Testing a Mediated Model of Modern Homonegativity
Kyle M. Erwin, M.A.
Department of Psychology & Philosophy, Texas Woman’s University,
Denton, TX
Jeffrey D. Strain, Ph.D.
Lake’s Crossing Center, Reno, NV
METHODS
INTRODUCTION RESULTS
PURPOSE & HYPOTHESES
HYPOTHESES
PARTICIPANTS (n = 526)
Ages among participants ranged from 18-72 (m = 33.9)
Race/Ethnicity:
383 (72.8%) White
60 (11.41%) Black
36 (6.84%) Latino or Hispanic
17 (3.23%) Biracial or Multiethnic
10 (1.9%) East Asian or Asian American
10 (1.9%) South Asian or Asian Indian
5 (< 1%) American Indian or Alaskan Native
4 (< 1%) Middle Eastern or West Asian
1 (< 1%) Native Hawaiian or Pacific Islander
Sex:
73 (13.9%) participants were men,
453 (86.1%) were women.
The average annual income among participants was $32,739
66% (n = 346) had earned at least one UG college degree
Religious affiliation:
281 (53.42%) Christian
74 (14.06%) indicated that they were spiritual
58 (11.03%) Catholic
44 (8.36%) Agnostics
27 (5.13%) Atheists
13 (2.47%) Buddhists
9 (1.71%) Jewish
7 (1.33%) Unitarian-Universalism
5 (.95%) Pagan/Neo-Pagan, Primal-indigenous, or Taoist
4 (.76%) Hindus
4 (.76%) Muslim
MEASURES
State-Trait Anger Expression Inventory-2 (STAXI-2; Spielberger)
Three Domain Disgust Scale (TDDS; Tybur, 2009)
Multidimensional Heterosexism Inventory (MHI; Walls, 2008)
PROCEDURES
Participants were recruited on social networking sites (e.g.,
Facebook) and list-serves (e.g., student, faculty).
ANALYSES
Mean, standard deviations, and zero order correlations for the 21
observed variables
Structural Equation Modeling (using LISREL 8.80)
STRUCTURAL MODEL
Again, we used the two-step method of SEM proposed by
Anderson and Gerbing (1988) to test the mediation effects for
our hypothesized model (see Figures 1 and 2). The final model
demonstrated robust goodness of fit (SRMR = 0.046, RMSEA =
0.039, CFI = 0.98). Moreover, examination of the structural
paths indicated that all paths were significant. A CFI greater
than .90 demonstrates an acceptable fit. According to Hu and
Benter (1999), and other researchers (e.g., Martens, 2005),
RMSEA values less than .06, and SRMR values less than .08,
indicate a good fit. Despite fit indices indicating a more than
acceptable fit, we recognize that absolute cutoff values to
accept or reject models fall short of their intended purposes,
and fit indices should be used to describe how well a model fits
the data (Quintana & Maxwell, 1999).
To analyze the data, Anerson and Gerbing’s (1988) two-
step method of structural equation modeling was followed (see
Table 2). Initially, a confirmatory factor analysis (CFA) was
conducted to test whether the measurement model fit the data.
After, acceptable fit of the measurement model was established,
the structural model was tested. In this study, both the
measurement model and structural model were tested using the
LISREL 8.80. Both procedures used the maximum likelihood
method. Three indexes were used to determine the goodness of
fit for the model (Hu & Bentler, 1999): the comparative fit index
(CFI), the root-mean square error of approximation (RMSEA),
and the standardized root-mean-square residual (SRMSR). It is
important to note that our sample size is appropriate for testing
our hypothesized model and meets suggested
recommendations (Hatcher, 1994). The results of the CFA
indicated an acceptable fit to the data Chi-square (df = 159, N =
526) = 288.56, p < .01; scaled Chi-square = 302.10, p < .01, CFI
= 0.98; RMSEA = 0.039 (90% confidence interval = 0.032;
0.047); SRMR = 0.046. Besides the fit of the measurement
model, the factor loadings of the 21 observed variables, for the
six latent variables were all significant at p < .01.
First, it is predicted that individuals with higher levels of anger
and disgust will have higher levels of modern homonegativity. Second,
it is predicted that increased exposure (direct and indirect) to queer
people and images will not affect levels of anger, and disgust. Third, it
is predicted that individuals with higher levels of conservative ideology,
religious beliefs, and certain beliefs about the cause of homosexuality,
will have higher levels of homonegativity. Forth, it is predicted that
increased exposure to queer people and images will significantly affect
ones’ beliefs about the cause of homosexuality, religious beliefs, and
conservative ideology. Fifth, it is predicted that homonegativity (i.e.,
the endogenous variable) is best explained by a structural model that
includes exposure to queer people and images, cognitive
characteristics, and affective characteristics (i.e., exogenous
variables). Finally, the researchers seek to find out if exposure,
cognitive characteristics, and affective characteristics mediate
homonegativity.
According to Willis (2004), “Hate crimes are customarily
defined as criminal acts based on the offender’s bias toward
individuals, families, groups, or organizations because of their
real or perceived racial, ethnic, religious, sexual orientation, or
disability status” (p. 117). He indicated that these
transgressions vary in overtness, intensity, and level of
destructiveness, with acts ranging from subtle intimidation to
rape and murder. The National Coalition of Anti-Violence
Programs (NCAVP; 2008) indicated that in 2007, law
enforcement agencies received 2,430 reports of hate crimes
based on sexual orientation and gender bias. NCAVP findings
from 2007, when compared to reported incidence in 2006, show
a 24% increase in crimes committed against lesbian, gay, and
bisexual (LGB; hereafter termed queer) men and women.
Herek (1989, 1990) suggested hate crimes aimed at queer men
and women resulted, at least in part, from homonegativity;
which, according to Cerny and Polyson (1984), is any
prejudicial affective or behavioral response directed toward an
individual because he or she is perceived to be queer.
Modern Homonegativity
Articulating a comprehensive understanding of
homonegativity is made more complex because, according to
Steffens (2005), overtly anti-queer attitudes and behaviors are
becoming increasingly less socially tolerable and, as a result,
may be expressed in more subtle and covert ways. The
increasingly positive public sentiment for diverse expressions of
sexual orientation may force current manifestations of
homonegativity (i.e., modern homonegativity) to be more
insidious and subdued than what was once considered
acceptable (i.e., old-fashioned homonegativity). Cowan and her
colleagues (2005) described old-fashioned homonegativity as
obvious, or overt, expressions of negative thoughts and
behaviors towards the queer community. Further, they reported
that increased visibility and subsequent acceptance, has only
made public expressions of homonegativity less socially
desirable and more subversive. In other words, old-fashioned
homonegativity is being replaced by more understated and
clandestine attitudes and behaviors that perpetuate oppression
based on sexual minority status while, at the same time,
appearing less corrosive (Walls, 2008).
Cognitive Characteristics:
Religious Beliefs, Conservative Ideology
Affective Characteristics:
Anger, Disgust
Exposure (direct and indirect)
MEASUREMENT MODEL
Results indicated that personal interaction with someone
who is gay or lesbian reduces anger, while indirect exposure
(e.g., via television, movies, the internet) has the opposite
effect, and causes anger levels to increase. In addition, direct
exposure had a positive and statistically significant impact on
cognitive characteristics. Both direct exposure and cognitive
characteristics were inversely related to modern
homonegativity. Specifically, 31% of the variance in cognitive
characteristics was explained by disgust and direct exposure,
31% of the variance in anger was explained by both direct and
indirect exposure, and 67% of the variance in modern
homonegativity was explained by cognitive characteristics and
direct exposure. Much of our findings are in keeping with
previous research and theory. The long-demonstrated
relationships between religious beliefs, conservative ideology,
and homonegativity were supported by this research.
Contrary to Parrot and Peterson’s (2008) findings,
however, our research results indicated direct personal
exposure to queer men and women did not evoke anger, as
was reported. The practical implications of the research
findings are clear. Direct exposure to queer men and women
mediates cognitive characteristics, anger, and homonegativity,
while, indirect exposure may actually increase anger and
homonegativity. Our results suggest that direct exposure,
indirect exposure, cognitive characteristics, and affective
characteristics are meaningful constructs worthy of
consideration when trying to examine the complex and
dynamic psychological mechanisms that promulgate modern
homonegativity.
The proposed structural model offers support for the
notion that statistically significant variance in modern
homonegativity can be explained by direct exposure, indirect
exposure, cognitive characteristics (i.e., religious beliefs,
conservative ideology, beliefs about the cause of
homosexuality), and affective characteristics (i.e., disgust,
anger).
LIMITATIONS & RECOMMENDATIONS
Caution should be used in generalizing our results to other
samples, as the current study is most accurately generalized to White,
well-educated women. In addition, structural equation modeling is a
statistical technique that allows researcher to either confirm or
disconfirm a specified model; however, it is important to note that other
possible models may exist that represent unique, and equally
important relationships among the observed and latent constructs
included in this study.
In spite of the lack of existing measures known to measure the
constructs of interest (e.g., direct and indirect exposure, conservative
ideology), the use of measures that were developed specifically for this
study is potentially problematic, as psychometric stability is
questionable. Consequently, we suggest that future researchers focus
on the development of psychometrically reliable and valid instruments.
ABSTRACT
This study examined a model in which direct (i.e., personal
contact), and indirect (i.e., images viewed inadvertently) exposure to
queer men and women mediated the relationship between, affective
characteristics, cognitive characteristics, and modern homonegativity.
Results from the structural equation modeling of data from 562
heterosexual men and women who participated in an online study,
indicated that personal interaction with someone who is gay or lesbian
reduces anger, while indirect exposure (e.g., via television, movies, the
internet) has the opposite effect, and causes anger levels to increase.
In addition, direct exposure had a positive and statistically significant
impact on cognitive characteristics. Both direct exposure and cognitive
characteristics were inversely related to modern homonegativity.
Specifically, 31% of the variance in cognitive characteristics was
explained by disgust and direct exposure, 31% of the variance in anger
was explained by both direct and indirect exposure, and 67% of the
variance in modern homonegativity was explained by cognitive
characteristics and direct exposure.
This exploratory study is designed to research whether
increased exposure to queer men and women (e.g., having
queer friends or coworkers, exposure to images of queer men
and women on television), cognitive characteristics (e.g.,
conservative ideology, religious beliefs), affective characteristics
(e.g., anger, disgust) affect modern homonegativity. This
research aims to test a theoretically informed model of
homonegativity as, it is anticipated that by doing so, our
collective understanding of this phenomenon will be increased,
which, hopefully, will aid in reducing incidence of hate crimes
and improve quality of life for queer men and women.
CONCLUSIONS

Erwin_Strain_APA_Poster_15

  • 1.
    Testing a MediatedModel of Modern Homonegativity Kyle M. Erwin, M.A. Department of Psychology & Philosophy, Texas Woman’s University, Denton, TX Jeffrey D. Strain, Ph.D. Lake’s Crossing Center, Reno, NV METHODS INTRODUCTION RESULTS PURPOSE & HYPOTHESES HYPOTHESES PARTICIPANTS (n = 526) Ages among participants ranged from 18-72 (m = 33.9) Race/Ethnicity: 383 (72.8%) White 60 (11.41%) Black 36 (6.84%) Latino or Hispanic 17 (3.23%) Biracial or Multiethnic 10 (1.9%) East Asian or Asian American 10 (1.9%) South Asian or Asian Indian 5 (< 1%) American Indian or Alaskan Native 4 (< 1%) Middle Eastern or West Asian 1 (< 1%) Native Hawaiian or Pacific Islander Sex: 73 (13.9%) participants were men, 453 (86.1%) were women. The average annual income among participants was $32,739 66% (n = 346) had earned at least one UG college degree Religious affiliation: 281 (53.42%) Christian 74 (14.06%) indicated that they were spiritual 58 (11.03%) Catholic 44 (8.36%) Agnostics 27 (5.13%) Atheists 13 (2.47%) Buddhists 9 (1.71%) Jewish 7 (1.33%) Unitarian-Universalism 5 (.95%) Pagan/Neo-Pagan, Primal-indigenous, or Taoist 4 (.76%) Hindus 4 (.76%) Muslim MEASURES State-Trait Anger Expression Inventory-2 (STAXI-2; Spielberger) Three Domain Disgust Scale (TDDS; Tybur, 2009) Multidimensional Heterosexism Inventory (MHI; Walls, 2008) PROCEDURES Participants were recruited on social networking sites (e.g., Facebook) and list-serves (e.g., student, faculty). ANALYSES Mean, standard deviations, and zero order correlations for the 21 observed variables Structural Equation Modeling (using LISREL 8.80) STRUCTURAL MODEL Again, we used the two-step method of SEM proposed by Anderson and Gerbing (1988) to test the mediation effects for our hypothesized model (see Figures 1 and 2). The final model demonstrated robust goodness of fit (SRMR = 0.046, RMSEA = 0.039, CFI = 0.98). Moreover, examination of the structural paths indicated that all paths were significant. A CFI greater than .90 demonstrates an acceptable fit. According to Hu and Benter (1999), and other researchers (e.g., Martens, 2005), RMSEA values less than .06, and SRMR values less than .08, indicate a good fit. Despite fit indices indicating a more than acceptable fit, we recognize that absolute cutoff values to accept or reject models fall short of their intended purposes, and fit indices should be used to describe how well a model fits the data (Quintana & Maxwell, 1999). To analyze the data, Anerson and Gerbing’s (1988) two- step method of structural equation modeling was followed (see Table 2). Initially, a confirmatory factor analysis (CFA) was conducted to test whether the measurement model fit the data. After, acceptable fit of the measurement model was established, the structural model was tested. In this study, both the measurement model and structural model were tested using the LISREL 8.80. Both procedures used the maximum likelihood method. Three indexes were used to determine the goodness of fit for the model (Hu & Bentler, 1999): the comparative fit index (CFI), the root-mean square error of approximation (RMSEA), and the standardized root-mean-square residual (SRMSR). It is important to note that our sample size is appropriate for testing our hypothesized model and meets suggested recommendations (Hatcher, 1994). The results of the CFA indicated an acceptable fit to the data Chi-square (df = 159, N = 526) = 288.56, p < .01; scaled Chi-square = 302.10, p < .01, CFI = 0.98; RMSEA = 0.039 (90% confidence interval = 0.032; 0.047); SRMR = 0.046. Besides the fit of the measurement model, the factor loadings of the 21 observed variables, for the six latent variables were all significant at p < .01. First, it is predicted that individuals with higher levels of anger and disgust will have higher levels of modern homonegativity. Second, it is predicted that increased exposure (direct and indirect) to queer people and images will not affect levels of anger, and disgust. Third, it is predicted that individuals with higher levels of conservative ideology, religious beliefs, and certain beliefs about the cause of homosexuality, will have higher levels of homonegativity. Forth, it is predicted that increased exposure to queer people and images will significantly affect ones’ beliefs about the cause of homosexuality, religious beliefs, and conservative ideology. Fifth, it is predicted that homonegativity (i.e., the endogenous variable) is best explained by a structural model that includes exposure to queer people and images, cognitive characteristics, and affective characteristics (i.e., exogenous variables). Finally, the researchers seek to find out if exposure, cognitive characteristics, and affective characteristics mediate homonegativity. According to Willis (2004), “Hate crimes are customarily defined as criminal acts based on the offender’s bias toward individuals, families, groups, or organizations because of their real or perceived racial, ethnic, religious, sexual orientation, or disability status” (p. 117). He indicated that these transgressions vary in overtness, intensity, and level of destructiveness, with acts ranging from subtle intimidation to rape and murder. The National Coalition of Anti-Violence Programs (NCAVP; 2008) indicated that in 2007, law enforcement agencies received 2,430 reports of hate crimes based on sexual orientation and gender bias. NCAVP findings from 2007, when compared to reported incidence in 2006, show a 24% increase in crimes committed against lesbian, gay, and bisexual (LGB; hereafter termed queer) men and women. Herek (1989, 1990) suggested hate crimes aimed at queer men and women resulted, at least in part, from homonegativity; which, according to Cerny and Polyson (1984), is any prejudicial affective or behavioral response directed toward an individual because he or she is perceived to be queer. Modern Homonegativity Articulating a comprehensive understanding of homonegativity is made more complex because, according to Steffens (2005), overtly anti-queer attitudes and behaviors are becoming increasingly less socially tolerable and, as a result, may be expressed in more subtle and covert ways. The increasingly positive public sentiment for diverse expressions of sexual orientation may force current manifestations of homonegativity (i.e., modern homonegativity) to be more insidious and subdued than what was once considered acceptable (i.e., old-fashioned homonegativity). Cowan and her colleagues (2005) described old-fashioned homonegativity as obvious, or overt, expressions of negative thoughts and behaviors towards the queer community. Further, they reported that increased visibility and subsequent acceptance, has only made public expressions of homonegativity less socially desirable and more subversive. In other words, old-fashioned homonegativity is being replaced by more understated and clandestine attitudes and behaviors that perpetuate oppression based on sexual minority status while, at the same time, appearing less corrosive (Walls, 2008). Cognitive Characteristics: Religious Beliefs, Conservative Ideology Affective Characteristics: Anger, Disgust Exposure (direct and indirect) MEASUREMENT MODEL Results indicated that personal interaction with someone who is gay or lesbian reduces anger, while indirect exposure (e.g., via television, movies, the internet) has the opposite effect, and causes anger levels to increase. In addition, direct exposure had a positive and statistically significant impact on cognitive characteristics. Both direct exposure and cognitive characteristics were inversely related to modern homonegativity. Specifically, 31% of the variance in cognitive characteristics was explained by disgust and direct exposure, 31% of the variance in anger was explained by both direct and indirect exposure, and 67% of the variance in modern homonegativity was explained by cognitive characteristics and direct exposure. Much of our findings are in keeping with previous research and theory. The long-demonstrated relationships between religious beliefs, conservative ideology, and homonegativity were supported by this research. Contrary to Parrot and Peterson’s (2008) findings, however, our research results indicated direct personal exposure to queer men and women did not evoke anger, as was reported. The practical implications of the research findings are clear. Direct exposure to queer men and women mediates cognitive characteristics, anger, and homonegativity, while, indirect exposure may actually increase anger and homonegativity. Our results suggest that direct exposure, indirect exposure, cognitive characteristics, and affective characteristics are meaningful constructs worthy of consideration when trying to examine the complex and dynamic psychological mechanisms that promulgate modern homonegativity. The proposed structural model offers support for the notion that statistically significant variance in modern homonegativity can be explained by direct exposure, indirect exposure, cognitive characteristics (i.e., religious beliefs, conservative ideology, beliefs about the cause of homosexuality), and affective characteristics (i.e., disgust, anger). LIMITATIONS & RECOMMENDATIONS Caution should be used in generalizing our results to other samples, as the current study is most accurately generalized to White, well-educated women. In addition, structural equation modeling is a statistical technique that allows researcher to either confirm or disconfirm a specified model; however, it is important to note that other possible models may exist that represent unique, and equally important relationships among the observed and latent constructs included in this study. In spite of the lack of existing measures known to measure the constructs of interest (e.g., direct and indirect exposure, conservative ideology), the use of measures that were developed specifically for this study is potentially problematic, as psychometric stability is questionable. Consequently, we suggest that future researchers focus on the development of psychometrically reliable and valid instruments. ABSTRACT This study examined a model in which direct (i.e., personal contact), and indirect (i.e., images viewed inadvertently) exposure to queer men and women mediated the relationship between, affective characteristics, cognitive characteristics, and modern homonegativity. Results from the structural equation modeling of data from 562 heterosexual men and women who participated in an online study, indicated that personal interaction with someone who is gay or lesbian reduces anger, while indirect exposure (e.g., via television, movies, the internet) has the opposite effect, and causes anger levels to increase. In addition, direct exposure had a positive and statistically significant impact on cognitive characteristics. Both direct exposure and cognitive characteristics were inversely related to modern homonegativity. Specifically, 31% of the variance in cognitive characteristics was explained by disgust and direct exposure, 31% of the variance in anger was explained by both direct and indirect exposure, and 67% of the variance in modern homonegativity was explained by cognitive characteristics and direct exposure. This exploratory study is designed to research whether increased exposure to queer men and women (e.g., having queer friends or coworkers, exposure to images of queer men and women on television), cognitive characteristics (e.g., conservative ideology, religious beliefs), affective characteristics (e.g., anger, disgust) affect modern homonegativity. This research aims to test a theoretically informed model of homonegativity as, it is anticipated that by doing so, our collective understanding of this phenomenon will be increased, which, hopefully, will aid in reducing incidence of hate crimes and improve quality of life for queer men and women. CONCLUSIONS