GHSS Cognitions “Being gay is part of my overall identity.”
GIISS Cognitions “Being gay is part of who I am as an individual.”
Added Items
Scale Indicator Items
GIISS Cognitions “Loving and being sexual with men is part of who I am.”
“Loving and being sexual with men is part of how I see
GIISS Cognitions myself.”
Revised items
Scale Indicator Items
“ When I meet someone I don ’t think solely of him in terms of
GGMSS Cognitions his sexual orientation. ” instead of “When I meet somebody, I
don ’t tend to think of him in terms of his sexual orientation. "
“Being gay or straight is one central part of who people
GGMSS Cognitions are. ” instead of “Being gay or heterosexual is one part of the

overall identity. "

 

Table V presents the final item placement per indicator in each subscale after all the

expert revisions were incorporated.

44

Table V

GIISS and GGMSS Item Placement Per Indicator

 

 

 

 

 

Indicator GIISS GGMSS
Same-sex “I feel a deep contentment about “I still get angry at the way
feelings my love for other men.” "‘ homosexuals get treated, but not as
much as once I did.” *
“I love and appreciate myself as a
gay man.” * “I feel comfortable interacting with
both straight and gay people.”
' “1 don’t feel guilty about my
attraction to and love for other
men.”
Same-sex “I feel proud of being gay as I am “Being gay or straight is one central
cognitions proud of other part of who people are.”
aspects/characteristics of me.”
“When I meet somebody, I don’t tend
”Loving and being sexual with to think solely of him in terms of his
men is part of who I am.” sexual orientation.”
Loving and being sexual with
men is part of how I see myself.”
Same-sex “My love for men is an important “Some heterosexuals are homophobic,
attitudes part of me but, it is not the only some are not.” "'
thing that defines me.” * “I enjoy having straight and gay
“Being in a relationship with friends.”
another man is important to me, “I think is important for me to get along
but it is not the only thing that with straight and gay people.”
defines me.”
Same-sex “I have successfully incorporated “I can relate comfortably to gays and
behaviors my love of men into my life.” * nongays.” *

“I’m doing what I want to do in
terms of love and sex; that makes
me feel more integrated as a
person.” *

“I rely on my gay/lesbian friends for
support but have some good
heterosexual fiiends as well.” *

“I am a member of the gay
community.” *

 

*Items by Miller & F assinger (1996)

45

The Gay Individual Identity — Synthesis Scale (GIISS) and the Gay Group Membership —
Synthesis Scale (GGMSS) are two of the eight subscales of the Gay Identity Inventory,
an instrument this researcher is developing. The scores of the Gay Identity Inventory will
be interpreted based on a criterion frame of reference. The purpose of the measure is to
classify individuals according to their phase of gay identity development in two different
processes: group identity formation and individual identity formation. Therefore, each
score will represent the individual’s relative position within each one of these processes.
Furthermore, theory supports the idea of gay identity development as a process that can
be unique for individuals even within the same identifiable group (e. g. gender, age
intervals, and ethnic identity). Thus the results will only show the person’s relative
position on gay individual identity and gay group membership rather than providing any
indication of how he scored relative to other people.
Data Collection

Once the revisions were incorporated in the GIISS and the GGMSS, this
researcher started the process of data collection.

Sample.

The targeted population of the study is gay men from different ethnic and race
backgrounds living in United States who are fluent in English. A sample of 110 gay men
from diverse cultural backgrounds was recruited through posts in LGBT-related web
pages, blogs and diverse web groups in three major Internet search engines. Data

collection was completed by February 2007.

46

Procedures.

This researcher used the services of the commercial online survey creator
www.createsurvey.com to develop a web-based questionnaire. Recruiting e-mails were
sent to listservs of gay men through different associations and groups in the United States
and advertisements of the study were posted in LGBT chat rooms. The e-mails and
advertisements introduced the researcher as an openly gay graduate student conducting a
study the purpose of which is to develop an instrument to measure gay identity
development. The advertisements explained that the researcher was interested in
examining how gay identity is related to how gay men think and feel about themselves. It
was further explained that this data would provide therapists and other mental health
professionals with valuable information that would foster increased sensitivity toward
their gay clients. The advertisements emphasized that participation in this study will be
held strictly confidential. Finally, it stated that participants would have the opportunity to
enter in a raffle for cash prizes and provided a link to the website for the study.

A link in the e-mails and advertisements connected to the study’s informed
consent. This consent form guaranteed that participation in the study would be entirely
voluntary, that their responses would be held confidential, and that they had the right to
withdraw from the study at any moment. The informed consent page asked the
participants to choose between agreeing to participate by proceeding to the questionnaires
pages or leaving the website. Once the participants completed the questionnaires, they
were directed to a separate website to provide their e—mail address in order to enter in a
raffle with three $50 cash prizes. The participants’ contact information was kept separate

from their responses without the possibility of any association and the raffle drawing was

47

conducted after data collection was completed. The prizewinners received e-mails stating
that they had won one of the prizes and explaining that they would receive their checks,
along with a receipt form, at the provided mailing addresses. The winners were asked to
sign the receipt form and sent it to the researcher in a pre-stamped envelope to verify that
they received their prizes. This researcher sent three emails to winners of the raffle. One
of the emails bounced back and an additional email was sent to another participant. After
two weeks and a reminding email, none of the winners reclaimed their prizes. This
researcher chose three other participants and sent the emails informing about the winning
prize. The prizes remained unclaimed.

This researcher chose the web-based design to collect data because its advantages
seemed to ensure the participation of a bigger sample. Overall, five advantages have been
noted regarding the use of web-based designs over traditional paper-and-pencil methods
(Hsu, 2005). First, the physical absence of the researcher eliminates the implied demand
for the participants to remain in the study against their true wishes. As a result, the
participants can exert more fi'eedom in withdrawing from the study. Second, the privacy
that a web-based design provides tends to decrease proneness to social desirability in the
participants. This is particularly important with the LGBT community as some of their
members and potential participants have not yet come out. This has been a problematic
issue for research on gay identity development as samples are mostly constituted by men
and women that have already engaged in some public disclosure of their sexual
orientation. Furthermore, some of the members of the LGBT community that have
already come out may find themselves sanctioning items in a way that is desirable to the

LGBT community even though that may not reflect their true attitudes, behaviors,

48

feelings or cognitions. Therefore the privacy that a web-based design provides may foster
more accurate responses from the participants. Third, difficult access to certain subjects
(e. g. gay men that have not come out of the closet yet) can be reduced through the use of
a web-based design. Fourth, the sample recruited through intemet is potentially more
representative of the population than the samples recruited at specific LGBT
organizations. F ifih, a web-based design increases the possibility of obtaining a larger
sample size, which results in the decrease of Type 11 error rate and an increased power.

Even though the web-based design has many advantages for the current study, it
is not immune to limitations. One of these limitations is that the conditions under which
the participants complete the survey are uncontrolled. They may complete the survey
while at work, between meetings or while engaging in another activity at home. Another
limitation is that the sample could be biased as it requires ownership or, at least, some
access to a computer.

Instruments.

The website contained the following instruments: A demographic questionnaire,
the Gay Individual Identity — Synthesis (GIISS) and the Gay Group Membership -
Synthesis (GMSS) scales, the Gay Identity Questionnaire (Brady and Busse, 1994), Six-
Factor Self-Concept Scale (Stake, 1994), Scales of Psychological Well-Being (Ryff,
1996) and the Multigroup Ethnic Identity Measure (Phinney, 1992). The instruments are
self-report measures that the participants can answer with little help.

Gay Individual Identity — Synthesis Scale (GIISS) and the Gay Group

Membership — Synthesis Scale (GMSS).

49

These are the instruments that were piloted. The scales are self-report measures
based on McCarn and F assinger’s (1996) Sexual Minority Identity Formation Model. The
GIISS and the GGMSS are actually subscales of the Gay Indentity Inventory (G11), 3
larger instrument under development that targets the complete McCarn and F assinger gay
identity developmental model. Each one of the scales included 10 items that represent the
Deepening/Synthesis phase in the individual gay identity and the gay group membership
proceses, respectively. To answer every item, the individual used a 4-point Likert scale
where 1 means strongly agree, 2 means agree, 3 means disagree and 4 means strongly
disagree. Once the individual completed the scales, the answers to each item in each
phase subscale are summed. Higher scores indicated that the person’s current gay identity
developmental phase is internalization/synthesis.

Gay Identity Questionnaire (Brady & Busse, 1994).

The Gay Identity Questionnaire (GIQ) is a self-report measure that is based on the
tenets of Cass’s (1979) Homosexual Identity Formation Model. The measure consists of
45 true-or-false items: 42 items used to assign individuals to one of the six stages of
Cass’s (1979) model and three items to verify that the individual has same-sex feelings,
thoughts or behavior. For each item that the individual marked true, one point was
accrued in the stage of HIF represented by the item. Items marked false were scored with
a zero. The subset of items in which the individual obtained the most points is his stage
designation. Inter-item consistency scores for the GIQ were obtained using the Kuder-
Richardson formula. For the items representing stage three (Tolerance), r = .76; for the
items representing stage four (Acceptance), r = .71; for the items representing stage five

(Pride), r = .44 and for the items representing stage six (Synthesis), r = .78. Since the

50

authors were not able to recruit enough individuals for the first two stages of the model,
interitem consistency scores were not obtained. The GIQ was included in the model as
one of the instrument variables to measure gay identity as a latent variable and was used
to establish convergent validity of the piloted scales (GIISS and GGMSS).

Six-F actor Self-Concept Measure (Stake, I 994).

The Six-F actor Self-Concept Scale (SF SCS) is a multidimensional self-report
measure of adult self-concept that has applicability across a wide range of life settings,
roles and activities (Stake, 1994). The measure consists of six subscales, each one
measuring one dimension of self-concept: Likeability, Morality, Task Accomplishment,
Gifiedness, Power and Vulnerability. The scale has a total of 36 items that are scored on a
seven-point scale that ranges from “never or almost never true for me” to “always or
almost always true for me. After six weeks, the test-retest coefficient for the composite
score was .97, whereas the coefficients for the subscales ranged from .68 (Vulnerability)
to .85 (Power and Task Accomplishment). A higher correlation of the measure with a
measure of self-esteem than with a measure of social desirability provided evidence of
convergent and discriminatory validity. Stake’s Self Concept Scales were included in the
model to establish divergent validity of the GIISS and the GGMSS. The proposed model
tested a correlation between the latent variables gay identity and self concept. Since Ryff
(1995) proposed self-concept as a multidimensional construct, the six subscales were
included as six different instrument variables of self concept as a latent variable.

Scales of Psychological Well-Being (Ryff; 1989).
The Scales of Psychological Well-Being (PSWB) are based on six dimensions

that are related to positive functioning: self-acceptance, positive relations with others,

51

autonomy, environmental mastery, purpose in life and personal growth. Each scale
consists of 14 items, equally split between positive and negative items. Items are scored
on a six-point scale ranging from strongly agree to strongly disagree. Internal consistency
coefficients for the scales range from .82 to .91 (Kaflca & Kozma, 2002; Ryff, 1989; Ryff
& Scmutte, 1997). Ryff’s Scales of Psychological Well-Being were included in the model
to establish convergent validity of the GIISS and the GGMSS. Given the fact that the
scarce research on gay identity has suggested a link between gay identity development
and psychological well-being, the proposed model included a structural equation where
gay identity is an independent variable and psychological well-being is a dependent
variable. Since Ryff (1995) proposed psychological well-being as a multidimensional
construct, the six subscales were included as six different instrument variables of
psychological well-being as a latent variable.

Multigroup Ethnic Identity Measure (Phinney, I 992).

The Multigroup Ethnic Identity Measure (MEIM) is a self-report measure based
on the elements of ethnic identity that are common across groups, so it can be used with
all ethnic groups. The measure consists of 14 items that assess three aspects of ethnic
identity: positive ethnic attitudes and sense of belonging (5 items), ethnic identity
achievement (7 items) and ethnic behaviors and practices (2 items). The questionnaire
also includes six items that assess other-group orientation. The items were rated on a
four-point scale that ranges from strongly agree to strongly disagree. Scores were
computed by reversing negatively worded items, summing across items and obtaining the
mean. Scores range from 4 (high ethnic identity) to 1 (low ethnic identity). The overall

reliability of the measure ranges from .81 to .90 (Phinney, 1992). The reliability

52

coefficients for the subscales range from .69 to .86. No coefficients were calculated for
the ethnic behaviors subscale because reliability cannot be computed with two items.
Phinney’s Multigroup Ethnic Identity Measure was included in the model to establish
convergent validity of the GIISS and the GGMSS. Given the fact that the scarce
theoretical papers published and qualitative studies on gay identity and ethnic identity
have suggested a link between these two constructs, a structural equation was included in
the model. Even though Phinney (1992) did not explicitly address ethnic identity as a
multidimensional construct, the structure of the MEIM suggests that general ethnic
identity has four dimensions: sense of belonging, ethnic identity achievement, ethnic
behaviors and other-group orientation. Therefore, this researcher included each subscale
as an instrumental variable to the latent variable ethnic identity.

Table VI summarizes the instruments used in this study and its correspondence to

the latent variable.

53

Table VI

Measurement Instruments Per Latent Variable

 

Latent Variable Instruments

 

Gay identity Gay Individual Identity — Synthesis Scale
Gay Group Membership — Synthesis Scale
Gay Identity Questionnaire (Brady and Busse, 1994)

Self-concept Six-F actor Self-Concept Scales (Stake, 1994)
- Likeability Scale (LS)
- Task Accomplishment Scale (TAS)
Power Scale (PS)
Vulnerability Scale (VS)
Giftedness Scale (GS)
Morality Scale (MS)

Psychological Well-Being Scales of Psychological Well-Being (Ryff, 1989)

- Autonomy Scale (AS)
Environmental Mastery Scale (EMS)
Personal Growth Scale (PGS)
Positive Relations With Others Scale (PROS)
- Purpose in Life Scale (PLS)
Self Acceptance Scale (SAS)

Ethnic Identity Multigroup Ethnic Identity Measure (Phinney, 1992)
- Identity Achievement Scale (IAS)
- Ethnic Behaviors Scale (EBS)
- Sense of belonging Scale (SBS)
- Other Group Orientation Scale (OGOS)

 

Data Analyses

To answer the first research question, Cronbach’s alpha indexes were computed to
examine the internal consistency of all the instruments used in this research.

To answer the first research question, a series of exploratory factor analyses

including the GIISS and the GGMSS were conducted to confirm the validity of gay

54

identity as a bidimensional construct comprised of gay individual identity formation and
gay group membership formation.

To answer the third and fourth research questions, an SEM analysis was
conducted to test the full proposed model. Structural equations describe the relationship
between dependent latent variables and independent latent variables. The firll model
includes not only the external model but also the measurement model that contains the
indicators used to measure the latent variables and the measurement errors. Figure 5

depicts the full model.

55

Figure 5
KEY:

GIISS — Gay Individual Identity — Synthesis Scale
GGMSS — Gay Group Membership — Synthesis Scale
GIQ-SI — Gay Identity Questionnaire, Stage 1: Confusion
GIQ-S2 — Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 — Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-SS — Gay Identity Questionnaire, Stage 5: Pride
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

TAS — Task Accomplishment Scale

PS — Power Scale

VS - Vulnerability Scale

GS — Giftedness Scale

MS — Morality Scale

IAS - Identity Achievement Scale

EBS — Ethnic Behaviors Scale

SBS — Sense of Belonging Scale

OGOS — Other-Group Orientation Scale

AS — Autonomy Scale

PGS -— Personal Growth Scale

EMS — Environmental Mastery Scale

PROS - Positive Relations W/Others Scale

PLS — Purpose in Life Scale

SAS - Self-Acceptance Scale

56

 

CHISS XXI

a GGMSS Xx2
a GIQ-S1 m

    

(Dr

'TAS hxlO

 

XXII

i

S Self Concept Psyc. Well Being

Xx12

<52 112

E

kx13

Xx14

hdS

II
U3

Figure 5. Full Model: Gay Identity, Self Concept, Ethnic Identity and Psychological Well Being

57

 

In this model, the latent variables are identified by circles. The independent latent
variables gay identity and self-concept are denoted 2;. and 2:2 , respectively while the
dependent latent variables ethnic identity and psychological well-being are denoted m
and 112, respectively. The instrumental or measurement variables are identified with
rectangles. The factor loadings of the measurement variables in the independent latent
variables are denoted 3.x] through M9 while the factor loadings of the measurement
variables in the dependent latent variables are denoted Xy] through 34/10. The errors of the
instrument variables that measure independent latent variables are denoted Sand 69 while
the errors of the instrument variables that measure dependent latent variables are denoted
81 through em. The correlation between the latent independent variables is denoted (1)].
The relationships portrayed by the structural equations including gay identity and ethnic
identity and gay identity and psychological well-being are denoted y; and 72, respectively.
Finally the measurement errors of the two structural equations are denoted l; 1 and C 2.

After concluding the model specification step, the next step was model estimation,
which consists of estimating the parameters of the relationships portrayed in the full
model. These parameters where included in the measurement equations as well as in the
structural modeling equations. The equations describing the measurement models are
given below.

Measurement equations:

GIISS=hx11§1+51
GGMSS = I» x21§1 + 52

GIQ-SI = hx31§ 1 + 53

58

GIQ-S2 = kx41§1 + 84
GIQ-S3 = 1x51!“ + 85
GIQ-S4=)tx61§1 +56
GIQ-SS = rmgl + 87
GIQ-S6=}.x81E,1 +58
LS=AX92§2+59
TAS=lx102§2 +810
PS=xln§2+5M
VS=AX122§2 +512
GS=l~x132§2 +513
MS=AX142§2 +814
IAS=ky11q1+8l
EBS=ky21q1 +82
SBS=ky31r|1 +83
OGOS=Xy4m1 +84
AS=hy51r|1 +85
PGS=ly6m1 +86

EMS =hy71nl + 87

59

Structural equations:

According to the firll model there were 43 parameters to be estimated: 14 factor

PROS = l» y81"11+ 88
PLS = It. y91'11 + 89

SAS = h. y101n1+ 810

(D: 1 <l>12]
(1)21 1

n1=§171+C1

112=§172+C2

1:01

loadings for two independent latent variables (gay identity and self-concept); 10 factor
loadings for two dependent latent variables (ethnic identity and psychological well
being); 24 error variances for the instrumental variables (measures); one correlation
between two latent variables (gay identity and self-concept); two structural equations
(gay identity and ethnic identity and gay identity and psychological well being); and two
error variances of the structural equations.
The next step in SEM was model identification, which consists in determining how many
degrees of fieedom are associated with the proposed model. To determine the degrees of

freedom we calculate the total number of observations in the lower diagonal of the

60

correlation matrix using the following equation: ‘/2 (p + q) ( p + q +1), where p + q
represents the total number of instrumental variables (measurements). The result of this
equation must be higher than the number of parameters to be estimated (overestimated
model). The full model contains 24 measurements, which means that the total number of
observations in the lower diagonal of the correlation matrix is 288 and the model has 247
degrees of freedom. It can be concluded that the model is overestimated and rendered

valid for firrther analysis.

61

CHAPTER III
Results
Demographic Data

The participants (n=109) completed a demographic questionnaire that included
items about age, sexual orientation, native language, ethnicity, religion, education,
occupation, partnership status, dating and participation in therapy.

Age.

From the total sample, 106 participants reported their age. The age mean in this
sample was 42.53 years old. The ages range from 18 to 77 years old: 17.9% were
between 18 and 27 years old, 16.0% were between 28 and 37 years old, 30.2% were
between 38 and 47 years old, 22.6% were between 48 and 57 years old, 11.3% were
between 58 and 67 years old and 1.9% were between 68 and 77 years old.

Sexual orientation.

From the total sample, 106 participants reported their sexual orientation. Most of
the participants reported a gay sexual orientation (88.7%). Ten percent of the sample
(10.4%) reported a bisexual sexual orientation while .9% reported being omnisexual. For
the purpose of this study, the participants that reported being bisexual or omnisexual were
removed from the analyses.

Language.

Participants were asked about their native language and 105 of them answered
this question. Most of the sample (84.8%) reported that English is their native language,

while 15.2% reported other native language.

62

Ethnicity.

Participants were asked to identify the ethnicity that best described them by
marking one of six alternatives. From the total sample, 107 participants answered this
question: 68.2% were White non Latino/Hispanic, 10.3% were Latino/Hispanic, 6.5%
were African American, 5.6% were Asian, 2.8% were Pacific Islander and 6.5%
answered Other.

Religion.

Participants were asked about their involvement with a religious community.
From the total sample, 106 answered this question: 69% of the sample reported not being
involved with a religious community while 31% indicated that they were involved with a
religious community.

Education.

Participants were asked about their highest achieved academic degree. From the
total sample, 106 participants answered this question: 17% of the sample had a High
School Diploma, 5.7% had a Technical College degree, 39.6% had a Bachelor’s degree,
17% had Master’s degree, 3.8% had a Doctoral degree, 1.9% had Post-doctoral studies
and 15.1% reported having another type of degree.

Occupation.

Participants were provided a domain in the web survey to write their current
occupation. From the total sample, 102 participants answered this question: 80.4%
reported having a job while 19.6% were not employed at the moment of the study. The
participants who had no employment were either students, on disability or retired. Due to

the wide variety of responses of the participants that reported being employed, the

63

researcher classified the participants’ responses using the Standard Occupational
Classification (SOC) System used by the US Department of Labor. The SOC System
organizes occupations in 23 major occupational groups. Sixteen of these groups were
represented in the sample. Table VII shows the distribution in terms of these major
occupational groups.

Table VII

Distribution of Occupations by Major Occupational Group

 

 

Major Occupational Groups Distribution (Percentage)
Management 19.6%
Business and Financial Operations 8.8%
Computer and Mathematical 7.8%
Architecture and Engineering .9%
Life, Physical and Social Science .9%
Community and Social Service 2.9%
Legal 1.9%
Education, Training and Library 7.8%
Arts, Design, Entertainment, Sports and Media 5.9%
Healthcare Practitioners and Technical 3.9%
Personal Care and Service .9%
Sales and Related 3.9%
Office and Administrative 5.9%
Installation, Maintenance and Repair 1.9%
Transportation and Material Moving 1.9%
Military Specific .9%

 

Partnership status.

The participants reported their current partnership status by choosing one of three
alternatives: Single, Committed Living Together and Committed Living Separate. From
the total sample, 106 participants answered this question: 55.7% were Single, 28.3% were

Committed Living Together and 16% were Committed Living Separate.

64

Dating.

Participants were asked about current and past dating experiences with same-sex
individuals with two questions: (1) Are you currently dating a same-sex person? and (2)
Have you dated a same-sex person before? From the total sample, 105 participants
answered the first question: 39% of the sample indicated that they were currently dating a
same-sex person while 61% reported not dating a same-sex person. From the total
sample, 106 participants answered the question about previous same-sex dating
experience: 90.6% indicated that they have dated a same-sex person before while 9.4%
reported not dating a same-sex person previously.

Disclosure of sexual orientation.

Participants were asked to indicate to whom they have disclosed their sexual
orientation by means of marking all that apply from a list including the following
categories: Nobody, Close friend(s), Acquaintance(s), Parent(s), Other Family
Member(s), Coworker(s), Supervisor(s) and Other. All the participants in the sample
answered this question. Close friend(s) is the category with the highest frequency as
84.4% of the participants reported that they have disclosed their sexual orientation to a
close friend. Participants disclosed their sexual orientation to people in the following
categories: Other Family Members (67%), Acquaintances (64.2%), Parents (59.6%),
Coworkers (51.4%), Supervisors (36.7), Other (1 1%) and Nobody (7.5%).

Therapy.

Participants were asked about previous experience in therapy with a helping

professional. From the total sample, 107 participants answered this question: 52.3%

65

indicated that they have participated in therapy before while 47.7 indicated that they have
never been to therapy prior to the study.
Research Question #1: Are the Items Reliable and Precise Enough to Grant Further
Development of the Items for Research and Clinical Uses?

To test the consistency of the items, Cronbach alphas (a) were computed for each
of the subscales of each instrument used in this research. The Cronbach alpha is a
correlation index that represents the relationship between the items in an instrument and
is calculated as follows:

k7
[1+(k—1)7']

 

a:

where k is the number of items used in the index, and ‘r' the average inter-item
correlation among k items. Cronbach’s alpha ranges from 0 to l, where higher indexes
indicate higher internal consistency of the instrument. Usually, a Cronbach’s alpha of .7
or higher is considered an acceptable level of scale reliability and internal consistency
(Lattin, Carroll & Green, 2003). Since Cronbach’s alpha is not robust against missing
cases, missing data was imputed using the SMEAN command of the statistical software
package SPSS 18. In this study, a missing case is an item left in blank by a particular
participant. Before imputing the data, this researcher examined the amount of missing
cases per instrument in the sample. Percentages of missing data in this sample ranged
from .5% (GGMSS) to 5.9% (GIQ). The percentage of missing cases for the total sample

in all of the 205 items was 3.1% (see Table VIII in p. 62). By imputing the data, this

66

researcher was able to increase the size of the sample (n) to 94 and reduce the variance
(.2).
Table VIII

Percentage of Missing Cases Per Instrument

 

 

 

Instrument Percentage Missing Cases
Gay Individual Identity -— Synthesis Scale (GIISS) .7
Gay Group Membership — Synthesis Scale (GGMSS) .5
Gay Identity Questionnaire (GIQ) 5.9
Scales of Psychological Well Being (SPWB) 2.4
Six-Factor Self-Concept Scales (SF SCS) 2.4
Multigroup Ethnic Identity Measure (MEIM) 5.9
TOTAL (all items) 3.1
GIISS and GGMSS.

The mean score of the GIISS in this sample was 34.76 and the standard deviation
was 5.18. The obtained Cronbach’s alpha ((1) with this sample was .89, which indicates
high reliability of the 10 items and, therefore, high internal consistency of the instrument.

The mean score of the GGMSS in this sample was 33.21 and the standard
deviation was 3.96. The obtained Cronbach’s alpha (or) with this sample was .79, which
indicates adequate reliability of the 10 items and, therefore, adequate internal
consistency of the instrument. Table IX summarizes the reliability results of the GIISS

and the GGMSS.

67

Table IX

Reliability of GIISS and GGMSS

 

 

 

Scale Mean Variance Standard deviation Cronbach’s alpha
GIISS 34.67 28.63 5.24 .89
GGMSS 33.47 15.71 3.54 .74

GIQ.

Brady and Busse (1994) developed this measure based on Cass’s developmental
model of gay identity. The instrument has six subscales, each one representing one of six
developmental stages: Confusion, Comparison, Tolerance, Acceptance, Pride and
Synthesis. The mean scores of these subscales in this sample ranged from .34
(Comparison) to 4.21 (Synthesis) and the standard deviations ranged from .69
(Confusion) to 2.40 (Synthesis). The obtained Cronbach’s alphas (a) ranged from .42

(Pride) to .85 (Synthesis). Table X summarizes these findings.

 

 

Table X
Reliability of GIQ
Subscale Mean Variance (Standard Cronbach’s alpha
evratron

Stage I: Confusion .45 .47 .69 .47

Stage 11: Comparison .34 .92 .96 .79

Stage 1H: Tolerance 1.04 1.74 1.32 .69

Stage IV: Acceptance 2.15 3.69 1.92 .79

Stage V: Pride 2.20 1.89 1.38 .42

Stage VI: Synthesis 4.21 5.77 2.40 .85

 

68

These results indicate only the subscales Stage 11, Stage IV and Stage VI have highly
reliable items and, therefore, have a high internal consistency. The items in Stage III
subscale have adequate reliability and, therefore, have adequate internal consistency.
Finally, the Confusion and Pride subscales have low reliability and, therefore, low
internal consistency. It is important to indicate that the reliability analysis discarded two
items included in the Confusion and Acceptance scales because they showed no variance.
These results suggested the removal of the scales of Confusion and Pride during the
model modification process.

Six-Factor Self-Concept Measure.

Stake’s measure consists of six subscales measuring the following dimensions:
Likeability, Task Accomplishment, Power, Vulnerability, Gifiedness and Morality. The
mean scores of these subscales in this sample ranged from 24.74 (Vulnerability Scale) to
37. 41 (Morality Scale) and the standard deviations ranged from 3.73 (Morality Scale) to
7.53 (Power Scale). The obtained Cronbach’s alphas (a) ranged from .79 (Morality Scale)
to .87 (Gifiedness Scale). Table XI summarizes these findings.

Table XI

Reliability of Six-F actor Self-Concept Measure

 

 

Scale Mean Variance Standard Cronbach’s alpha
devratron
Likeability 35.00 23.69 4.87 .84
Task Accomplishment 34.53 27.16 5.21 .84
Power 32.75 56.77 7.53 .84
Vulnerability 24.74 51.50 7.18 .82
Giftedness 28.20 30.39 5.51 .87
Morality 37.41 13.88 3.73 .79

 

69

These results indicate that, for this sample, each scale has highly reliable items
and, therefore, each scale has high internal consistency.

Scales of Psychological Well-Being.

Ryff” s (1989) instrument consists of six scales that measure six dimensions:
Autonomy, Environmental Mastery, Personal Growth, Positive Relations With Others,
Purpose in Life and Self-Acceptance. The mean scores of these subscales in this sample
ranged from 26.15 (Personal Growth Scale) to 39.95 (Environmental Mastery Scale) and
the standard deviations ranged fi'om 8.12 (Personal Growth Scale) to 13.69 (Self-
Acceptance Scale). The obtained Cronbach’s alphas (a) ranged from .82 (Autonomy and
Personal Growth scales) to .92 (Sel-Acceptance Scale). Table XII (in page 65)
summarizes these findings.

Table XII

Reliability of the Scales of Psychological Well-Being

 

 

Scale Mean Variance (Standard Cronbach’s alpha
evratron
Autonomy 37.13 103.77 10.19 .82
Enviromnental Mastery 39.95 163.26 12.78 .89
Personal Growth 26.15 65.92 8.12 .82
Positive Relations 35.37 168.76 12.99 .88
Purpose in Life 36.09 117.13 10.82 .84
Self-Acceptance 36.90 1 87.33 13.69 .92

 

These results indicate that, for this sample, each scale has highly reliable items and,

therefore, each scale has high internal consistency.

70

Multigroup Ethnic Identity Measure.

The Multigroup Ethnic Identity Measure has four subscales that measure the
following aspects: Affirmation and Belonging, Ethnic Identity Achievement, Ethnic
Behaviors and Other Group Orientation. The mean scores of these subscales in this
sample ranged from 4.70 (Ethnic Behaviors with 2 items) to 19.81 (Other Group
Orientation with 6 items) and the standard deviations ranged fiom 1.96 (Ethnic
Behaviors) to 5.11 (Ethnic Identity Achievement). The obtained Cronbach’s alphas (a)
ranged from .74 (Ethnic Behaviors) to .88 (Affirmation and Belonging). Table XIII
summarizes these findings.

Table XIII

Reliability of Multigroup Ethnic Identity Measure

 

 

Scale Mean Variance Emmi“! Cronbach’s alpha
evratron
Affirmation and Belonging 15.32 13.83 3.72 .88
Ethnic Identity Achievement 18.57 26.16 5.11 .83
Ethnic Behaviors 4.70 3.86 1.96 .74
Other Group Orientation 19.81 11.93 3.45 .77

 

These results indicate that, for this sample, all the subscales of the MEIM have highly
reliable items and, therefore, have high internal consistency.
Research Question #2: Is gay identity a bidimensional construct that comprises
individual identity formation and group membership formation?

To answer this question, this researcher ran a series of exploratory factor analyses
using SPSS 18, as part of the SEM analysis. Exploratory factor analysis is a non-

theoretically driven application that seeks to find the latent factors that account for the

71

patterns of collinearity among multiple metric variables, or items (Lattin, Carroll &
Green, 2003). This analysis is commonly used in the development of measurement
instruments as it allows examining how a certain set of items, once applied to a sample,
group into factors that measure some latent trait. Even if the measure under development
may be based on some theory, the first step is always examining how these items will
behave without any preconceived ideas. The factor solution consists in orienting the
factors in ways that the least amount of factors can account for the greatest possible
variance. In exploratory factor analysis is usually desirable to reorient the factors to
facilitate their interpretation. This reorientation is achieved by a process called rotation.
There are two types of rotation: orthogonal and oblique. The orthogonal rotation assumes
that the factors are independent while the oblique assumes that there might be some
correlation between them.

To test the validity of the bidimensional model of gay identity, this researcher’s
first step was to examine the behavior of the 20 items that comprise the piloted scales
GIISS and GGMSS. To accomplish this, a series of factor analyses were run with all 20
items: unrotated factors, unrotated forcing 2 factors, rotated orthogonal factors, rotated
oblique factors, rotated forcing 2 orthogonal factors and rotated forcing 2 oblique factors.
Based on the results of the factor analyses of the 20 items, the same steps were taken to
examine the behavior of the 10 items published by McCarn and Miller (1996) and the 10
items developed by this researcher.

Since data were imputed, this researcher also wanted to examine if there were
differences between the factor solutions of imputed and not imputed data. Therefore, each

one of the exploratory factor analyses was conducted with both imputed and not imputed

72

data. The results guided decisions regarding the use of imputed or not imputed data to run
the confirmatory factor analyses to answer research questions #3 and #4.

The interpretation of the results took into consideration five specific indexes: the
Keiser-Mayer-Olkin Measure of Sampling Adequacy, the Bartlett’s Test of Sphericity,
the determinant of the correlation matrix, the number of iterations and the factor loadings
of each item. The KMO evaluates the adequacy of sample size. A KMO index lower than
.50 indicates that the sample size is not adequate. Usually, an index equal or higher than
.80 indicates that the sample size is more than adequate. The Bartlett’s Test of Sphericity
is a test of intercorrelation of the variables. The variables should be somewhat correlated
but they should not be too highly correlated (extreme multicollinearity). In the case of
Bartlett’s Tests, it is desirable that the significance is less than .01 (p < .01). The
determinant of the correlation matrix examines the multicollinearity of the variables. If
the determinant is greater than .0000], then there is no multicollinearity. The number of
iterations is the number of attempts to find a satisfactory factor solution. The lower the
number of iterations, the more robust is the factor solution.

The factor loadings, expressed as correlation indexes, represent the contribution
of each item to every factor extracted by the analysis. In an exploratory factor analysis,
each item has one factor loading per factor. The highest of these factor loadings indicates
to which factor the item belongs. Furthermore, the higher the factor loading, the bigger is
its contribution to that factor. To evaluate the adequacy of the factor loadings, this
researcher used Creswell (2005) criteria: .00 to .20 — non existent or too low; .21 to .35 —

low; .35 to .65 — moderate; .66 to .85 — high; .86 to 1.00 — very high.

73

Exploratory Factor Analyses Series with 20 items.

A comparison of the KMO indexes indicated that the sampling adequacy of the
EFA with imputed data (KMO = .841) was better than the sampling adequacy of the EFA
without imputed data (KMO = .817). The Bartlett’s test was significant in both EFAs.
However, the determinants of the matrix are different in both analyses, with the
determinant of EF A of the non-imputed data being relatively bigger (4.81 E-005) than the
determinant of the EF A of the imputed data (1.29 E-005). These results are the same for
each EFA analysis in this series, as they are all based on the same correlation matrix.

First step.

The first step in this series was to run exploratory factor analyses (EF A) without
rotating factors with both the imputed and non-imputed data of the GIISS and the
GGMSS. Without imputed data, the factor solution extracted 5 factors after 20 iterations,
while the factor solution with imputed data also extracted 5 factors after only 5 iterations.

Table XIV shows the factor loading of each item and their factor location.

74

 

 

 

2 m- nwm. v5. Sum.- NE. omaO
woo. 3 ~ .- ommr mmm. nwm. o ~ #0
woo. RN. woo. mwm. Sum. in
w: .- ~w_. wt: now. how. :35
vmo. moor ._. w _ .- bwo. oov. o _ a0
3 _. momr KN. ovo. Eh. m :5
3:. $~ .- wwo. o_o. who. 3&0
owmr 02. _ov.- wwmr wmv. 2&0
mm? oofi. 2m. 02.- com. ~35
wit mwo. vow. mwo. 2.9. _ $0
moor _ w _ . who: mow. wmm. o _ no
Nmm. mom- mwm. .mw. NNV. $0
o~ w. wow. vwo. mm _ .- mow. wad
RN. So. So. 02.- 3o. 55
to. mom.- omo. mmmr 5%. £0
mw _ .- Kor wvo. an: oow. £0
2»..- wmmr 2 m 3;. mwv. 35
ww _. ooor o2 .- E _.- won. 935
Q- _. gov Novr Em..- Sfi. Q0
owo. vw _ .- ow? mg .- 3o. :5
m 8.8”— v 88$ w .883 N 868m ~ 88mm Eu:

 

 

mom..- Nnor mom. on? $5.
nmo. wit :1 .- wmm. w_w.
oofi .- woo. Nwo. EN. omm.
mm o .- mmor wo_ .- oov. now.
go. moo. SN.- mom. wmm.
vow. coo. ow fl. nwo. vow.
EN. m2. 3o.- hmor v3.
wnmr wofi- oRr mm? 5mm.
mwor wvo. :m. omor new.
two.- moor mmw. ooo. how.
32.- oofi. wwor wmm. o3.
oom. E3. w_N. mmw. :v.
Nvmr 3w. w_ _. wm _ .- mwv.
wwo. ohm. owo.- Q. _ .- mg.
NNN. mmfi .- So. womr our.
3o. oom- moo. win.- Sh.
2m. oovr mi. m2. 3v.
v2. m2. nw~ .- wwfi- omb.
2o. vmo. womr mowr New.
co. oo... 37 >2.- wwo.
m 88a..— v 88am w 888m N 883 _ 888m
Sea 3395 <mm

83 8238782 Em

 

EQQ 323%: 36 Katina-.82 62$: om .3290 ES» Viv-kw \e .8333... Shock QSESBQ Losxwwfihgd 383%

>Cm 2an

75

In the factor solution of the EFA with non-imputed data, 17 of the 20 items loaded
on the first factor while one item (#8) loaded in the fourth factor and the two remaining
items (#16 and #19) loaded in the second factor. Based on Creswell (2005) criteria, most
of the factor loadings are moderate to high with one exception. Item #7 had a factor
loading of .022. This low factor loading suggests that the item contribution to the factor is
not significant. Item #8 is also problematic because even though it loaded on the fourth
factor, this factor loading (.468) is basically the same as its factor loading in the first
factor (.465).

In the factor solution of the EFA with imputed data, 17 of the 20 items loaded on
the first factor. One item (#7) loaded on the fourth factor while the two remaining items
(#17 and #19) loaded in the second factor. Based on Creswell (2005) criteria, all factor
loadings are moderate to high with one exception.

None of these factor solutions suggested the structure of two different instruments
(GIISS and GGMSS) that measure two different developmental phases in individual gay
identity and gay group membership. Based on the adequacy of the sample and the
number of iterations, the factor solution of the imputed data appears to be stronger.

Second step.

The second step in this series was to run 2-factor exploratory factor analyses
(EFA) without rotating factors with both the imputed and non-imputed data of the GIISS
and the GGMSS. In these analyses, the factor solutions were forced to a maximum of two
factors. Without imputed data, the 2-factor solution was extracted after 8 iterations, while
the 2-factor solution with imputed data was extracted afier only 5 iterations. Table XV

shows the factor loading of each item and their factor location.

76

Table XV
Factor Loadings for 2-F actor Exploratory Factor Analyses of GIISS and GGMSS 20

Items, Non-Imputed and Imputed Data

 

 

 

 

 

EFA Non-imputed Data EFA Imputed Data
Item Factor 1 Factor 2 Factor 1 Factor 2
Q#1 .617 -. 160 .640 -.161
Q#2 .720 -.312 .729 -359
Q#3 .705 -.1 18 .747 -. 186
Q#4 .429 .120 .413 .095
Q#5 .803 -.303 .767 -359
Q#6 .750 -.273 .739 -.313
Q#7 .021 -.150 .160 -.155
Q#8 .443 -.11 1 .474 -.121
Q#9 .408 .279 .413 .292
Q#10 .557 .403 .652 .355
Q#1 1 .466 .014 .492 .045
Q#12 .485 -. 132 .526 -.053
Q#13 .442 -.185 .540 -.125
Q#14 .680 -.004 .741 -.O65
Q#15 .468 .023 .591 .072
Q#16 .405 .654 .552 .545
Q#17 .594 .484 .695 .467
Q#18 .541 .281 .554 .278
Q#19 .337 .354 .318 353
Q#20 .734 -.240 .753 -. 150

 

In the factor solution of the EF A with non-imputed data, 18 of the 20 items loaded
on the first factor while the two remaining items (#16 and #19) loaded on the second
factor. Based on Creswell (2005) criteria, most of the factor loadings are moderate to
high with one exception. Item #7 had a factor loading of .021. This low factor loading

suggests that the item contribution to the factor is not significant. Item #19 is also

77

problematic because even though it loaded on the second factor, this factor loading (.354)
is not too different from its factor loading in the first factor (.337).

In the factor solution of the EF A with imputed data, 19 of the 20 items loaded on
the first factor. The remaining item (#19) loaded in the second factor. Based on Creswell
(2005) criteria, all factor loadings are moderate to high with one exception. Item #7 (“I
still get angry at the way homosexuals are treated, but not as much as once I did.”) had a
factor loading of .160. This low factor loading suggests that the item contribution to the
factor is not significant. Item #19 is also problematic because even though it loaded on
the second factor, this factor loading (.353) is not too different from its factor loading in
the first factor (.318).

None of these factor solutions suggested the structure of two different instruments
(GIISS and GGMSS) that measure two different developmental phases in individual gay
identity and gay group membership. Based on the adequacy of the sample and the
number of iterations, the factor solution of the imputed data appears to be stronger.

Third step.

The third step in this series was to run exploratory factor analyses (EF A) with a
rotated orthogonal factor solution for both the imputed and non-imputed data of the
GIISS and the GGMSS. Without imputed data, the factor solution extracted 5 factors
after 8 iterations, while the factor solution with imputed data also extracted 5 factors after
only 6 iterations. Table XVI shows the factor loading of each item and their factor

location.

78

 

 

 

mom. o8. moo. o2. o3. oz. 39. o8. of. E... omoo
so. 8: .- moor mom. 2:. o: .- E. voo- E... cm. 2.5
woo. woo. omm. oo... o:. o2. So. So. om. E. o :5
me. So: moo. ooh m2. so: so: an. ok. 9:. too
to: woo. omo. woo. me. o:- No _. moo: one. go. Boo
mom- ooo. woo. Em. own. so. on... mom. 5. oom. Soc
2 7 so. woo. oom. oom. moo. oom. mom. :3. mum. zoo
Em. ms. me. mom. 9%. omo. o3- o2. oz. Sm. 2.5
So; ooo. ofi. moo. me. omo. 5. So. So. so «:5
So. moo. mam. SN. o3. ooo.- : _. oam. E. of. E5
o2. 9:. so. «mo. oom. So. So. oNN. 98. E. ooao
oom. :o. 2:- mo... so. of. 8m. woo. woo. moo. $0
3. on... So. o2. SN. oom. to; own. mom. oNN. woo
moo; o3. o8. o8... m2. m3. o8. so; woo- to. So
So; moo. mom. #2. So. So. ME. o2. ooo. ooh £0
woo. woo: 3.. no. as. :o- ooo. mom. 2 o. :3. 30
_o_ .- o3- o8. on. «on. Sn- E. 5. on... so. £0
ooo. 8N. me. mom. 3o. o2. 5. we. o2. N3. 25
So. woo. woo. 9:. 2o. mom. So... so. o2. o3. .36
E. So. of. on. oom. 2o. mo. 3;. oz. 2o. :5
m 88o"— v SSE m .883 N Sousa ~ 583 m 5.8m v “Boom .6. Soon“— N 86mm _ 88mm 88
38 8395 Em 38 oaooéaoz Em _

 

gm: om mmgww VS» 5.2% \o 23332 583.8: 5?: omwéomw x823 bSEQERMmeMRfiaoN SEQ-n.

~>x O~n~mr~r

79

Once the factors were rotated, the distribution of items across factors was more
diverse. In the orthogonal EFA with non-imputed data, nine items loaded on the first
factor, 5 items loaded in the second factor, 2 items loaded in the third factor, 2 items
loaded on the fourth factor and 2 items loaded on the fifth factor. Based on Creswell
(2005) criteria, all factor loadings are moderate to high with one exception. Item #4 had a
factor loading of .246. This low factor loading suggests that the item contribution to the
factor is not significant.

In the orthogonal EF A with imputed data, 9 items loaded on the first factor, 6
items loaded in the second factor, 2 items loaded in the third factor, 2 items loaded on the
fourth factor and 1 item loaded on the fifth factor. Based on Creswell (2005) criteria, all
factor loadings are moderate to high with one exception. Item #4 had a factor loading of
.336. This low factor loading suggests that the item contribution to the factor is not
significant.

In both factor solutions 9 of the 10 items of the GIISS loaded on the first factor.
However, the items of the GGMSS and the remaining item of the GIISS loaded on
different factors. Based on the adequacy of the sample and the number of iterations, the
factor solution of the imputed data appears to be stronger.

Fourth step.

The fourth step in this series was to run exploratory factor analyses (EFA) with a
rotated oblique factor solution for both the imputed and non-imputed data of the GIISS
and the GGMSS. Without imputed data, the factor solution extracted 5 factors afier 21
iterations, while the factor solution with imputed data also extracted 5 factors afier only

13 iterations. Table XVII shows the factor loading of each item and their factor location.

80

 

 

 

 

 

ovor ooo. Noo. ooo. ooo.
ooo; moor oo_; oom. boo.
moo. __o. moo. ope. o_o.
oo_c ooo- ooo. ooh. ooo;
oo_. ooo. oo_: ooo. ooo.
who. ooo. ooo. oo_. boo.
oo_. ooo. ooo. moo. ooo.
ooor _oo- o__. ooo. moo.
ooo. moor ooh. ooo- woo.
moo. _oo; Now. moo. moo.
moo. ooo. moo. ooo. ooo.
ooo. woo. ooo. oVo. ooo;
oo_r o_o. one. moo. ooo.
moo. hoe. ooo- ooo. oo_.
ooo. ooo: woo. moo: o_o.
ovo; ooo- ooo. _o_a omo.
oo_. ooo- oo_. oo_. Nam.
moo. woo. ooox. No_. ooo.
moo; moo. oo_r moo. moo.
ooo: o_or ooo. ooo. mom.
m ooooom v “Soon— w Sooom N ooooom _ “Boom
son 8595 Em 26:8

2» _. m3. owm . ooo. 9mm. omo0
3: .- moor :Nr mmv. oofi o$0
mm o. mmo. mom. omm. w _o. w 50
moor 2 o. moo. woo. 8o. 2, #0
N2 .- fin _ .- ow _ .- wg. ooor 030
«not oovr vow. woo. oom. m o #0
Vmo. mmmr ooo. m2. vom. v50
voor vom. wvo. 8:. wow. M 2.0
m S. Q _.- now. So.- mmo. N50
owor woor wow. Q o. wmo. g 30
So. 2o. N2. woo. woo. o$0
woo. flow.- 21. own. w _o.- o¢0
bwm. fiwo. Km. mom. moo. wa0
mom. moor omor owor moo. $0
vwor 2 fl- EN. woor our. 30
ooor who. m :V. mvor 5%. £0
vwwr owor KN. m2. :N. $0
ooo. ooor Nwo. 3:. one. 30
N3. o2. om fl .- woo. woo. $0
woo. woo. moor wmo. N3. 50
m ooooom v .885— w 88o”— N Boom-o _ Boos”—
:5:

38 oaooééoz «mm 25:3

 

oEmh om 5.390 has MMNNGKG notoooz 2.2330 5%: 8.3354 383% bEESQoQ xQoMEhooQ Shook

=>X 2an

81

In the oblique EF A with non-imputed data, 10 items loaded on the first factor,.6
items loaded in the second factor, 2 items loaded in the third factor, 1 item loaded on the
fourth factor and 1 item loaded on the fifth factor. This factor solution loaded 9 items of
the GIISS and 1 item of the GGMSS in the first factor. The remaining items loaded
across the rest of the factors. Based on Creswell (2005) criteria, most of factor loadings
are moderate to high with four exceptions. Item #4 had a factor loading of .246, item #9
had a factor loading of .337, item #15 had a factor loading of .300 and item #20 had a
factor loading of .334. These low factor loadings suggest that the items contribution to
the factor was not significant.

In the oblique EF A with imputed data, 9 items loaded on the first factor, 5 items
loaded in the second factor, 4 items loaded in the third factor, 1 item loaded on the fourth
factor and 1 item loaded on the fifth factor. This factor solution loaded 9 of the 10 items
from the GIISS in the first factor, while the remaining GIISS item and the items fiom the
GGMSS loaded across all factors. Based on Creswell (2005) criteria, most factor loadings
were moderate to high with three exceptions. Item #4 had a factor loading of .342, item
#13 had a factor loading of .332 and item #15 had a factor loading of .297. These low
factor loadings suggest that the items contribution to the factor is not significant. Based
on the adequacy of the sample and the number of iterations, the factor solution of the
imputed data appears to be stronger.

Fiflh step.

The fifth step in this sen’es was to run orthogonal 2-factor exploratory factor
analyses (EFA) with both the imputed and non-imputed data of the GIISS and the

GGMSS. In these analyses, the factor solutions were forced to a maximum of two factors.

82

In both EFAs, the 2-factor solution was extracted after 3 iterations. Table XVIII. shows
the factor loading of each item and their factor location.

Table XVIII

Factor Loadings for 2-F actor Exploratory Analyses With Varimax Rotation of GIISS and

 

 

 

 

 

GGMSS 20 items
It EF A Non-Imputed Data EFA Imputed Data
em Factor 1 Factor 2 Factor 1 Factor 2

Q#1 .613 .173 .604 .266
Q#2 .780 .093 .796 .164
Q#3 .668 .253 .704 .31 1
Q#4 .310 .320 .268 .329
Q#5 .846 .142 .826 .187
Q#6 .785 .143 .776 .206
Q#7 .094 -.1 19 .222 -.024
Q#8 .439 .128 .449 .196
Q#9 .21 1 .446 .147 .484
Q#10 .278 .629 .297 .680
Q#1 1 .395 .247 .361 .338
Q#12 .485 .130 .448 .281
Q#13 .475 .063 .503 .232
Q#14 .590 .339 .625 .403
Q#15 .393 .255 .422 .420
Q#16 .021 .769 .101 .769
Q#17 .269 .718 .262 .795
Q#18 .326 .515 .266 .560
Q#19 .1 13 .476 .035 .474
Q#20 .755 .163 .687 .344

 

In the 2-factor orthogonal solution of the EF A with non-imputed data, 14 of the
20 items load on the first factor while the remaining 6 items loaded in the second factor.
Based on Creswell (2005) criteria, most of the factor loadings are moderate to high with

two exceptions. Item #4 had a factor loading of .329 and item #7 had a factor loading of

83

.074. These low factor loadings suggest that the items contribution to the factor is not
significant.

In the 2-factor orthogonal solution of the EF A with imputed data, 13 of the 20
items loaded on the first factor while the remaining 7 items loaded in the second factor.
Based on Creswell (2005) criteria, most factor loadings are moderate to high with two
exceptions. Item #4 had a factor loading of .329 and item #7 had a factor loading of .222.
These low factor loadings suggest that the item’s contribution to the factor is not
significant.

None of these factor solutions suggested the structure of two different instruments
(GIISS and GGMSS) that measure two different developmental phases in individual gay
identity and gay group membership. Based on the adequacy of the sample and the
number of iterations, the factor solution of the imputed data appears to be stronger.

Sixth step.

The final step in this series was to run a 2-factor oblique solution EFA with both
non-imputed and imputed data. In these analyses, the factor solutions were forced to a
maximum of two factors. In both EF As, the 2-factor solution was extracted after 7

iterations. Table XIX shows the factor loading of each item and their factor location.

84

Table XIX

Factor Loadings for 2-Factor Exploratory Analyses With Oblimin Rotation of GIISS and

 

 

 

 

 

GGMSS 20 items
I EFA Non-imputed Data EFA Imputed Data
tem Factor 1 Factor 2 Factor 1 Factor 2

Q#1 .628 .028 .608 .097
Q#2 .810 -095 .846 -.075
Q#3 .678 .097 .708 .1 15
Q#4 .295 .253 .218 .274
Q#5 .876 -.060 .874 -.060
Q#6 .812 -045 .813 -.023
Q#7 .110 -.145 .253 -.098
Q#8 .449 .024 .452 .071
Q#9 .179 .407 .045 .483
Q#10 .232 .577 .165 .649
Q#11 .392 .157 .319 .254
Q#12 .497 .015 .430 .164
Q#13 .493 -.051 .504 .093
Q#14 .587 .204 .597 .241
Q#15 .389 .166 .368 .324
Q#16 -052 .784 -.075 .809
Q#17 .214 .671 .097 .786
Q#18 .293 .450 .160 .527
Q#19 .073 .461 -.O78 .508
Q#20 .778 -.017 .680 .156

 

In the 2-factor oblique solution of the EF A with non-imputed data, 14 of the 20
items load on the first factor while the remaining 6 items loaded in the second factor.
Based on Creswell (2005) criteria, most of the factor loadings are moderate to high with
ttwo exceptions. Item #4 had a factor loading of .295 and item #7 had a factor loading of
.110. These low factor loadings suggest that the items’ contribution to the factor is not

significant.

85

In the 2-factor oblique solution of the EFA with imputed data, 13 of the 20 items
loaded on the first factor while the remaining 7 items loaded in the second factor. Based
on Creswell (2005) criteria, most factor loadings are moderate to high with three
exceptions. Item #4 had a factor loading of .274, item #7 had a factor loading of .253 and
item #11 had a factor loading of .319. These low factor loadings suggest that the items’
contribution to the factor is not significant.

None of these factor solutions suggested the structure of two different instruments
(GIISS and GGMSS) that measure two different developmental phases in individual gay
identity and gay group membership. Based on the adequacy of the sample and the
number of iterations, the factor solution of the imputed data appears to be stronger.

Exploratory Factor Analyses Series with McCarn & Miller’s (1996) 10 Items

A comparison of the KMO indexes indicated that the sampling adequacy of the
EFA with non-imputed data (KMO = .803) was better than the sampling adequacy of the
EFA with imputed data (KMO = .79). The Bartlett’s test was significant in both EFAs.
The determinants of the matrix are the same in both analyses (.023). It is important to
notice that these determinants improved significantly when compared with the
determinants of the correlation matrix of the 20 items (4.81 E-005 for non-imputed data
and 1.29 E-005 for imputed data). These results are the same for each EF A analysis in
this series, as they are all based on the same correlation matrix.

First step.

The first step in this series was to run exploratory factor analyses (EF A) without
rotating factors with both the imputed and non-imputed data of the GIISS and the

GGMSS. Without imputed data, the factor solution extracted 3 factors afier 14 iterations,

86

while the factor solution with imputed data also extracted 3 factors after 27 iterations.
Table XX shows the factor loading of each item and their factor location.
Table XX

Factor Loadings for Exploratory Factor Analyses of McCarn & Miller ’s (I 995) 10 items

 

 

 

 

 

from GIISS and GGMSS

I EF A Non-imputed Data EF A Imputed Data

tem

Factor 1 Factor 2 Factor 3 Factor 1 Factor 2 Factor 3

Q#1 .710 -.285 .228 .673 -.171 .165
Q#2 .763 -.446 .088 .803 -.552 .158
Q#4 .463 .117 -.097 .433 .195 -.l78
Q#5 .860 —.027 -344 .868 -.004 -.397
Q#6 .806 -.138 -.139 .800 -.111 -.131
Q#7 .116 -.064 .003 .130 -. 169 .083
Q#9 .337 .264 .145 .341 .293 .191
Q#10 .528 .199 .399 .526 .200 .358
Q#1 l .526 .406 .024 .505 .423 .047
Q#12 .524 .425 -093 .512 .331 .002

 

According to Creswell (2005) criteria, the factor loading in both EF As are
between moderate and high with two exceptions. Item #7 has a factor loading of .116 in
the EFA without imputed data and a factor loading of .130 in the EF A with imputed data.
Item #9 has a factor loading of .337 in the EFA without imputed data and a factor loading
of .340 in the EF A with imputed data. However, the factor loadings of item #9 are very
close to the minimum suggested by Creswell (.360). In both analyses all the items loaded
on the first factor, even though they were purported to measure two different processes:
individual gay identity (items 1, 2, 4, 5 and 6) and gay group membership (items 7, 9, 10,
11 and 12). These results suggest at least two possibilities. Either the items measure one

process or gay identity is a unidimensional construct.

87

Based on the adequacy of the sample and the number of iterations, the factor
solution with non-imputed data seems to be more robust.

Second step.

The second step in this series was to run a non-rotated 2-factor exploratory factor
analyses (EF A) with both the imputed and non-imputed data of the GIISS and the
GGMSS. Without imputed data, the 2-factor solution was extracted after 13 iterations,
while the 2-factor solution with imputed data was extracted after 19 iterations. Table XXI
shows the factor loading of each item and their factor location.

Table XXI
Factor Loadings for 2-F actor Exploratory Factor Analyses of McCarn & Miller ’s (I 995)

10 items from GIISS and GGMSS

 

 

 

 

I EF A Non-imputed Data EF A Imputed Data
tern Factor 1 Factor 2 Factor 1 Factor 2
Q#1 .702 -.264 .676 -. 180
Q#2 .776 -.464 .797 -.526
Q#4 .465 .122 .433 .186
Q#5 .826 -.01 1 .820 -.004
Q#6 .808 -.127 .807 -.1 17
Q#7 .117 -.066 .132 -. 167
Q#9 .337 .249 .342 .273
Q#10 .503 .147 .509 .159
Q#11 .533 .419 .514 .440
Q#12 .529 .430 .521 .346

 

Overall, the results of these analyses are not too different from the analyses of the
previous step. According to Creswell (2005) criteria, the factor loading in both EF As are
between moderate and high with two exceptions. Item #7 has a factor loading of .117 in

the EFA without imputed data and a factor loading of .132 in the EF A with imputed data.

88

Item #9 has a factor loading of .337 in the EF A without imputed data and a factor loading
of .342 in the EF A with imputed data. However, the factor loadings of item #9 are very
close to the minimum suggested by Creswell (.360). In both analyses all the items loaded
on the first factor, even though they were purported to measure two different processes:
individual gay identity (items 1, 2, 4, 5 and 6) and gay group membership (items 7, 9, 10,
11 and 12). These results further suggest the previously metioned two possibilities. Either
the items measure one process or gay identity is a unidimensional construct.

Based on the adequacy of the sample and the number of iterations, the factor
solution with non-imputed data seems to be more robust.

Third step.

The third step in this series was to run exploratory factor analyses (EF A) with a
rotated orthogonal factor solution for both the imputed and non-imputed data of the
GIISS and the GGMSS. Without imputed data, the factor solution extracted 3 factors
after 24 iterations, while the factor solution with imputed data also extracted 3 factors
after only 5 iterations. Table XXII shows the factor loading of each item and their factor

location.

89

Table XXII
Factor Loadings for Exploratory Analyses With Varimax Rotation of McCarn & Miller’s

(1995) I 0 items from GIISS and GGMSS

 

 

 

 

 

I EF A Non-imputed Data EF A Imputed Data
tern Factor 1 Factor 2 Factor 3 Factor 1 Factor 2 Factor 3
Q#1 .730 .098 .310 .286 .571 .317
Q#2 .862 .157 .146 .326 .925 .125
Q#4 .237 .372 .208 .437 .041 .254
Q#5 .589 .702 .134 .858 .356 .222
Q#6 .652 .477 .186 .599 .492 .262
Q#7 .127 .032 .017 .000 .229 -.001
Q#9 .074 .193 .403 .1 12 .051 .473
Q#10 .284 .085 .623 .093 .286 .595
Q#1 1 .096 .445 .485 .342 -.004 .565
Q#12 .067 .536 .415 .369 .052 .483

 

In the EF A with non-imputed data 4 items loaded on the first factor, 3 items
loaded on the second factor and 3 items loaded on the third factor. In the EF A with
imputed data 3 items loaded on the first factor, 3 items loaded on the second factor and 4
items loaded on the third factor. None of these factor solutions suggested the structure of
two different instruments (GIISS and GGMSS) that measure two different developmental
phases in individual gay identity and gay group membership.

According to Creswell (2005) criteria, the factor loading in both EFAs are
between moderate and high with one exception. Item #7 has a factor loading of .127 in
the EF A without imputed data and a factor loading of .229 in the EFA with imputed data.
These factor loadings indicate that this item not does contribute significantly to the factor.
Based on the adequacy of the sample and the number of iterations, the factor solution of

the non-imputed data appears to be stronger.

90

Fourth step.

The fourth step in this series was to run a rotated oblique exploratory factor
analyses (EFA) for both the imputed and non-imputed data of the GIISS and the
GGMSS. Without imputed data, the factor solution extracted 3 factors after 16 iterations,
while the factor solution with imputed data extracted 3 factors after only 9 iterations.
Table XXIII shows the factor loading of each item and their factor location.

Table XXIII
Factor Loadings for Exploratory Analyses With Oblimin Rotation of McCarn & Miller’s

(19.95) 10 items from GIISS and GGMSS

 

 

 

 

 

I EFA Non-imputed Data EF A Imputed Data
tem Factor 1 Factor 2 Factor 3 Factor 1 Factor 2 Factor 3
Q#1 .778 .425 .1 1 1 .484 —.607 .460
Q#2 .883 .360 -.025 .534 -.954 .339
Q#4 .368 .432 —. 187 .493 -.090 .358
Q#5 .772 .643 -492 .939 -440 .480
Q#6 .781 .553 -.275 .739 -554 .473
Q#7 .132 .055 -.014 .046 -.227 .033
Q#9 .208 .440 .065 .262 -.080 .488
Q#10 .424 .578 .283 .332 -.316 .635
Q#1 1 .314 .664 -.097 .493 -.050 .625
Q#12 .298 .662 -.213 .504 -.104 .562

 

In the EF A with non-imputed data 5 items loaded on the first factor and 5 items
loaded on the second factor. In the EF A with imputed data 6 items loaded on the first
factor and 4 items loaded on the third factor. However, these distributions do not
correspond to the initial distribution of items per process. The factor solution of the non-
imputed data EFA grouped 4 items of the GIISS and 1 item of the GGMSS in one factor

and 4 items of the GGMSS and 1 item of the GIISS in other factor. The factor solution of

91

the imputed data EFA grouped 5 items of the GIISS and 1 item of the GGMSS in one
factor and 4 items of the GGMSS in other factor. None of these factor solutions
suggested the structure of two different instruments (GIISS and GGMSS) that measure
two different developmental phases in individual gay identity and gay group membership.

According to Creswell (2005) criteria, the factor loading in both EF As are
between moderate and high with one exception. Item #7 has a factor loading of .132 in
the EFA without imputed data and a factor loading of .046 in the EF A with imputed data.
These factor loadings indicate that this item does not contribute significantly to the factor.
Based on the adequacy of the sample and the number of iterations, the factor solution of
the non-imputed data appears to be slightly better.

Fifth step.

The fifth step in this series was to run an orthogonal 2-factor exploratory factor
analyses (EF A) with both the imputed and non-imputed data of the GIISS and the
GGMSS. In these analyses, the factor solutions were forced to a maximum of two factors.
In both EFAs, the 2-factor solution was extracted after 3 iterations. Table XXIV shows

the factor loading of each item and their factor location.

92

Table XXIV

Factor Loadings for 2-Factor Exploratory Analyses With Varimax Rotation of McCarn &

Miller’s (1 995) 10 items from GIISS and GGMSS

 

 

 

 

 

I EFA Non-imputed Data EFA hnputed Data

tern Factor 1 Factor 2 Factor 1 Factor 2
Q#1 .708 .247 .355 .603
Q#2 .893 .141 .198 .934
Q#4 .279 .391 .439 .172
Q#5 .641 .521 .581 .578
Q#6 .702 .420 .492 .650
Q#7 .132 .025 -.024 .211
Q#9 .099 .407 .435 .045
Q#10 .292 .435 .474 .244
Q#1 1 .141 .663 .675 .047
Q#12 .130 .669 .614 .119

 

In the EF A with non-imputed data 5 items loaded on the first factor and 5 items
loaded on the second factor. In the EF A with imputed data 6 items loaded on the first
factor and 4 items loaded on the third factor. However, these distributions do not
correspond to the initial distribution of items per process. The factor solution of the non-
imputed data EFA grouped 4 items of the GIISS and 1 item of the GGMSS in one factor
and 4 items of the GGMSS and I item of the GIISS in other factor. The factor solution of
the imputed data EFA grouped 5 items of the GIISS and 1 item of the GGMSS in one
factor and 4 items of the GGM SS in other factor.

According to Creswell (2005) criteria, the factor loading in both EFAs are
between moderate and high with one exception. Item #7 has a factor loading of .132 in
the EF A without imputed data and a factor loading of .211 in the EF A with imputed data.

These factor loadings indicate that this item does not contribute significantly to the factor.

93

Sixth step.

The sixth step in this series was to run an oblique 2-factor exploratory factor
analyses (EF A) with both the imputed and non-imputed data of the GIISS and the
GGMSS. In the non-imputed data EF A, the 2-factor solution was extracted after 7
iterations. In the imputed data EF A, the 2-factor solution was extracted after 8 iterations.
Table XV shows the factor loading of each item and their factor location.

Table XV
Factor Loadings for 2-F actor Exploratory Analyses With Oblimin Rotation of McCarn &

Miller ’s (I 995) I 0 items from GIISS and GGMSS

 

 

 

 

 

I EF A Non-imputed Data EF A Imputed Data

tem Factor 1 Factor 2 Factor 1 Factor 2
Q#1 .718 .059 .279 -.548
Q#2 .958 -.1 17 .058 -.932
Q#4 .192 .353 .438 -.076
Q#5 .558 .387 .524 -.468
Q#6 .658 .255 .417 -.564
Q#7 .141 -.013 -.059 -.227
Q#9 -.015 .426 .455 .056
Q#10 .193 .398 .464 -.143
Q#11 -.048 .701 .710 .111
Q#12 -.062 .711 .632 .021

 

In the non-imputed data EFA, five items loaded in each factor. However, these
results are consistent with previous steps in that this distribution does not reflect the
original distribution of items in the GIISS and the GGMSS. The first factor grouped 4
items from the GIISS and 1 item from the GGMSS while the second factor grouped 4
items of the GGMSS and 1 item from the GIISS. The factor solution in the imputed data

EF A grouped all of the items in the first factor.

94

According to Creswell (2005) criteria, the factor loading in both EFAs are
between moderate and high with one exception. Item #7 has a factor loading of .141 in
the EF A without imputed data and a factor loading of -.059 in the EPA with imputed
data. These factor loadings indicate that this item does not contribute significantly to the
factor. This finding has been consistent across all the exploratory factor analyses
conducted up this point. Based on the adequacy of the sample and the number of
iterations, the factor solution of the non-imputed data appears to be slightly better.

Exploratory factor analyses series with 10 items developed by researcher

A comparison of the KMO indexes indicated that the sampling adequacy of the
EF A with both non-imputed and imputed data is the same (KMO = .760). The Bartlett’s
test was significant in both EFAs. These factor solutions may not be as robust with a
KMO index of less than .80. The determinant of the correlation matrix fi'om the non-
imputed data was higher (.039) than the determinant of the correlation matrix from the
imputed data (.018). It is important to notice that these determinants improved
significantly when compared with the determinants of the correlation matrix of the 20
items (4.81 E-005 for non-imputed data and 1.29 E-005 for imputed data). However,
there is a difference between these indexes that is not observed between the determinants
of the fi'om the correlation matrix of the 10 items from McCarn & Miller (1996). These
results are the same for each EFA analysis in this series, as they are all based on the same
correlation matrix.

First step.

The first step in this series was to run exploratory factor analyses (EF A) without

rotating factors with both the imputed and non-imputed data of the GIISS and the

95

GGMSS. Without imputed data, the extraction of factors was terminated after 999
iterations, while the factor solution with imputed data also extracted 2 factors after 8
iterations. Table XVI shows the factor loading of each item and their factor location.
Table XVI

Factor Loadings for Exploratory Factor Analysis of A gosto 's (2004 ) 10 items from GIISS

 

 

 

and GGMSS
EF A Imputed data
Item
Factor 1 Factor 2

Q #3 .716 -.274
Q #8 .463 -.315
Q #13 .539 -.075
Q #14 .710 -.191
Q #15 .573 -.040
Q #16 .625 .436
Q #17 .801 .278
Q #18 .576 .165
Q #19 .382 .421
Q #20 .681 -.308

 

The results of the EF A with the imputed data grouped 9 out of the 10 items in one
factor. This factor solution did not suggest the structure of two different instruments
(GIISS and GGMSS) that measure two different developmental phases in individual gay
identity and gay group membership. According to Creswell (2005) criteria, all of the
factor loadings are between moderate and high.

Second step.

The second step in this series was to run a non-rotated 2-factor exploratory factor
analysis (EF A) with both the imputed and non-imputed data of the GIISS and the

GGMSS. Without imputed data, the 2-factor solution was extracted after 11 iterations,

96

while the 2-factor solution with imputed data was extracted after 8 iterations. Table XVII
shows the factor loading of each item and their factor location.
Table XXVII

Factor Loadings for 2-F actor Exploratory Factor Analyses of A gosto ’s (2004 ) 1 0 items

 

 

 

 

 

from GIISS and GGMSS

I EF A Non-imputed Data EF A Imputed Data

tem

Factor 1 Factor 2 Factor 1 Factor 2

Q #3 .671 -.276 .716 -.274
Q #8 .448 -313 .463 -.315
Q #13 .433 -095 .539 -.075
Q #14 .626 —.197 .710 -.191
Q #15 .449 -.051 .573 -.040
Q #16 .522 .507 .625 .436
Q #17 .719 .310 .801 .278
Q #18 .603 .132 .576 .165
Q #19 .440 .399 .382 .421
Q #20 .629 -.364 .681 -.308

 

The results of the 2-factor solution with the non-imputed data indicated that all
items loaded on the first factor, while the results of the 2-factor solution with the imputed
data loaded 9 of the 10 items in the first factor and the remaining item in the second
factor. None of these factor solutions suggested the structure of two different instruments
(GIISS and GGMSS) that measure two different developmental phases in individual gay
identity and gay group membership. Based on the adequacy of the sample and the
number of iterations, the factor solution of the imputed data appears to be stronger.

According to Creswell (2005) criteria, all of the factor loadings in both EF As are
between moderate and high. However, item #17 poses a problem in the non-imputed data

EFA as its loadings in both factors are very similar. None of these factor solutions

97

suggested the structure of two different instruments (GIISS and GGMSS) that measure
two different developmental phases in individual gay identity and gay group membership.
Based on the adequacy of the sample and the number of iterations, the factor solution of
the imputed data appears to be slightly stronger.

Third step.

The third step in this series was to run a rotated orthogonal exploratory factor
analyses (EF A) with both the imputed and non-imputed data of the GIISS and the
GGMSS. Without imputed data, the extraction of factors was terminated after 999
iterations, while the factor solution with imputed data extracted 2 factors after 3
iterations. Table XXVIII shows the factor loadings of each item and their factor location.
Table XXVIII
Factor Loadings for Exploratory Factor Analysis With Varimax Rotation of A gosto ’s

(2004 ) 1 0 items from GIISS and GGMSS

 

 

 

I EF A Imputed data
tem Factor 1 Factor 2
Q #3 .731 .233
Q #8 .558 .042
Q #13 .468 .277
Q #14 .674 .294
Q #15 .473 .326
Q #16 .217 .731
Q #17 .453 .716
Q #18 .347 .488
Q #19 .036 .567
Q #20 .725 .184

 

The factor solution grouped 6 items in the first factor and 4 items in the second

factor. All of the items of the GIISS loaded in one factor with one item from the GGMSS.

98

The four items that loaded on the second factor belong to the GGMSS. This factor
solution did not suggested the structure of two different instruments (GIISS and GGMSS)
that measure two different developmental phases in individual gay identity and gay group
membership. According to Creswell (2005) criteria, all factor loadings are between
moderate and high.

Fourth step.

The fourth step in this series was to run a rotated oblique exploratory factor
analyses (EFA) with both the imputed and non-imputed data of the GIISS and the
GGMSS. Without imputed data, the extraction of factors was terminated after 999
iterations, while the factor solution with imputed data extracted 2 factors after 6
iterations. Table XXIX shows the factor loadings of each item and their factor location.
Table XXIX
Factor Loadings for Exploratory Factor Analysis With Oblimin Rotation of Agosto ’s

(2004) 10 items from GIISS and GGMSS

 

 

 

I EF A Imputed data
tem Factor 1 Factor 2
Q #3 .757 .020
Q #8 .612 -.133
Q #13 .453 .153
Q #14 .679 .105
Q #15 .446 .205
Q #16 .060 .731
Q #17 .327 .639
Q #18 .266 .423
Q #19 -.101 .610
Q #20 .763 -.032

 

99

The factor solution grouped 6 items in the first factor and 4 items in the second
factor. All of the items of the GIISS loaded in one factor with one item from the GGMSS.
The four items that loaded on the second factor belong to the GGMSS oblique
exploratory factor analyses. According to Creswell (2005) criteria, all factor loadings are
between moderate and high.

Fifth step.

The fifth step in this series was to run an orthogonal 2-factor exploratory factor
analyses (EFA) with both the imputed and non—imputed data of the GIISS and the
GGMSS. In these analyses, the factor solutions were forced to a maximum of two factors.
In both EFAs, the 2-factor solution was extracted after 3 iterations. Table XXX shows the
factor loading of each item and their factor location.

Table XXX
Factor Loadings for 2-F actor Exploratory Factor Analyses With Varimax Rotation of

Agosto 's (2004) 10 items from GIISS and GGMSS

 

 

 

 

 

I EFA Non-imputed Data EF A Imputed Data
tem Factor 1 Factor 2 Factor 1 Factor 2
Q #3 .695 .207 .731 .233
Q #8 .545 .038 .558 .042
Q #13 .397 .198 .468 .277
Q #14 .611 .240 .674 .294
Q #15 .381 .242 .473 .326
Q #16 .088 .722 .217 .731
Q #17 .365 .692 .453 .716
Q #18 .386 .481 .347 .488
Q #19 .091 .587 .036 .567
Q #20 .718 .112 .725 .184

 

100

In the EF A with non-imputed data 6 items loaded on the first factor and 4 items
loaded on the second factor. In the EF A with imputed data 6 items loaded on the first
factor and 4 items loaded on the third factor. However, these distributions do not
correspond to the initial distribution of items per process. The factor solution of both
EFAs grouped 5 items of the GIISS and 1 item of the GGMSS in one factor and 4 items
of the GGMSS in the second factor.

According to Creswell (2005) criteria, the factor loading in both EFAs are
between moderate and high.

Sixth step.

The sixth step in this series was to run an oblique 2-factor exploratory factor
analyses (EF A) with both the imputed and non-imputed data of the GIISS and the
GGMSS. In both non-imputed and imputed data EFAs, the 2-factor solution was
extracted after 6 iterations. Table XXXI shows the factor loading of each item and their

factor location.

101

Table XXXI
Factor Loadings for 2-F actor Exploratory Factor Analyses With Oblimin Rotation of

Agosto ’s (2004) 10 items from GIISS and GGMSS

 

 

 

 

 

I EF A Non-imputed Data EF A Imputed Data
tem Factor 1 Factor 2 Factor 1 Factor 2
Q #3 .714 .024 .757 .020
Q #8 .589 -.116 .612 -.133
Q#13 .389 .101 .453 .153
Q #14 .614 .084 .679 .105
Q #15 .361 .153 .446 .205
Q #16 —.072 .757 .060 .731
Q #17 .239 .645 .327 .639
Q #18 .312 .410 .266 .423
Q #19 -.O36 .609 -.101 .610
Q #20 .761 -.085 .763 -032

 

In the EFA with non-imputed data 6 items loaded on the first factor and 4
items loaded on the second factor. In the EF A with imputed data 6 items loaded on the
first factor and 4 items loaded on the third factor. However, these distributions do not
correspond to the initial distribution of items per process. The factor solution of both
EFAs grouped 5 items of the GIISS and 1 item of the GGMSS in one factor and 4 items
of the GGMSS in the second factor.

According to Creswell (2005) criteria, the factor loading in both EFAs are
between moderate and high.

Unidimensionality vs. bidimensionality of gay Identity

Before addressing the issue of dimensionality of the construct of gay identity, it is
important to decide which data base should be considered before reaching any

conclusions. Based on the factor loadings, it was concluded that all but one of the items

102

are good when it comes to measuring the construct. Item #7 (“I still get angry at the way
homosexuals are treated, but not as much as I once did.”) was the only item that obtained
very low factor loadings across the 18 exploratory analyses conducted in this study. This
item was removed from the data during the model modification phase.

When comparing the factor solutions of non-imputed data with the imputed data,
overall, the results of the imputed data were more stable with higher adequacy of
sampling and less iterations per exploratory factor analysis. As noted before, some of the
exploratory analyses conducted with the non-imputed data could not extract a factor
solution after 999 iterations.

When comparing the factor solutions of the items published by McCarn and
Miller (1996) with the items developed with this researcher, the items by McCarn and
Miller appear to be better measures of the gay identity construct. It is important to notice
that the determinant of the correlation matrix improved when the items developed by this
researcher were removed from the analyses. Furthermore, the exploratory factor analyses
that could not obtain a factor solution were all conducted with data from the items
developed by this researcher.

Therefore, the research question about the dimensionality of the gay identity
construct was answered based on the series of imputed-data exploratory factor analyses
based on the data from the items of McCarn and Miller (1996). From this series, none of
the factor solutions explained satisfactorily the structure proposed by the measures GIISS
and GGMSS. Actually, none of the 18 exploratory analyses in all three series explained
satisfactorily this structure. The best factor solution suggested was the orthogonal three

factor model that explained 63.3% of variance in the latent variable. When this model

103

was forced to two factors, the explained variance was reduced to 52.7%. Moreover, the
factor loadings did not support the structure of the GIISS and the GGMSS. Both factors
loaded items from both measures.

Even when some of the factor solutions grouped most of the items in one factor,
they failed to explain more than 50% percent of the variance of the latent trait. Although
we cannot make definite conclusions about the dimensionality of the construct, the
sample data suggested a multidimensional construct as the best solution includes three
independent factors.

Research Question #3: Do the responses to the items correlate with other constructs
in a manner that is consistent with the external model?

This research question was answered by executing an SEM analysis consisting of
a series of confirmatory factor analyses to test and modify the proposed full model. In
general, SEM analysis is a strategy to test how much of the proposed model actually fits
the sample data. In this study, the proposed model was tested by means of a confirmatory
analysis. The successive analyses represent an attempt to modify the model in a way that
best fits the sample data (model modification). A total of 9 confirmatory analyses were
run. To evaluate model’s fit to sample data, three indexes were examined: the goodness-
of-fit index (GFI), the adjusted goodness-of-fit for degrees of freedom (AGFI) and the
root mean square residual (RMR). An adequate fit is characterized by an RMR lower
than .10, and a GFI/AGFI higher than .95. The model modifications introduced in each
successive confirmatory analysis were determined by the p value. Those scales with p

values significantly different from .000 were removed in each successive step.

104

First Step.

Regardless of the reliability results, the first step in SEM is testing the full
proposed model. Therefore, this step consisted in conducting a confirmatory analysis to
test the full proposed model (see Figure 5, pp. 54-55). This model included two latent
independent variables (gay identity and self-concept) and two latent dependent variables
(ethnic identity and psychological well being). Gay identity had 8 observable variables
with their respective measurement errors: GIISS, GGMSS and the 6 subscales of the GIQ
(Stage I: Confusion, Stage 2: Comparison, Stage 3: Tolerance, Stage 4: Acceptance,
Stage 5: Pride, and Stage 6: Synthesis). Self-Concept had Stake’s Self-Concept Scales as
its 6 observable variables with their respective measurement errors: Likeability Scale,
Task Accomplishment Scale, Power Scale, Vulnerability Scale, Giftedness Scale and
Morality Scale. Ethnic identity had the MEIM subscales as its 4 observable variables with
their respective measurement errors: Identity Achievement Scale, Ethnic Behaviors
Scale, Affirmation and Sense of Belonging Scale and Other Group Orientation Scale.
Psychological well being had Ryff’s Scales of Psychological Well Being as its 6
observable variables with their respective measurement errors: Autonomy Scale, Personal
Growth Scale, Environmental Mastery Scale, Positive Relations With Others Scale,
Purpose in Life Scale and Self-Acceptance Scale.

With a RMR of .918, a GFI of .672 and an AGF I of .597, it was concluded that

the full proposed model did not fit the sample.

105

Second Step.

The second step includes the first modification to the original full proposed
model. This modification was based on both the reliability results and the exploratory
factor analyses conducted previously. The reliability results indicated that the scales
Stage 1: Confusion and Stage 5: Pride, were not reliable enough as they lacked internal
consistency. Therefore, these scales were removed as part of the first modification. On
the other hand, the exploratory factor analyses suggested that 9 out of Miller and
F assinger’s (1995) 10 items included in the GIISS and the GGMSS were the best items,
even though they were measuring one dimension, rather than the proposed two.
Therefore, this researcher grouped these items into one scale (Gay Identity Inventory —
G11) and included it in the model to substitute the GIISS and the GGMSS as they were

formulated (see Figure 6).

106

Figure 6
KEY:

GII — Gay Identity Inventory

GIQ-S2 -— Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 - Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

TAS — Task Accomplishment Scale

PS — Power Scale

VS -— Vulnerability Scale

GS — Giftedness Scale

MS - Morality Scale

IAS - Identity Achievement Scale

EBS — Ethnic Behaviors Scale

SBS - Sense of Belonging Scale

OGOS — Other-Group Orientation Scale

AS — Autonomy Scale

PGS — Personal Growth Scale

EMS — Environmental Mastery Scale

PROS — Positive Relations W/Others Scale

PLS - Purpose in Life Scale

SAS - Self-Acceptance Scale

107

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

801

1 rapow P9IIIP0W 491101411103 '9 942281.;

 

 

 

 

éé

 

 

 

ZIQ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11 116

 

 

 

 

V”)
60

 

 

 

 

 

 

 

 

 

 

 

013 svs sw
0/396'
6 UN
83 808d , ,/ SA
8' zrxx
. ‘_—k g d
89“ 9 01m 8V1
93 sv 6W S'I
/ 98-019
8’06
173 sooo / 178-019
911%
‘53 SBS __...——-————-* 88-019
SW
883 fi“ * “cs-019
vxx

 

 

 

 

 

 

ll

 

I At
4t

(Y

KY

KY

9M

th

811198119111 "Ms-1

 

”N

Z?

1
£1

Ill

8199901 0111103

/

l

801

119110141 11901110141 Irepow 1m '9 94228.15

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

sw
17m
so
cm
a S"
zm
1deouoonas ‘ - .1
IN S
0m S“
10)
98-019
8W
\ 178-019
I? ES'OID
9W
81111191111199 * 88-019
17w

 

 

 

IXY

601

 

 

taint: 090' 1761' 59d 3
4....- 991' 198' 8111981 119711 18018010908919 89d 8
#:124- 920' 1761' SW3 3
.4.- 988' 9881 801981 119111 1801801011084» 8713 Y
4:414: [90' L99 SV 9

0001 8111981 119m 19018010908889 sv Y
ZOO’ 190' L9 1' 8111981 11910 19913010110st 5
ZOO' 981' 999- A11111911181299 8111981 119M 19313010119st 1
an“: {170' 917?." SW 9
o...- 9EZ' L06' 1919011033199 SW 8
4..- 881' 1811 899
4....- SSV' ZIS'I 10190110331989 89 Y
tau: 681° [1721 SA?
L00' SZE' 681'1 1dGOUOD‘JPS-) SA 1L
41:11:11 Z01. LZ9° Sci 9
.4.- 8Z8' 681'1 1d93003'319S-) Sd “t
1700' 280' L88' svr 9
.4.- 8117' 99L'1 11193003211989 svr Y
m 6L0' 9817' S1 9

000'1 19190110991989 S’I Y
#414: L09 L91 8 9S'OID 9
.4.- 986' 891 '9 819091118899 9s-019 x
.4.- v8Z' 980'1 VS'OIE) 9
o... 808' 6SZ'S‘ 811111991 899-) 178-019 Y
o..- 801' LLV' 88-019 9
m 1799' 171798 8191191118999 88-019 Y
m 1701' 089' zs-019 9
o... 998' 2891- 811111991 £899 zs-Oro Y
#11111: [20' Z91. 1199

0001 4111119111 589-) 119 Y
1787' 910' z 10' 1daou09 1198 / 8111119131 £99 19

d 9181111183 pmpums 9181111183 191'.)me

 

1 [apozu pag/gpowfo sarowgzsa Jazaumxod

IIXXX 9199.].

serum d Sugpuodsanoo

119111 1111216 saieumsa Jaiotumed our 8116qu "xxx alqal 'oldtues 9111113 rou p1p [spout 81111

18q1p9pn101103 59’“ 1! ‘9L9' .10 139V “9 pun 6EL' 10 :119 9 ‘981'30 HWtI “'9 1111M

 

Table XXXII (cont.)

Parameter estimates of modified model I

 

 

 

Parameter Estimate Standard Estimate P
APROS {-Psychological Well Being 1.664 .279 ***
e PROS .334 .054 ***
A PLS é-Psychological Well Being 1.525 .240 ***
e PLS .156 .028 ***
A SAS (-Psychological Well Being 2.107 .316 ***
a SAS .116 .032 *** 1“
7 Ethnic Identity (—Gay Identity .201 .210 .340 ‘
1; Ethnic Identity .315 .078 ***
A SBS é-Ethnic Identity 1.000
a SBS .229 .045 ***
k IAS {-Ethnic Identity 1.252 .163 ***
a IAS .028 .048 .552
A EBS {-Ethnic Identity 1.192 .172 ***
e EBS .502 .085 ***
)1 OGOS éEthnic Identity .186 .107 .084
e OGOS .317 .047 ***

 

The next successive modifications were based on the P value. According to the
results, the P values of 8 estimates were higher than .000: the correlation between Gay
Identity and Self-Concept ((1) = .434), the measurement error of the Task
Accomplishment Scale (8 TAS = .004), the regression weight of the Vulnerability Scale
and the variable Self-Concept (k VS(-SC = .007), the regression weight of the structural
equation of Psychological Well Being and Gay Identity (y PWBé-GI = .002), the
measurement error of the variable Psychological Well Being (1; PWB = .002), the
regression weight of the structural equation of Ethnic Identity and Gay Identity (7 EK-Gl
= .340), the measurement error of the Identity Achievement Scale (8 IAS = .552) and the
regression weight of the Other Group Orientation Scale) and Ethnic Identity (1

OGOS(-El = .084). This researcher decided not to eliminate any latent variable at this

110

point as it was a too radical solution at this point. Eliminating either Self-Concept or
Ethnic Identity, would have removed between 4 and 6 scales at once. From the remaining
estimates, this researcher decided to keep in the model those scales with p values very
close to .000. Two P values were significantly high: 5 IAS = .552 and 71 OGOS(-El =
.084. One of these values is associated with the measurement error of the Identity
Achievement Scale (IAS) from the Multigroup Ethnic Identity Measure (MEIM) and the
other is the regression of the Other Group Orientation Scale (OGOS) of the MEIM. Even
though the measurement error has a higher P value, this researcher decided to keep the
IAS scale in the model and removed the OGOS to examine how this would affect the
measurement errors related to the MEIM.

Third Step.

The next modification consisted in removing the Other Orientation Group Scale

of the MELM from the model. Figure 7 presents the modified model.

111

Figure 7
KEY:

GII — Gay Identity Inventory

GIQ-82 — Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 — Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

TAS — Task Accomplishment Scale

PS - Power Scale

VS — Vulnerability Scale

GS — Giftedness Scale

MS — Morality Scale

IAS — Identity Achievement Scale

EBS — Ethnic Behaviors Scale

SBS - Sense of Belonging Scale

AS — Autonomy Scale

PGS — Personal Growth Scale

EMS — Environmental Mastery Scale

PROS — Positive Relations W/Others Scale

PLS — Purpose in Life Scale

SAS - Self-Acceptance Scale

 

112

GII

 

(1)1

 

 
   
   
 

Self Concept Psyc. Well Being

1x12 n2
We

1x13
@—

XX14
@—

Pigure Z Fuilhdodelzhdodified.h10delll

113

 

 

 

With an RMR of .139, a GIF of .752 and an AGFI of .688, it was concluded that

this model did not fit the sample. Table XXXIII shows the parameter estimates with their

corresponding P values.
Table XXXIII

Parameter estimates of modified model I]

 

 

Parameter Estimate Standard Estimate P

(1) Gay ldentity/ Self Concept .012 .016 .434
1. G11 {-Gay Identity 1.000

8 G11 .132 .021 ***
71 GIQ-S2 {-Gay Identity -1.582 .365 ***
8 GIQ-82 .680 .104 ***
7t. GIQ-S3 (-Gay Identity -3.643 .554 ***
8 GIQ-S3 .478 .108 ***
k GIQ-S4 {-Gay Identity -5.237 .803 ‘**
8 GIQ-S4 1.085 .234 ***
2» GIQ-S6 (-Gay Identity 6.161 .985 ***
8 GIQ-S6 2.167 .407 ***
1 LS é-Self-Concept 1.000

8 LS .485 .079 ***
k TAS {-Self-Concept 1.755 .418 ***
8 TAS .237 .082 .004
71 PS {-Self-Concept 1.139 .328 ***
8 PS .627 .102 ***
1 VS {-Self-Concept -1.026 .383 .007
8VS 1.241 .189 ***
1 GS {-Self-Concept 1.512 .437 ***
8GS 1.131 .183 ***
1 MS {-Self-Concept .907 .236 ***
8 MS .245 .043 ***
y Psychological Well Being (-Gay Identity -.565 .185 .002
C Psychological Well Being .157 .051 .002
2» AS {-Psychological Well Being 1.000

8 AS .337 .051 ***
1 EMS (-Psychological Well Being 1.835 .286 ***
3 EMS .194 .036 ***
l PGS (-Psychological Well Being .861 .165 ***
e PGS .194 .030 ***

 

114

Table XXXlll (cont.)

Parameter estimates of modified model 11

 

 

Parameter Estimate Standard Estimate P

XPROS (-Psychological Well Being 1.664 .279 ***
e PROS .334 .054 ***
A PLS (-Psychologica1 Well Being 1.525 .240 ***
e PLS .156 .028 ***
7» SAS (-Psychological Well Being 2.107 .316 ***
a SAS .116 .032 ***
7 Ethnic Identity (-Gay Identity .209 .219 .340
C Ethnic Identity .330 .079 ***
k SBS é-Ethnic Identity 1.000

a SBS .213 .044 ***
k IAS é-Ethnic Identity 1.191 .151 ***
e IAS .053 .045 .227
A EBS {-Ethnic Identity 1.193 .167 ***
e EBS .‘480 .083 ***

 

According to the results, the P values of 7 estimates were higher than .000: the
correlation between Gay Identity and Self-Concept (d) = .434), the measurement error of
the Task Accomplishment Scale (8 TAS = .004), the regression weight of the
Vulnerability Scale and the variable Self-Concept (k VS(-SC = .007), the regression
weight of the structural equation of Psychological Well Being and Gay Identity (7‘
PWB(-GI = .002), the measurement error of the variable Psychological Well Being (C
PWB = .002), the regression weight of the structural equation of Ethnic Identity and Gay
Identity (7 EI(-GI = .340) and the measurement error of the Identity Achievement Scale
(8 IAS = .227). These are the same results of the previous step with two exceptions: the
OGOS is not included in this model and the measurement error of the IAS dropped to
.227. Even though this P value is still significantly higher than .000, this researcher

decided to keep this scale in model for two reasons. First, from a theoretical perspective,

115

it is important to have in the model a scale that is analogous to the gay identity scales.
Second, since the P value of the measurement error dropped, it is important to examine if
removing other scales could lower this scale’s measurement error. As a result the next
higher P values were .007 obtained by the Vulnerability Scale and .004 obtained by the
measurement error of the Task Accomplishment Scale. Both scales were removed from
the model..
Fourth Step
The next modification consisted in removing from the model the Vulnerability "‘
and Task Accomplishment scales of the Ryff’s Scales of Self-Concept. Figure 8 presents

the modified model.

116

Figure 8
KEY:

G11 — Gay Identity Inventory

GIQ-S2 — Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 — Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

PS — Power Scale

GS — Gifiedness Scale

MS — Morality Scale

IAS — Identity Achievement Scale

EBS — Ethnic Behaviors Scale

SBS — Sense of Belonging Scale

AS — Autonomy Scale

PGS — Personal Growth Scale

EMS — Environmental Mastery Scale

PROS — Positive Relations W/Others Scale

PLS — Purpose in Life Scale

SAS - Self-Acceptance Scale

117

 

1x1
G11

71x4

1

GIQ—$2

GIQ-86

2919
a

1x11
GD

70x13
GS

1

1x14
S

a

Figure 8. Full Model: Modified Model 111

118

    
   

Gay Identity

5.1

    
   
 

Self Concept

E2

Yl

   
   

Psyc. Well Being

‘12

 

 

With an RMR of .117, a GIF of .775 and an AGFI of .709, it was concluded that

this model did not fit the sample. Table XXXIV shows the parameter estimates with their

corresponding P values.
Table XXXIV

Parameter estimates of modified model 111

 

 

Parameter Estimate Standard Estimate P

4) Gay Identity / Self Concept .027 .019 .115
2» G11 (-Gay Identity 1.000

8 G11 .132 .021 ***
A GIQ-$2 (-Gay Identity -1.577 .364 ***
8 GIQ-S2 .680 .104 ***
1. GIQ-$3 (-Gay Identity -3.61 l .549 ***
8 GIQ-S3 .492 .109 ***
k GIQ-S4 {-Gay Identity -5.226 .798 ***
8 GIQ-S4 1.081 .234 ***
2t GIQ-S6 (-Gay Identity 6.169 .981 ***
8 GIQ-S6 2.138 .404 ***
1 LS {-Self-Concept 1.000

8 LS .468 .086 ***
it PS {-Self-Concept 1.247 .371 ***
8 PS .557 .112 ***
A GS {-Self-Concept 2.027 .601 ***
8 GS .759 .223 ***
7» MS (-Self-Concept .568 .213 ***
8 MS .322 .052 ***
y Psychological Well Being {-Gay Identity -.569 .184 .002
C Psychological Well Being .157 .051 .002
1 AS {-Psychological Well Being 1.00

8 AS .336 .051 ***
1 EMS {-Psychological Well Being 1.834 .286 ***
8 EMS .194 .036 ***
k PGS (-Psychological Well Being .861 .165 ***
e PGS .194 .030 ***
XPROS (-Psychological Well Being 1.664 .279 ***
e PROS .334 .054 ***
it PLS (-Psychological Well Being 1.525 .240 ***
e PLS .156 .028 ***
2» SAS (-Psychological Well Being 2.107 .316 ***
8 SAS .116 .032 ***

 

119

Table XXXIV (cont.)

Parameter estimates of modified model 111

 

 

Parameter Estimate Standard Estimate P

7 Ethnic Identity (-Gay Identity .211 .218 .333
C Ethnic Identity .330 .079 ***
7t SBS é-Ethnic Identity 1.000

e SBS .213 .044 ***
l IAS é-Ethnic Identity 1.191 .151 ***
e IAS .053 .045 .236
7. EBS {-Ethnic Identity 1.193 .167 ***
a EBS .479 .083 ***

 

According to the results, the P values of 5 estimates were higher than .000: the
correlation between Gay Identity and Self-Concept (d) = .434), the regression weight of
the structural equation of Psychological Well Being and Gay Identity (7 PWB 6G1 =
.002), the measurement error of the variable Psychological Well Being (C PWB = .002),
the regression weight of the structural equation of Ethnic Identity and Gay Identity (7
E1(-Gl = .340) and the measurement error of the Identity Achievement Scale (8 IAS =
.227). These values exhibit some notable changes. The P values of correlation between
Gay Identity and Self-Concept and the regression weight of the structural equation of
Ethnic Identity and Gay Identity are smaller when compared with the P values obtained
in the previous step. The same is true for the measurement error of the Identity
Achievement Scale. Since the P value related with the IAS is still higher than .000, this
researcher decided to remove the scale from the model to examine its influence in the

regression weight of Ethnic Identity and Gay Identity.

120

Fifth Step.
The next modification consisted in removing from the model the Identity
Achievement Scale of the Multigroup Ethnic Identity Measure. Figure 9 presents the

modified model.

 

121

Figure 9
KEY:

GII — Gay Identity Inventory

GIQ-$2 - Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 - Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

PS — Power Scale

GS — Giftedness Scale

MS - Morality Scale

EBS — Ethnic Behaviors Scale

SBS — Sense of Belonging Scale

AS — Autonomy Scale

PGS — Personal Growth Scale

EMS - Environmental Mastery Scale

PROS —- Positive Relations W/Others Scale

PLS — Purpose in Life Scale

SAS - Self-Acceptance Scale

122

   
    
     
 
 

1x1
0

71x41 11
9 GIQ-82 Gay Identity

i1

GIQ—S3

Xx6

71x8

GIQ—S6

(D1

70x9
LS

11

 

kxll

    
   

p Psyc. Well Being

Self Concept

1x13 ”2

<32

1x14

Figure 9. Full Model: Modified Model IV

123

 

 

 

With an RMR of .119, a GIF of .770 and an AGF I of .7697, it was concluded that

this model did not fit the sample. Table XXXV shows the parameter estimates with their

corresponding P values.
Table XXXV

Parameter estimates of modified model I V

 

 

Parameter Estimate Standard Estimate P

<1) Gay Identity / Self Concept .027 .019 .153
71 GII (-Gay Identity 1.000

8 G11 .132 .021 ***
it GIQ-$2 (-Gay Identity -l.576 .363 ***
8 GIQ-$2 .681 .104 ***
71 GIQ-S3 (-Gay Identity . -3.608 .549 ***
8 GIQ-S3 .494 .109 ***
1 GIQ-S4 {-Gay Identity -5.226 .798 ***
8 GIQ-S4 1.080 .234 ***
it GIQ-S6 (-Gay Identity 6.171 .981 ***
8 GIQ-S6 2.135 .404 ***
71 LS é-Self-Concept 1.000

8 LS .467 .086 ***
it PS (-Self-Concept 1.247 .371 ***
8 PS .557 .112 ***
2» GS {-Self-Concept 2.026 .601 ***
8 GS .759 .223 ***
1. MS {-Self-Concept .568 .213 .008
8 MS .322 .052 ***
y Psychological Well Being {-Gay Identity -.570 .184 .002
C Psychological Well Being .157 .051 .002
1 AS (-Psychological Well Being 1.000

8 AS .336 .051 ***
7» EMS (-Psychological Well Being 1.834 .286 ***
8 EMS .194 .036 ***
7» PGS (-Psychological Well Being .861 .165 ***
e PGS .194 .030 ***
XPROS (-Psychological Well Being 1.664 .279 ***
e PROS .334 .054 ***
l PLS (-Psychological Well Being 1.525 .240 ***
e PLS .156 .028 ***
7» SAS é-Psychological Well Being 2.106 .316 **"‘
8 SAS .116 .032 ***

 

124

Table XXXV (cont.)

Parameter estimates of modified model IV

 

 

Parameter Estimate Standard Estimate P
7 Ethnic Identity ('Gay Identity .288 .267 .280
1; Ethnic Identity .629 1.042 .546
it SBS ('Ethnic Identity 1.000
e SBS -.O90 1.038 .931
k EBS (-Ethnic Identity .624 1.022 .542
e EBS .707 .417 .090

 

According to the results, the P values of 7 estimates were higher than .000: the
correlation between Gay Identity and Self-Concept (cl) = .153), the regression weight of
the structural equation of Psychological Well Being and Gay Identity (7 PWBé-GI =
.002), the regression weight of the Morality Scale and Self Concept (1 MS€-SC = .008),
the measurement error of the variable Psychological Well Being (C PWB = .002), the
regression weight of the structural equation of Ethnic Identity and Gay Identity (7 EIé-GI
= .542), the measurement error of the Sense of Belonging Subscale (8 SBS = .931) and
the measurement error of the Ethnic Behavior Scale (8 EBS = .090). The most salient
aspect of these results is the significant increase of the P values associated with scales
from the Ethnic Identity variable. Due to this situation, this researcher decided to remove
the latent variable Ethnic Identity from the model.

Sixth Step
The next modification consisted in removing from the model the latent variable.

Ethnic Identity. Figure 10 presents the modified model.

125

Figure 10
KEY:

GII — Gay Identity Inventory

GIQ-S2 — Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 - Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

PS — Power Scale

GS — Gifiedness Scale

MS — Morality Scale

AS — Autonomy Scale

PGS — Personal Growth Scale

EMS — Environmental Mastery Scale

PROS — Positive Relations W/Others Scale

PLS — Purpose in Life Scale

SAS - Self-Acceptance Scale

126

 

XXI
o

 

 
   
   
 

Self Concept

kxl3 é
@- '

1x14
@-

Figure 10. Full Model: Modified Model V

127

   
   

Psyc. Well Being

‘12

 

 

 

With an RMR of .119, a GIF of .803 and an AGFI of .731, it was concluded that

this model did not fit the sample. Table XXXVI shows the parameter estimates with their

corresponding P values.
Table XXXVI

Parameter estimates of modified model V

 

 

Parameter Estimate Standard Estimate P

(1) Gay Identity/ Self Concept .027 .019 .158
1 G11 (-Gay Identity 1.00

8 G11 .133 .021 ***
A GIQ-82 {-Gay Identity -1.577 .367 ***
5 GIQ-S2 .684 .105 ***
X GIQ-S3 (-Gay Identity -3.628 .558 ***
8 GIQ-S3 .499 .109 ***
it GIQ-S4 (-Gay Identity -5.290 .813 ***
8 GIQ-S4 1.056 .232 ***
k GIQ-S6 {-Gay Identity 6.244 .997 ***
8 GIQ-S6 2.104 .400 ***
1 LS (-Self-Concept 1.00

8 LS .468 .086 ***
A PS {-Self-Concept 1.248 .372 ***
8 PS .558 .112 ***
71 GS é-Self-Concept 2.031 .603 ***
8 GS .757 .223 ***
1 MS {-Self-Concept .569 .213 .008
8 MS .322 .052 ***
y Psychological Well Being (-Gay Identity -.565 .185 .002
C Psychological Well Being .158 .051 .002
it AS {-Psychological Well Being 1.000

8 AS .336 .051 ***
)1 EMS (-Psychological Well Being 1.834 .286 ***
8 EMS .194 .036 ***
1 PGS {-Psychological Well Being .861 .165 ***
e PGS .194 .030 ***
XPROS {-Psychological Well Being 1.664 .279 ***
e PROS .334 .054 ***
71 PLS {-Psychological Well Being 1.525 .240 ***
8 PLS .156 .028 ***
71 SAS é-Psychological Well Being 2.106 .316 ***
eSAS .116 .032 ***

 

128

According to the results, the P values of 4 estimates were higher than .000: the
correlation between Gay Identity and Self-Concept (d) = .158), the regression weight of
the structural equation of Psychological Well Being and Gay Identity (y PWB(-GI =
.002), the regression weight of the Morality Scale and Self Concept (1 MS(-SC = .008)
and the measurement error of the variable Psychological Well Being (C PWB = .002).
Based on these P values, this researcher decided to remove the Morality Scale fi'om the
model to examine its effect on the correlation between Gay Identity and Self-Concept.

Seventh Step.

The next modification consisted in removing from the model the Morality Scale

from the Stake’s Six Factor Self-Concept Scales. Figure 11 presents the modified model.

129

Figure 1 1
KEY:

GlI — Gay Identity Inventory

GIQ-$2 — Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 —- Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

PS — Power Scale

GS — Gifiedness Scale

AS — Autonomy Scale

PGS —— Personal Growth Scale

EMS — Environmental Mastery Scale

PROS - Positive Relations W/Others Scale

PLS - Purpose in Life Scale

SAS - Self-Acceptance Scale

130

 

Gay Identity

£1

70x9
9
a 1x11

 
   
   
 

Self Concept

5.2

Xx13
®
Figure I 1. Full Model: Modified Model VI
1 3 1

 
   
   

   
   

Psyc. Well Being

T12

 

 

With an RMR of .122, a GIF of .806 and an AGFI of .729, it was concluded that

this model did not fit the sample. Table XXXVII shows the parameter estimates with

their corresponding P values.
Table XXXVII

Parameter estimates of modified model VI

 

 

Parameter Estimate Standard Estimate P

(1) Gay Identity / Self Concept .026 .018 .150
1 G11 {-Gay Identity 1.000

8G1] .133 .021 ***
it GIQ-S2 é-Gay Identity -1.578 .367 ***
8 GIQ-82 .683 .105 ***
it GIQ-S3 ('Gay Identity -3.627 .557 ***
8 GIQ-S3 .499 .109 ***
it GIQ-S4 (-Gay Identity -5.287 .812 ***
8 GIQ-S4 1.058 .232 ***
it GIQ-S6 é-Gay Identity 6.244 .997 ***
8 GIQ-S6 2.102 .400 ***
)1 LS (-Self-Concept 1.000

8 LS .497 .089 ***
1 PS (-Self-Concept 1.333 .420 .002
8 PS .570 .121 ***
)1 GS {-Self-Concept 2.370 .829 .004
8 GS .650 .290 ***
y Psychological Well Being (-Gay Identity -.564 .185 .002
1; Psychological Well Being .158 .051 .002
1 AS (-Psychological Well Being 1.000

8 AS .336 .051 ***
)1 EMS (-Psychological Well Being 1.834 .286 ***
8 EMS .194 .036 ***
lt. PGS (-Psychological Well Being .861 .165 ***
e PGS .194 .030 **"‘
XPROS (-Psychological Well Being 1.664 .279 ***
e PROS .334 .054 ***
71 PLS {-Psychological Well Being 1.525 .240 ***
e PLS .156 .028 ***
it SAS {-Psychological Well Being 2.106 .316 ***
eSAS .116 .032 ***

 

132

 

According to the results, the P values of 5 estimates were higher than .000: the
correlation between Gay Identity and Self-Concept (d) = .150), the regression weight of
the structural equation of Psychological Well Being and Gay Identity (7 PWB(-GI =
.002), the regression weight of Power and Self Concept 01. PS(-SC = .002), the
regression weight of Gifiedness and Self Concept (A GS(-SC = .004) and the
measurement error of the variable Psychological Well Being (C PWB = .002). With the
removal of the Morality Scale, the P value of correlation between Gay Identity and Self-
Concept dropped to .150 from .158. Based on these results, this researcher decided to
remove the Gifiedness Scale.

Eighth Step
The next modification consisted in removing from the model the Giftedness Scale

from the Stake’sSix Factor Self-Concept Scales. Figure 12 presents the modified model.

133

Figure 12
KEY:

GII — Gay Identity Inventory

GIQ-82 — Gay Identity Questionnaire, Stage 2: Comparison
GIQ-S3 — Gay Identity Questionnaire, Stage 3: Tolerance
GIQ-S4 - Gay Identity Questionnaire, Stage 4: Acceptance
GIQ-S6 — Gay Identity Questionnaire, Stage 6: Synthesis
LS — Likeability Scale

PS — Power Scale

AS - Autonomy Scale

PGS - Personal Growth Scale

EMS — Environmental Mastery Scale

PROS — Positive Relations W/Others Scale

PLS — Purpose in Life Scale

SAS - Self-Acceptance Scale

134

 

0 GH
Xx4
Q GIQ-S2 Gay Identity

71x5

§
1

    

$1

 

 
   
   

Self Concept Psyc. Well Being

£2 112

71x11

@-

Figure 12. Full Model: Modified Model VII

135

 

 

With an RMR of .119, a GIF of .813 and an AGFI of .730, it was concluded that

this model did not fit the sample. Table XXXVIII shows the parameter estimates with

their corresponding P values.
Table XXXVIII

Parameter estimates of modified model VII

 

 

Parameter Estimate Standard Estimate P

(1) Gay Identity/ Self Concept .038 .028 .173
2» G11 (-Gay Identity 1.000

8 G11 .132 .021 ***
it GIQ-82 (-Gay Identity -l.583 .366 ***
8 GIQ-82 .680 .104 ***
it GIQ-S3 (-Gay Identity -3.636 .555 ***
8 GIQ-S3 .484 .108 ***
71 GIQ-S4 (-Gay Identity -5.244 .805 ***
8 GIQ-S4 1.083 .234 ***
it GIQ-S6 (-Gay Identity 6.188 .988 ***
8 GIQ-S6 2.143 .404 ***
2» LS {-Self-Concept 1.000

8 LS .434 .225 .054
it PS (-Self-Concept .905 .896 .313
8 PS .665 .202 ***
y Psychological Well Being (-Gay Identity -.567 .185 .002
C Psychological Well Being .157 .051 .002
2» AS (-Psychological Well Being 1.000

8 AS .336 .051 ***
1 EMS (-Psychological Well Being 1.835 .286 ***
8 EMS .194 .036 ***
it PGS (-Psychological Well Being .861 .165 ***
e PGS .194 .030 ***
XPROS {-Psychological Well Being 1.665 .279 ***
e PROS .334 .054 ***
71 PLS {-Psychological Well Being 1.525 .240 ***
a PLS .156 .028 ***
2» SAS (-Psychological Well Being 2.107 .316 ***
8 SAS .116 .032 ***

 

According to the results, the P values of 5 estimates were higher than .000: the

correlation between Gay Identity and Self-Concept ((1) = .173), the regression weight of

136

the structural equation of Psychological Well Being and Gay Identity (7 PWBé-GI =
.002), the regression weight of Power and Self Concept (7» PS(-SC = .313), the
regression weight of Giftedness and Self Concept (it GS(-SC = .004), the measurement
error of the variable Psychological Well Being (C PWB = .002) and the measurement
error of the Likeability Scale (8 LS = .054). These results suggested the removal of the
Self-Concept Scale from the model, which supported the Hypothesis lof this research
question. In this sample data, no significant relationship was observed between Gay
Identity and Self Concept.

Research Question #4: Do the correlation between the responses to the items and
other constructs exhibit significant differences based on ethnicity?

Since the SEM analysis suggested the removal of the latent variable Ethnic
Identity from the model, this research question could not be answered with this analysis
results. Moreover, since the model modification suggested that the items piloted were
measuring one factor, the results could not be used to answer Hypothesis 1 of this
question: A stronger positive relationship will be observed between the scores of the
Individual Identity — Synthesis scale and ethnic identity when compared with the
relationship between the responses to the Group membership — Synthesis Scale and
ethnic identity among gay men of color located in the Synthesis phase.

This researcher conducted a post hoc ANOVA analysis to examine if there was a
relationship between Gay Identity and Ethnic Identity. The Table XXXIX shows the

results of this analysis.

137

Table XXXIX

ANO VA of Ethnic Identity With Gay Identity As Dependent Variable

 

 

Source Sum Of Df Mean Square F Sig.
Squares

Corrected Model 2.760 4 .690 3.332 .014
Intercept 9.894 1 9.894 47.779 .000
Sense of belonging .374 1 .374 1.805 .183
Identity Achievement .040 1 .040 .193 .661
Ethnic Behaviors .474 1 .474 2.289 .134
Other Group Orientation .829 l .829 4.001 .049
Error 18.430 89 .207
Total 1077.245 94
Corrected Total 21.190 93

 

With the exception of the intercept, none of the F tests were significant. Therefore, it can
be concluded that, in this sample data, there is not a significant relationship between gay

identity and ethnic identity.

138

CHAPTER IV
Discussion

The purpose of this study was to empirically test a part of McCarn and
F assinger’s Sexual Minority Identity Formation Model. More specifically, this study
preliminarily tested the Gay Individual Identity — Synthesis (GIISS) and the Gay Group
Membership — Synthesis (GMSS) scales, each one containing 10 items purported to
measure the Synthesis phases in the Individual Identity Formation and Group
Membership processes depicted by McCarn and F assinger’s (1996) model.

The first research question was: Are the items reliable and precise enough to grant
further development of the items for research and clinical uses? Results indicated that all
but two measurement instruments (Stage 1: Confusion and Stage 2: Pride) have good
internal consistency and consist of items that are reliable enough. The reliability results
indicated that the items from the piloted measures GIISS and the GGMSS are consistent
and reliable.

The second research question was: Is gay identity a bidimensional construct that
comprises individual identity formation and group membership formation? The results
suggested the presence of three factors. Moreover, the analyses indicated that the items of
the GIISS and the GGMSS were not grouped as expected. Each factor grouped items
from both the GIISS and the GGMSS. This research question had a related hypothesis:
Significant differences will be observed between the scores of the Gay Individual Identity
— Synthesis Scale (GIISS) and Gay Group Membership — Synthesis Scale (GGMSS). The
results did not support this hypothesis. The results from the factor analysis suggested that

McCarn and Miller’s items included in both the GIISS and the GGMSS measure only one

139

dimension of gay identity, whereas the items developed by this researcher measure two
dimensions. However, these dimensions did not correspond to individual gay identity and
gay group membership as the factors grouped items from both the GIISS and the
GGMSS.

The third research question was: Do the responses to the items correlate with
other constructs in a manner that is consistent with the external model? This research
question had four related hypotheses. The first hypothesis was that no significant
relationship will be observed between the scores of the Gay Individual Identity —
Synthesis Scale (GIISS) and the Gay Group Membership — Synthesis Scale (GGMSS)
and the scores of the Six-F actor Self-Concept Scale. The results from the SEM analyses
supported this hypothesis, as they suggested that Self Concept should have been removed
from the model. Furthermore, the correlation between Gay Identity and Self Concept in
every step of the SEM suggested a very low relationship between these variables. The
second hypothesis was that a significant positive relationship will be observed between
the responses to the items included in the Gay Individual Identity - Synthesis Scale and
the Psychological Well Being Scales. This hypothesis could not be answered with the
sample data because it did not support the structure of the measure GIISS and GGMSS.
This situation was similar for the third hypothesis: A significant positive relationship will
be observed between the scores of the Gay Group Membership - Synthesis Scale and the
Psychological Well Being Scales. However, the modified model suggested a possible
relationship with Psychological Well Being. Interestingly, the relationship purported by
this model is not positive but inverse. The fourth hypothesis was that differences in the

scores in the Gay Group Membership — Synthesis Scale will be observed based on the

140

scores in the Multigroup Ethnic Identity Measure. This hypothesis could not be tested
with the sample data as it did not support the presence of the variable Ethnic Identity in
the model. This suggested that, in the case of the sample data, there is little or no
relationship between gay identity and ethnic identity.

The fourth research question was: Do the correlation between the responses to the
items and other constructs exhibit significant differences based on ethnicity? This
question had a related hypothesis: A stronger positive relationship will be observed
between the scores of the Individual Identity — Synthesis scale and ethnic identity when
compared with the relationship between the responses to the Group membership —
Synthesis Scale and ethnic identity among gay men of color located in the Synthesis
phase. This hypothesis could not be tested because the sample data did not support the
structure of the GIISS and the GGMSS. Furthermore, the results suggested that gay
identity and Ethnic Identity are not correlated. Still, a post hoc analysis was run to
examine any possible relationship between these two variables. However, these results
did not support a relationship between these variables in the sample data.

The reliability results indicated the items included in the GIISS and the GGMSS
are items reliable and precise enough to grant further development of the items for
research and clinical uses. However, the item #7 (“I still get angry at the way
homosexuals are treated but not as much as once 1 did.”) exhibited some problems
regarding its factor loadings. during the exploratory factor analyses. This result might be
explained in terms of the item complexity. This item actually contains two ideas (i.e.
“getting angry” and “but not as much”) and a negation particle (“not”). Also, it is

possible that the sample participants may have had some problems with the item because

141

 

the way it is redacted may be misconstrued as not caring enough for other gay men. In
other words, it could be read as if once the individual used to care for other gay men but
now that is not necessarily true.

The results from the exploratory factor analyses did not support the idea of gay
identity as a bidimensional construct. The best description of this construct according to
the sample data is a three-dimensional variable. Furthermore, the item location in factors
did not support the structure of the GIISS and the GGMSS. An examination of the mean
scores of these two instruments revealed that the sample appeared to be in the higher
levels of gay individual identity and gay group membership development. Furthermore, a
considerable majority of the participants had already done some type of disclosure about
their sexual orientation. It is possible that the responses to the items of both GIISS and
GGMSS were highly correlated due to characteristics of the sample while picking up
some other undocumented differences resulting in the presence of three dimensions.

The results indicating a lack of significant relationship between Self Concept and
Gay Identity are consonant with this researcher’s expectations. As it was noted
previously, self concept is a broader construct, while identity is more specific and related
to particular situations (Troiden, 1984). The scarce research done in this area (Larson,
1985) revealed that no significant changes occur in self concept based solely on sexual
orientation. An individual has only one self concept constituted by multiple identities.
Therefore, elements like self concept complexity (how many identities constitute the self
concept) and self concept clarity (how defined is the self concept) may have more

importance than a single identity.

142

The results suggested an inverse relationship between gay identity and
psychological well being. This is probably one of the most written about and researched
areas in the field of LGBT issues. Previous theoretical papers and published research
findings have suggested a positive relationship between gay identity and psychological
well being based on the negative effects of internalized homophobia (DeLuca, 1967;
Dunkle, 1994; Hooker, 1965; Halpin & Allen, 2004; Igartua, Gill & Montoro, 2003;
Kertzner, Meyer & Frost, 2009; Miranda & Storms, 1989; Sagir, Robins, Walbran &
Gentry, 1970; Szymanski & Gupta, 2009). However, this work paid little attention to the
role of social support and the perception by members of the LGBT community. This
situation may bring additional stressors as these participants may or may not have enough
family or social support. Being gay has been conceptualized as a minority status.
Actually, McCarn and Fassinger’s model is called Sexual Minority Identity Formation
model. Along this line, a gay man can experience what is known as minority stress. The
visibility as a member of the LGBT community may add stressors that undermine the
person’s perceived psychological well being. As it was previously stated, the sample
appears to be in the higher levels of gay identity development and has already engaged in
some form of public disclosure of its sexual orientation. It is possible, then, that the
participants in this study may feel lack of adequate social support, which translates into
feelings of stress detrimental to their well being.

The results did not suggest a relationship between gay identity and ethnic identity.
Even though the sample reported higher scores in the gay identity measures and lower
scores in the ethnic identity subscales, this difference did not translate into an inverse

relationship between these two variables among gay men of color suggested by the scarce

143

published theoretical papers. One possible explanation is that the Multigroup Ethnic
Identity Measure may not be a good instrument to measure ethnic identity within gay
men of diverse ethnic extractions. Examining the perceptions of the importance of
perceived ethnic identity was beyond the scope of this study. However, since the post hoc
analysis examined differences based on ethnicity, no significant differences should be
expected if the participants did not give much importance to this aspect, either because
they do not think about it or they decided to be gay over belonging to their ethnic
community. Another explanation for this result is that it may be a consequence of the
sample size. Even though the confirmatory analyses were run by the software program,
the goodness-of-fit index of the model did not reach the desired levels. For confirmatory
factor analysis, a ratio of 5 participants per item is recommended. However, since this
investigation is a pilot study, the sample size of these analyses was only 94. This may
have introduced statistical problems that are reflected in an apparent lack of relationship
between these two variables.
Implications

The results of this pilot study indicate the need for further research to examine
issues of dimensionality of gay identity as a construct. Some evidence was found to
support the idea of gay identity as a multidimensional construct and its relationship to
psychological well being. Furthermore, as expected, it suggested little or no relationship
with the construct of self concept. However, issues related with sample size prevented
this researcher from making any definitive conclusions. It is important to remember that

this study was a pilot conducted to test a research design.

144

The sample data did not support the proposed full model, specifically the presence
of ethnic identity. Again, sample size and its influence on the statistical analyses may
have affected this result. Furthermore, the fact the instruments used to measure gay
identity were either in development (GIISS and GGMSS) or had exhibited problems in
previous studies (GIQ), calls for caution when reaching conclusions about this
relationship. The scarce qualitative research on gay identity and ethnic identity suggests
an interaction between these two variables. Solid measurement instruments may tap into
this relationship in ways that findings can generalize to a larger population.

In terms of applied implications, it is too soon to ascertain that the piloted scales
are an adequate measure of individual gay identity and gay group membership status.
However, results suggested that the relationship between gay identity and psychological
well being might be a complex one that requires openness and a critical attitude of our
own perceptions when working with LGBT clients. More than following a particular
model of gay identity development, it is important that the therapist becomes aware of
how other identities may interact with that developmental process.

Limitations

This pilot study had some limitations that require attention. One of these
limitations was sample size. The analyses required a ratio of 5 participants per item. With
205 items to answer, that would have required a sample of, at least, 1,025 participants.
However, since this was a pilot study, this researcher did not recruit that number of
participants. The final sample for this study was constituted by 94 participants, which
barely provided adequate variance for running analyses. The intention of conducting a

pilot study was to test the research design to examine its adequacy and suggesting

145

possible improvements. Therefore, caution should be exerted in generalizing these results
to the population. Another limitation is related to the recruitment technique. Using a
website to collect the data did not improve significantly the possibility of recruiting a
more varied sample. The demographic data showed that the sample was located in the
higher levels of gay identity development at the moment of data collection. Finally, the
instruments used to measure gay identity were either under development (GIISS and
GGMSS) or have presented problems in previous studies. There are only two instruments
to measure gay identity development. Cass (1984) developed an instrument when testing
her model but warned against its use as it did not support her model. Brady and Busse
(1994) developed the GIQ based on Cass’s model. This instrument has not been
frequently used as it may have some validity issues. However, this researcher included
this instrument in the research design because it was the only one with which the GIISS
and the GGM SS could be compared. Since the GIQ has a binomial scale (only one of two
possible responses per item), variance may be compromised.
Future directions

This study was a pilot to develop the GIISS and the GGMSS. In terms of the
development of the piloted measures, it is important to recruit a sample with bigger size
and more variance in order to improve the statistical analyses. Also, it is recommended to
step back and start by testing a simpler model including only psychological well being as
a variable. Once there is a sufficiently solid instrument to measure gay identity, then the
relationship with ethnic identity may be examined from a quantitative perspective.
Certainly more exploratory analyses of the piloted measures with bigger samples are

needed before conducting confirmatory analyses. Finally, a longer period of data

146

collection combining various techniques would be recommended to increase the size of
the sample. These techniques will have to consider the physical presence of the
researcher in the process of collecting data. For example, visiting LGBT interests groups
will imply having a face attached to a name, which can facilitate data collection through
the snowballing technique. Also conducting a focus group with members of the
community may foster an exchange of ideas that, when implemented, will improve the
research design, particularly the data collection process.

From a theoretical perspective, once the measures are developed, it is important to
further examine the relationship between gay identity and ethnic identity. Since the
research in this area is currently scarce, future studies should combine quantitative and

qualitative techniques in a way their results can inform each other.

147

Appendix A
GIISS and GGMSS

Gay Individual Identity Scale — Synthesis and Gay Group Membership Scale - Synthesis

1— I feel a deep contentment about my love for other men. (Individual)

a. Strongly Agree (1)

b. Agree (2)

c. Disagree (3)

d. Strongly Disagree (4)

2- I love and appreciate myself as a man that loves and is sexual with other men.
(Individual)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

3- Loving and being sexual with other men is part of how a see myself.
(Individual)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

4- My love for men is an important part of me, but is not the only thing that
defines me. (Individual)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

5- I have successfiilly incorporated my love of men into my life. (Individual)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

148

10-

11-

I’m doing what I want to do in terms of love and sex; that makes me feel more
integrated as a person. (Individual)

a. Strongly Agree (1)

b. Agree (2)

c. Disagree (3)

d. Strongly Disagree (4)

I still get angry at the way homosexuals are treated, but not as much as once I
did. (Group)

Strongly Agree (1)

Agree (2)

Disagree (3)

Strongly Disagree (4)

i”

999'

Being gay or straight is one central part of who people are. (Group)
Strongly Agree (1)

Agree (2)

Disagree (3)

Strongly Disagree (4)

9’

9.0.6

Some straights are homophobic, some are not. (Group)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

I can relate comfortably to gays and nongays. (Group)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

I rely on my gay/lesbian friends for support but have some good straight friends
as well. (Group)

a. Strongly Agree (1)

b. Agree (2)

c. Disagree (3)

d. Strongly Disagree (4)

149

12-

13-

14-

15-

16-

17-

I am a member of the gay community. (Group)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

I don’t feel guilty about my attraction to and love for other men. (Individual)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

Loving and being sexual with men is part of who I am. (Individual)
Strongly Agree (1)

b. Agree (2)

c. Disagree (3)

d. Strongly Disagree (4)

9’

Being in a relationship with another man is important to me, but it is not the
only thing that defines me. (Individual)

a. Strongly Agree (1)

b. Agree (2)

c. Disagree (3)

d. Strongly Disagree (4)

I feel comfortable interacting with both straight and gay people. (Group)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

I enjoy having straight and gay friends. (Group)
Strongly Agree (1)

Agree (2)

Disagree (3)

Strongly Disagree (4)

9’

P-F’P‘

150

18- I think it is important for me to get along with straight and gay people. (Group)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

19- When I meet somebody I don’t think solely of him/her in terms of hi/her sexual
orientation. (Group)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

20— I feel proud of being gay, as I am proud of other aspects/characteristics of me.
(Individual)
a. Strongly Agree (1)
b. Agree (2)
c. Disagree (3)
d. Strongly Disagree (4)

151

Appendix B
Informed Consent

. We would like you to participate in a research study titled “ A pilot study to
develop the Gay Individual Identity — Synthesis (GIISS) and the Gay Group
Membership — Synthesis (GGMSS) scales for gay men based on McCarn and

F assinger’s (1996) Sexual Minority Identity Formation Model”. The purpose of
this study is to develop an instrument that will measure gay identity development.
We are also interested in examining how gay identity development is related to
how gay men feel about themselves and how they relate to friends and family. By
participating in this research you will help to develop a measure that will enable
mental health practitioners to identify the specific needs of gay clients in ways
that will foster a healthy gay identity development.

. If you decide to participate in the study, your involvement will take about 40-45
minutes of your time. You will be asked to answer the questionnaires included
this website.

. We do not anticipate that your participation will result in any risk for you. Your
participation will be completely voluntary and you will be free to refuse or stop at
any time without penalty.

All information will be number-coded and strictly confidential. Your privacy will
be protected to the maximum extent allowable by law.

. At the end of the survey, you will have the opportunity to enter a raffle for one of
three $50.00 cash prizes. You only have to provide a valid email address so the
researcher could contact you. These email addresses are collected in a different
webpage to protect your anonymity.

. If you have any questions about this study, please, feel free to contact us:

 

 

 

Gloria Smith, Ed.D. Ariel I. Agosto-Cepeda, M.Ed.
433 Erickson Hall 433 Erickson Hall

CEPSE Department CEPSE Department

School of Education School of Education

Michigan State University Michigan State University
Phone (517) 355-8508 Phone: (517) 355-3053

 

152

 

7. If you have any questions or concerns regarding your rights as a study participant,
or are dissatisfied at any time with any aspect of his study, you may contact-
anonymously, if you wish, Ashir Kumar, M.D., Chair of the University
Committee on Research Involving Human Subjects (UCHRIS) by phone: (517)
432-4503, email: uchris@msu.edu, or regular mail: 202 Olds Hall, East Lansing
MI 48823.

 

8. Before you leave this window please print a copy of this informed consent for
your records.

9. Click on Accept to start participating in the study.

10. Thank you for your cooperation.

153

REFERENCES

Beals, K.P., Peplau, L.A. & Gable, S.L. (2009). Stigma management and well-being: The
role of perceived social support, emotional processing, and suppression.
Personality and Social Psychology Bulletin, 35, 867-879.

Berger, R. M. & Mallon, D. (1993). Social support networks of gay men. Journal of
Sociology and Social Welfare, 20, 155-174.

Bieschke, K. J ., Eberz, A. B., Bard, C. C. & Croteau, J. M. (1998). Using social cognitive
career theory to create affirmative lesbian, gay, and bisexual research training
environments. Counseling Psychology, 26, 735-753.

Brady, S. & Busse, W. J. (1994). The gay identity questionnaire: A brief measure of
homosexual identity formation. Journal of Homosexuality, 26, 1-22.

Cass, V. C. (1979). Homosexual identity formation: A theoretical model. Journal of
Homosexuality, 4, 218-235.

Cass, V. C. (1984). Homosexuality identity formation: Testing a theoretical model.
Journal of Sex Research, 20, 143-167.

Chapman, B. E. & Branock, J. C. (1987). Proposed model of lesbian identity
development: An empirical examination. Journal of Homosexuality, 14, 69-80.

Creswell, J. (2005). Educational Research: Planning, conducting, and evaluating
quantitative and qualitative research. Upper Saddle River: Pearson.

Coleman, E. (1982). Developmental stages in the coming out process. In J. C. Gonsoriek

(Ed.), A guide to psychotherapy with gay and lesbian clients. New York:
Harrington Park Press.

Deluca, J. N. (1967). The structure of homosexuality. Journal of Projective Techniques
and Personality Assessment, 30, 187-191.

Dunkle, J. H. (1994). Counseling gay male clients: A review of treatment efficacy
research l975-present. Journal of Gay and Lesbian Therapy, 2, 1-19.

F assinger, R. E. (1991). The hidden minority: Issues and challenges in working with
lesbian women and gay men. Counseling Psychologist, 19, 157-176.

F assinger, R. E. & Miller, B. A. (1996). Validation of an inclusive model of sexual
minority identity formation on a sample of gay men. Journal of Homosexuality,
32, 53-78.

Gamets, L. D. & Kimmel, D. C. (1993). Psychological perspectives on lesbian and gay

154

male experiences. New York: Columbia University Press.

Goode-Cross, D.T. & Good GE. (2009). Managing multiple-minority identities: African
American men who have sex with men at predominately white universities.
Journal of Diveristy in Higher Education, 2, 103-112.

Grossman, A. H. & Keener, M. S. (1998). Support networks of gay male and lesbian
Youth. Journal of Gay, Lesbian, and Bisexual Identity, 3, 27-46.

Grov, C., Bimbi, D.S., Nanin, J .E. & Parsons, J .T. (2006). Race, ethnicity, gender and
generational factors associated with the coming-out process among gay, lesbian,
and bisexual individuals. Journal of Sex Research, 43, 1 15-121.

Halpin, S.A. & Allen, M.W. (2004). Changes in psychosocial well-being during stages of
gay identity development. Journal of Homosexuality, 47, 109-126.

Harper, G.W., J emewall, N. & Zea (2004). Giving voice to emerging science and theory
for lesbian, gay and bisexual people of color. Cultural Diversity and Ethnic
Minority Psychology, 10, 187-199.

Hooker, E. (1957). The adjustment of the male homosexual. Journal of Projective
Techniques, 21, 18-31.

Igartua, K.J., Gill K. & Motoro, R. (2003). Intemalized homophobia: A factor in
depression, anxiety, and suicide in the gay and lesbian population. Canadian
Journal of Community Mental Health, 22, 15-30.

Jamil, 08., Harper, G.W. & Fernandez, M.I. (2009). Sexual and ethnic development
among gay-bisexual-questioning (GBQ) male ethnic minority adolescents.
Cultural Diversify and Ethnic Minority Psychology, 1 5, 203-214.

Jordan, K. M. & Deluty, R. H. (1995). Clinical interventions by psychologists with
lesbians and gay men. Journal of Clinical Psychology, 51, 448-456.

Kaflra, G. J. & Kozma, A. (2002). The construct validity of Ryff’s Scales of
psychological Well-Being (SPWB) and their relationship to measures of
subjective well-being. Social Indicators Research, 57, 171-202.

Kertzner, R.M., Meyer, I.H., Frost, D.M. & Stirrat, M.J. (2009). Social and psychological
well-being in lesbians, gay men, and bisexuals: The effects of race, gender, age,
and sexual identity. American Journal of Orthopsychiatry, 79, 500-510

Lark, J. S. & Croteau, J. M. (1998). Lesbian, gay, and bisexual doctoral students’

mentoring relationships with faculty in counseling psychology: A qualitative
study. Counseling Psychologist, 26, 754-776.

155

Levy, E. F. (1989). Lesbian motherhood: Identity and social support. Aflilia, 4, 40-53.

Linacre, J. M. (1998). Detecting multidimensionality: Which residual data-type works
best? Journal of Outcome Measurement, 2, 266-283.

McCarn, S. R. and Fassinger, R. E. (1996). Revisioning sexual minority identity
formation: A new model of lesbian identity and its implications for counseling
and research. Counseling Psychologist, 24, 508-534.

Messick, S. (1995). Validity of psychological assessment: Validation inferences from
persons’ responses and performances as scientific inquiry into score meaning.
American Psychologist, 50, 741-749.

Millman, J. & Greene, J. (1993). The specification and development of tests of
achievement and ability. In R. L. Linn (Ed.) Educational measurement, third
edition @p. 335-366). Phoenix: Oryx Press.

Minton, H. L. & McDonald, G. J. (1984). Homosexual identity formation as a
developmental process. Journal of Homosexuality, 9, 91-104.

Miranda, J. & Storms, M. (1989). Psychological adjustment of lesbians and gay men.
Journal of Counseling and Development, 68, 41-45.

Moe, J .L., Dupuy, P.J. & Laux, J.M. (2008). The relationship between LGBQ identity
development and hope, optimism, and life engagement. Journal of LGBT Issues in
Counseling, 2, 199—215.

Nesmith, A. A., Burton, D. L. & Cosgrove, T. J. (1999). Gay, lesbian and bisexual youth
and young adults: Social support in their own words. Journal of Homosexuality,
37, 95-108.

Phillips, J. C. & Fischer, A. R. (1998). Graduate students’ training experiences with
lesbian, gay and bisexual issues. Counseling Psychologist, 26, 712-734.

Phinney, J. S. (1992). The Multigroup Ethnic Identity Measure: A new scale for use with
diverse groups. Journal of Adolescent Research, 7, 156-176.

Procidiano, M. E. & Heller, K. (1983). Measures of perceived social Support from
friends and family: Three validation studies. American Journal of Community
Psychology, 11, 1-24.

Rosario, M., Schrimshaw, E.W. & Hunter, J. (2004). Ethnic/racial differences in the
coming-out process of lesbian, gay, and bisexual youths: A comparison of sexual

identity development over time. Cultural Diversity and Ethnic Minority
Psychology, 1 0, 21 5-228.

156

Ryff, C. D. (1989). Happiness is everything, or is it? Explorations on the meaning of
psychological well-being. Journal of Personality and Social Psychology, 5 7,
1069-1081.

Ryff C. D. & Keyes C. L. M. (1995). The structure of psychological well-being revisited.
Journal of Personality and Social Psychology, 69, 719-727.

Saghir, M. T., Robins, E., Walbran, B. & Gentry, K. A. (1970). Homosexuality: III.
Psychiatric disorders and disability in the male homosexual. American Journal of
Psychiatry, 126, 197-215.

Scasta, D. (1998). Historical perspectives on homosexuality. Journal of Gay and Lesbian
Psychotherapy, 2, 3-17.

Schmutte, P. S. and Ryff, C. D. (1997). Personality and well-being: Reexamining
methods and meanings. Journal of Personality and Social Psychology, 73, 549-
559.

Smith, A. (1997). Cultural diversity and the coming-out process. In B. Greene (Ed.)
Ethnic and cultural diversity among lesbians and gay men mp. 27 9—3 00). London:
Sage.

Sophie, J. (1985/1986). A critical examination of stage theories of lesbian identity
formation. Journal of Homosexuality, 12, 39-51.

Stake, J .E. (1994). Development and validation of the Six-factor Self-Concept Scale for
adults. Educational and Psychological Measurement, 54, 56-72.

Szymanski, D. M. & Gupta, A. (2009). Examining the relationship between multiple
internalized oppressions and African American lesbian, gay, bisexual, and
questioning person’s self-esteem and psychological distress. Journal of
Counseling Psychology, 56, 1 10-1 18.

Troiden, R. R. (1984). Self, self-concept, identity, and homosexual identity: Constructs in
need of definition and differentiation. Journal of Homosexuality, 10, 97-109.

Troiden, R. R. (1989). The formation of homosexual identity. Journal of Homosexuality,
1 7, 43-73

157

 

”111111111111[111111111111111I1’