Breaking Down Homophobic Attitudes:
The effect of gender and level of contact on homophobic
Attitudes in college students
St. Olaf College
Sociology/Anthropology Department
Fall 2008
Anne Steeves, Martha Stuckey, Selina Betcher, Jon Mahnke
Abstract
This study measured levels of homophobia among heterosexual college students at a small, liberal arts college in the Midwest. Our survey examined how the number of homosexual contacts respondents’ have affects their attitudes, beliefs, and levels of comfort towards lesbian, gay, bisexual, transsexual, and queer (LGBTQ) individuals and the community. We found statistically significant evidence that students with close LGBTQ contacts (close friends and family) have more positive attitudes and beliefs and a higher level of comfort towards the LGBTQ community, on average, than students who have no close LGBTQ contacts. We also found no evidence that levels of homophobia differ between male and female respondents at St. Olaf, which diverges from previous research. Overall, we found that St. Olaf students generally have positive attitudes, beliefs, and are comfortable with LGBTQ individuals and the community.
Introduction
November 4th, 2008 was a momentous day for the entire nation and a celebratory time for many. But for the LGBTQ community and allies, it was a bleak and disheartening reminder of pervasive homophobia that continues to suppress efforts to achieve equality. Proposition 8 was a California ballot proposition that changed California's constitution to limit the definition of marriage to the coupling of a man and woman, eliminating the ability of same-sex couples to marry. The proposition passed by a small margin, yet it is a powerful statement for a dominantly liberal population to pass such legislation. Our culture has become more accepting since the feminist and gay pride movements in the 1970's but it is clear that negative attitudes still exist. For this reason, our research is especially imperative to discern and improve attitudes towards the LGBTQ community.
Review of Literature
The term homophobia was coined in 1972 by psychologist George Weinberg who examined how American society thought about sexuality by looking at “antigay hostility” (Herek 2004). At the time homophobia was defined as “the fear of being in close quarters with homosexuals.” However, in the past three decades the definition and study have evolved both socially and scientifically. For instance, Agnew et al.’s 1993 study concluded that factors associated with homophobia include a lack of contact, religiosity, and the importance of traditional sex/gender roles.
Past research reports those with higher levels of contact with homosexuals have more positive attitudes regarding homosexuality (Agnew et al. 1993; Finlay &Walther 2003; Gentry 1987). Herek’s 1993 and 2002 studies found a positive correlation between comparing levels of contact and acceptability, but could not prove causality between them.
In general, females have more contact and therefore more positive attitudes regarding LGBTQ issues (Herek & Glunt, 1993). Basow (2000), Marsiglio (1993), and Herek (2002) found a strict adherence to traditional gender roles fostered more negative attitudes towards homosexuals. Belief in normative gender orientation influences males more than females by creating a fear around preserving and protecting one’s masculinity because of higher levels of discomfort around LGBTQ individuals (Basow 2000; Herek 2002; Herek & Glunt, 1993; Marsiglio 1993).
After reviewing and conceptualizing this information we came up with two hypotheses: 1) that heterosexuals with close LGBTQ contacts are less homophobic than those with no close LGBTQ contacts, 2) that females have lower levels of homophobia than males.
Research Methods
We define homophobia as adherence to negative stereotypes, prejudiced beliefs about the LGBTQ community, and a low level of comfort within a medium level of social distance with LGBTQ individuals. To reach a large and fully representative survey sample, data were collected through survey questionnaire distributed through Form Creator, an online program. The central concepts investigated how an individual’s gender and level of contact with LGBTQ individuals (independent variables) affect her/his level of homophobia (dependent variable).
Respondents recorded the number of known LGBTQ individuals they had been in contact with over the past year. We provided social categories for respondents to choose from: roommate, casual acquaintance, teacher/professor, close friend, intimate dating relationship, family member, co-worker, and other (to be specified by the respondent). We later collapsed the categories into two groups, creating those with “close contact” which included close friend and/or family member; and “no close contact” which included all other responses. Because we measured heterosexuals’ levels of homophobia we discarded answers in the “intimate dating relationship” category. Respondents reported their gender, sexual orientation and class year.
To measure levels of homophobia, we provided two sets of five statements that studied respondents’ attitudes, beliefs and level of comfort. We later combined responses from both sets to create a Homophobic Scale score. The Homophobic Scale score is what was used to compare attitudes amongst the different levels of contact and genders. The first set of statements, based on a study by Gentry (1986: 199-208), investigated attitudes and beliefs through statements like “LGBTQ individuals are mentally ill” and “Same sex couples have the right to marry.” Participants responded using a Likert scale in order to measure levels of stereotypical beliefs and opinions concerning the civil rights of the LGBTQ community.
The second set of statements asked questions about level of comfort at specific social distances, as comfort was part of our conceptual definition of homophobia. Those with a low level of comfort were considered more homophobic. Again, we used a Likert scale to measure agreement with statements including “I would feel uncomfortable attending a party where known LGBTQ individuals would be in attendance” and “I would feel comfortable staying overnight at a known LGBTQ individual’s house.” We modeled our scale items on the social distance scales of prejudice developed by Bogardus in 1925 (Neuman 2007: 132-133).
By consulting Dr. Ryan Sheppard, the professor of our sociological research methods class, we were able to achieve face validity, or a type of measurement validity in which an indicator ‘makes sense’ (Neuman 2007: 366). Our research methods class critiqued and observed our conceptual and operational definitions as well, to ensure that our indicators accurately measure our constructs. We also administered pilot tests to our classmates to ensure clarity. These helped us create precise measurements to test our hypotheses accurately. To ensure content validity we had represented all aspects of our conceptual definition, we checked all parts of our construct of homophobia, with our indicators (Neuman 2007: 363). For instance, we gauged a variety of activities and behaviors to measure levels of homophobia. We used Likert scales, a modified version of Gentry’s homosexuality indicators (1986), and a Bogardus’ social distance scale (1925) since they are widely accepted methods of measurement and achieved concurrent validity, a measurement of validity relying on preexisting and already accepted measures to verify the indicator of a construct (Neuman 2007: 363). We achieved criterion validity which relies on some independent, outside verification, for our hypothesis that which stated higher level of contact will foster more positive attitudes and comfort with LGBTQ individuals (Neuman 2007: 363).
Reliability means consistency in results in comparable circumstances; according to Neuman, “It suggests that the same thing is repeated or recurs under the identical or very similar conditions” (Neuman 2007: 115). Neuman also suggests four ways to improve reliability: Clearly conceptualize definitions and concepts, use known precise levels of measurement, pick indicators that are varied and informative, and administer pilot tests to ensure consistency and accuracy (Neuman 2007, 116). Firstly, we clearly conceptualized our concepts through discussion and much consideration. After pilot-tests and editing we further conceptualized to make sure our definitions were unambiguous and clear. Secondly, we used a Likert scale and the Bogardus social distance scale. Thirdly, we consciously picked indicators that were varied and informative using statements that involved a range of LGBTQ issues like civil rights and stereotypes, and real life social situations with varying social distances with LGBTQ individuals. Lastly, we analyzed the pilot-tests for consistency in responses and accuracy of wording.
To test our hypotheses and best represent our target population, we used a simple random sample survey of heterosexual students at a Midwestern, Liberal Arts College. Before choosing our sample we excluded students that participated in a focus group to help us formulate our questions prior to formal administration to ensure unbiased responses. We also excluded students studying abroad and special populations, or a group of people that requires another authority’s permission for them to legally respond, including anyone under the age of 18. After distributing 540 surveys, we received 315 completed giving us a 58.3% response rate. We then excluded non-heterosexual responses since we were trying to measure heterosexual levels of homophobia. Overall our respondents were 38% male and 61% female, and 21% seniors, 23% juniors, 26% sophomores, and 24% freshmen.
Our main ethical concern was to protect the privacy and well-being of respondents, maintain anonymity, avoid coercion, ensure voluntary and informed consent, and protect special populations. Because we asked for private beliefs, backgrounds and behaviors on sensitive issues, privacy and anonymity were important ethical issues. We hoped that respondents would take the survey in a safe and unbiased environment to protect their privacy. To protect anonymity, the computer program used to create our survey did not attach a name to the completed responses. The nature of a survey lends itself to the avoidance of any physical or psychological harm, and we avoided questions that were unnecessarily stress-inducing. Legal harm was not a concern as our subject matter does not deal with illegal practices.
On the cover letter that preceded our survey we described the purposes of the study, the benefits to respondents if they completed the study, the study’s dissemination, and the voluntary nature of their participation. With this information respondents could choose to complete or not complete the survey, thus ensuring informed, voluntary consent. Finally, as mentioned previously, we also asked any respondent under the age of 18 to refrain from completing the survey. In order to fulfill our legal and ethical obligations to project participants we applied for and received approval from the Institutional Review Board (IRB).
Results
Close Contacts
In order to test our first hypothesis, we asked respondents to quantify the number of LGBTQ contacts into various categories which we then collapsed into close and non-close contacts. The mean number of close contacts was 2.22 with a standard deviation of 3.181, a minimum of 0 and maximum of 23 creating a right skew (Figure 1). Based on these results, our respondents, on average, tend to have about two close LGBTQ contacts.
The distribution for non-close contacts was also skewed to the right. The mean number of non-close contacts was 7.94 with a standard deviation of 7.995 and minimum of 0 and a maximum of 60 (Figure 2). Based on these results, our respondents, on average, tend to have over 7 non-close LGBTQ contacts.
Figure 1: Distribution of total close LGBTQ contacts per respondent.
Figure 2: Distribution of total non-close LGBTQ contacts per respondent.
|
|
Mean Total Scale Scores
In order to quantify total homophobic attitudes we created a Homophobic Scale. The least homophobic or most accepting responses, received a maximum score of 40, while the most homophobic or least accepting responses, received a minimum score of 10. The mean of the Homophobic Scale scores did not have a normal distribution and were skewed left, indicating low levels of homophobia among respondents (See Figure 3). The mean homophobic scale score was 32.00, with a standard deviation of 6.414 and minimum of 12 and maximum of 40.
Figure 3: This histogram shows the distribution of the homophobic scale scores. 40 is the least homophobic score.
Contact with LGBTQ Individuals
Our first hypothesis stated that respondents with close contacts (family and close friends) would be less homophobic than those with no close contacts. In an independent-sample t test we found significance (t(276) = -5.59, p<0.01). This confirmed that respondents with close LGBTQ contacts (M= 33.40, SD= 5.23), on average, have lower homophobic scale scores than those without close contacts (M= 28.30, SD= 7.43). The 95% confidence interval for the difference in mean scores was -6.91 to -3.30. Therefore, we reject the null hypothesis that there is no difference in homophobic scale scores for those with close contacts and those with no close contacts. Instead, our results support the alternative hypothesis that respondents with close contacts are less homophobic. Figure 4 shows the distribution of Homophobic Scale scores of respondents with close contacts and those with no close contacts.
A positive correlation was found in a Spearman’s rho test (rho(276)= 0.373, p< 0.01) indicating a significant relationship between the two variables of total LGBTQ contacts and total Homophobic Scale scores. Therefore, those with more LGBTQ contacts tend to have higher Homophobic Scale scores, or low levels of homophobia. We conducted another Spearman’s rho test between total close LGBTQ contacts and homophobic scale scores again finding a positive correlation (rho(276)= 0.377, p<0.01). This indicates a significant relationship between total close LGBTQ contacts and total Homophobic Scale scores showing those with more close contacts tend to have higher scale scores indicating lower levels of homophobia. Therefore, we reject the null hypothesis that contact has no influence on levels of homophobia.
Figure 4: Distribution of total homophobic scale scores according to contact.
Gender differences in the Homophobic Scale
Based on past studies, our second hypothesis stated that women are less homophobic than men. We used independent-samples t tests to evaluate this hypothesis based on Homophobic Scale scores. The tests were all non-significant at the p=0.05 level, meaning that there was no significant difference between the male scores and females scores with regards to Homophobic Scale scores, attitudes and beliefs scale scores, and social distance scale scores. Based on these results, we cannot reject the null hypothesis that men and women have equal levels of homophobia. Figure 6 shows the mean total scores based on gender.
Figure 6:
Mean Homophobic Scale scores by gender.
Discussion
Our data support the hypothesis that those with close LGBTQ contact are less homophobic than those with no close contact. After collapsing our respondents into those with "close contact" and "no close contact" we found those in the former group had significantly higher Homophobic Scale scores, signifying they are less homophobic than those in the latter group. This finding reflected past research. Herek & Glunt (1993), Agnew (1993), Basow (2000), Finlay & Walther (2003), and Gentry (1987) also found level of contact was an indicator of homophobia. Presumably, the cause of this is that time spent with LGBTQ individuals’ dissolves harmful stereotypes and increases empathy and understanding between the individuals in the relationship.
Our data did not support the hypothesis that females are less homophobic than males. Although we found statistical significance that showed females had more LGBTQ contact than males, the difference between female and male Homophobic Scale scores was not significant. On the whole, the Homophobic Scale mean score was fairly high denoting mostly positive attitudes and beliefs, and a high level of comfort. This finding was unanticipated given past literature. Many studies found that males are generally more homophobic than females as inferred by self-reported higher levels of discomfort around LGBTQ individuals (Basow 2000; Herek & Glunt 1993; Lachlan & Lannutti 2007; Marsiglio 1993; Gentry 1987). We infer that males adhere to their masculine gender norms more than women, making them less comfortable when masculinity is put into question. From this information we inferred females would have more positive attitudes regarding LGBTQ individuals and community.
Although this discovery was unanticipated it is easily explained when considering the left-leaning tendency of students at St. Olaf College. Because of this perspective, it is plausible that nearly all students will have open attitudes towards the LGBTQ community given the supportive stance Democrats and liberals take in debates surrounding civil rights and opportunities.
Conclusion
Our research goal was to study how gender and level of contact with LGBTQ individuals affect levels of homophobia among heterosexual college students at a small, liberal arts college in the Midwest. Our first hypothesis was supported by statistically significant results, but our second hypothesis was not. As stated in previous literature and as we hypothesized, a greater amount of contact is correlated with higher levels of acceptances of LGBTQ individuals and the community. The data did not support our second hypothesis that females are less homophobic than males. We found that males and females had similar low levels of homophobia. This may be due to St. Olaf’s liberal-leaning partiality.
This study adds to the scientific body of knowledge by supporting previous studies that suggested that higher levels of contact between heterosexuals and LGBTQ individuals lead to more positive attitudes and beliefs. By understanding this, we can encourage more contact between heterosexuals and LGBTQ individuals. This contact, in theory, will help dissolve stereotypes that lead to negative attitudes and beliefs and provide a less discriminatory perspective towards the LGBTQ community.
A strength of our study was that we had a very high response rate of 58.3% which would suggest that our sample is fairly representative of the St. Olaf population. We attribute the higher response rate to the use of an online survey, which was convenient for students. We were also able to create the Homophobic Scale which allowed us to compare different levels of contact and responses among the genders. The scale created from the two sets of five questions was key in comprehending our results.
One weakness of our study is that we cannot establish causality between contact and homophobia. As Herek’s 1993 and 2002 studies pointed out, negative attitudes may come before any relationships are formed, which would deter the formation of relationships with LGBTQ individuals, reinforcing negative attitudes. Also, if positive attitudes come first, this would facilitate more contact with LGBTQ individuals. A final weakness of our study is that we are not able to generalize to the larger population of colleges beyond St. Olaf because of strong political views and a lack of diversity.
Since the social movements of the 1970’s, discrimination has dissipated, yet new issues and new kinds of homophobia have been created. In our study, we examined attitudes toward a community whose only similarity is their lack of adherence to traditional gender norms. It is entirely possible that breaking masculine gender norms may produce different kinds of attitudes than breaking feminine gender norms. It would be helpful to the LGBTQ community to study how homophobia differs between genders to best reduce specific forms of discrimination. We would also be interested in looking at the influence St. Olaf has in changing any attitudes and beliefs towards homosexuals.
Ultimately, we hope that our study could facilitate effortsto reduce homophobia in any community. This could be facilitated by studying what specific kinds of contact with LGBTQ individuals’ increases understanding and empathy while dissolving discrimination and harmful stereotypes.
Appendix: Additional Data Analysis for Breaking Down Homophobic Attitudes: The effect of gender and level of contact on homophobic attitudes in college students
Elizabeth Zahn, Consulting Statistics Fellow
In addition to investigating whether or not gender and level of contact with LGBTQ individuals affected the levels of homophobia among heterosexual college students, we wished to investigate several other hypotheses. First, we wondered whether or not religiosity was associated with lower scores on the Homophobic Scale. In a single Likert-item question, respondents were asked to strongly agree, somewhat agree, somewhat disagree, or strongly disagree with the statement “religion is very important in my life.” Comparing mean Homophobic Scale scores across responses in a one-way ANOVA did not yield significant results, however (p-value = 0.397).
We also wondered whether other demographic variables, such as ethnicity or class year, would be significantly associated with scores on the Homophobic Scale. Because of the homogeneity of the St. Olaf College population, ethnicity was merged into a binary variable comparing White non-Hispanic students and non-White students. Using an independent samples t-test comparing the two groups mean Homophobic Scale scores, we found that ethnicity was not significantly related to the level of homophobia within an individual (t= -0.592, p-value = 0.874). An ANOVA comparison of means indicated that class year was also not significantly associated with level of homophobia (F= 0.180, p-value = 0.857).
A considerable part of our analysis investigated the amount and type of contact a respondent had with LGBTQ individuals. We found that the level of homophobia was lower for students who had close contact with LGBTQ individuals as opposed to those that did not (t=5.85, p-value = 0.000). Building on this result, we wondered if the level of homophobia would differ between genders when close contact was held constant using multiple linear regression. We found that, after adjusting for differences in close contact, gender was not significantly related to level of homophobia (t= -0.72, p-value = .468). However, holding gender constant, the indicator variable for whether or not a student had close contact with an LGBTQ individual remained significantly associated with the Homophobic Scale score (t=5.85, p-value = 0.000).
We further investigated which types of contact relationships with LGBTQ individuals were most highly associated with levels of homophobia within college students. Independent sample t-tests were used to test whether any amount of interaction with each type of contact had an effect on the Homophobic Scale score. The mean scale scores of individuals with a close friend was significantly lower than individuals without a close friend (p-value = 0.000). This relationship seems to have the greatest effect on the level of homophobia within college students. Having a casual acquaintance or a teacher in the LGBTQ community were also both significant at the .05 level. There was also marginally significant evidence (p-value<0.1) that individuals with roommates, family members, or co-workers who are LGBTQ, were more likely to have lower mean Homophobic Scale scores.
Sources Cited
Agnew, C. R., et. Al. 1993. “Proximal and Distal Predictors of Homophobia: Framing the Multivariate Roots of Outgroup Rejection.” Journal of Applied Social Psychology 23(24): 2013-2042.
Basow, Susan A., 2000. “Predictors of Homophobia in Female College Students.” Sex Roles: A Journal of Research Vol. 42(5-6): 391-404.
Finlay, Barbara and Carol S. Walther. 2003. "The Relation of Religious Affiliation, Service Attendance, and Other Factors to Homophobic Attitudes among University Students." Review of Religious Research 44: 370-393.
Gentry, Cynthia S. 2007. “Social Distance Regarding Male and Female Homosexuals.” Journal of Social Psychology 127(2): 199-208.
Lachlan, Kenneth A. & Lannutti, Pamela J. 2007. “Assessing Attitude Toward Same-Sex marriage: Scale Development and Validation.” Journal of Homosexuality 53(4): 113-133.
Herek, Gregory M. 2004. “Beyond ‘Homophobia’: Thinking About Sexual Prejudice and Stigma in the Twenty-First Century.” Journal of National Sexuality Resource Center 1(2): 6-24.
Herek, Gregory M. and Eric K. Glunt 1993. “Interpersonal Contact and Heterosexuals Attitudes toward Gay Men: Results from a National Survey.” The Social Sex Research 30(3): 239-244.
Marsiglio, William 1993. “Attitudes toward Homosexual Activity and Gays as Friends: A National Survey of Heterosexual 15- to 19-Year-Old Males” The Journal of Sex Research 30(1): 12-17.
Weinberg, George 1972. Society and the Healthy Homosexual. New York, NY: St Martins Press.
