There are many studies examining the issue of homosexuality from many different disciplines like mental health, psychology, clinical perspective, sociologists, law, communication, history, religion, market research, or multidisciplinary approaches etc. In particular, research on negative attitudes toward homosexuals has rapidly increased in the past decades. Researchers have been seeking insight into factors that differentiate between individuals who have a generally positive attitude toward homosexuals and those who have a generally negative toward homosexuals. One of common interest is gender differences and their homophobic attitudes.
Homophobia is a personality trait with deleterious effects, also labeled as homosexphobia, heterosexism, homoerotophobia. It is generally refer to the personal and institutional prejudice and fear or negative attitudes toward lesbian and gay men (Herek, 1988). Although not proven empirically, people who hold negative attitudes toward homosexuals also tend to disagree same-sex marriage whereas people who hold positive attitudes toward homosexuals tend to support same sex marriage.
Canada has recently become the third country in the world that legalized same-sex marriage after Netherland and Belgium. However, there are still overwhelmingly negative attitudes toward homosexuals in the society. Thus, the present paper investigated the basis for differences among Canadians that explain their negative attitudes towards homosexual that also lead them to oppose same-sex marriage.
II. Literature Review
One of the most important scholar that study gender differences towards homosexuality is Herek, Gregory. M. (1988, 1994,2002). First, he found out that male and female heterosexuals differ in their attitudes, women generally hold more favorable and less condemning attitudes toward gay people. Second, aggregate attitudes tend to be more hostile toward gay men than lesbians. Affective reactions towards gay men were significantly more negative than reactions to lesbians. Third, whereas heterosexuals tend to express more negative attitudes toward homosexual people of their same sex, this pattern occurs mainly among men. Heterosexual men responded significantly more negatively towards gay men than to lesbians in questions about recognition of same-sex relationships and adoption rights.
Herek (1988) also summarized heterosexuals with negative attitudes are more likely to express traditional, restrictive attitudes about gender roles, more likely to manifest high levels of authoritarian personality characteristic and more likely to subscribe to a conservative religion ideology.
Johnson et al (1998) also reported gender was related to homophobia in his study with women being less homophobic than men and more willing to grant human tights to gay individuals. Age was also related to homophobia with older respondents being less homophobic than younger respondents. The study also shows that other personality traits such as empathy, religiosity and coping styles are related to negative attitudes towards homosexuals. Higher levels of empathy are less homophobia toward gay individuals regardless of perceived origins of their homosexuality. Religious beliefs and religion-consistent behaviors were related to higher level of homophobia. Relative to coping style, individuals who tend to use denial and to isolate or turn away from others were more homophobic.
Lim (2002) studied gender differences in attitudes toward homosexuals in a non-Western and Confucian ideology based society – Singapore also shows that gender differences in attitudes towards homosexuals exist. The results of this study suggest women are more tolerance towards homosexuality in general than men.
In sum, one of the most consistent findings is that male manifest more homophobic hostility on average than do female. This includes being more supportive of employment protection, human rights, adoption rights and were more willing to extend some form of recognition to same-sex couples.
In light of the previous studies, the present study is interested to explain the gender differences in supporting the issue of same sex marriage. The main objective of this study is to understand and try to explain reasons of the recognition or opposition of same-sex marriage. It is hypothesized there is a gender differences in supporting same sex marriage. In other word, female or male are more likely to agree that ‘gay and lesbian should be allowed to get married’. The null hypothesis is that there is no relationship between gender (female or male) and their supporting (or opposition) to same sex marriage. Besides, the present paper also explore some other variables that associated with supporting of same sex marriage.
The data used in the present study is based on the 2000 Canadian National Election Survey (CNES). The 2000 CNES included three components – Campaign-Period Survey (CPS), Post-Election Survey (PES) and Mailback Survey (MBS). Of those respondents who completed the CPS, almost 80% also completed PES, but only 42% completed the MBS.
The research is concerned with gender differences and attitudes toward homosexuals. More particularly, the study explore the gender differences in recognition and/ or supporting of same sex marriage. The measurement of attitudes toward homosexuality, or the recognition of same-sex marriage is base on the following question ‘Gays and lesbian should be allowed to get married’ from the dataset. Cpsf18 is an ordinal variable with rank ordered answers from strongly agree, somewhat agree, somewhat disagree to strongly disagree.1 Those who strongly disagree represent opposition of same-sex marriage (hold negative attitudes toward homosexuals) whereas those who strongly agree represent recognition of same-sex marriage (hold positive attitudes toward homosexuals).
There are two main parts of this study. First, the bivariate relationship of gender differences in supporting of same sex marriage is explored by using cross tabulation and chi-square test. According to previous literature, other personal traits could explain attitudes toward homosexuality. Therefore, the study explores other possible predictors of supporting of same-sex marriage in the second part of the analysis by using multiple regression.
In the multiple regression, many others variables that associated with religiosity, individual backgrounds and personality traits are included. The purpose of the multiple regression is to identify the most important explanatory variables and examine how much of the variation is cause by the independent variables. The main goal is to increase predictive accuracy by identifying a set of independent variables that allowed for the best prediction about the dependent variable. Meanwhile, ‘gender’ is also included in the multiple regression for hypothesis testing.
IV. Data Analysis
The respondents consist of 1762 male and 1889 female. 20.7 % of the respondents strongly agree that ‘gays and lesbian should be allowed to get married’ while 28.7 % somewhat agree; 11.1% somewhat disagree and 30.7 % strongly disagree (see appendix 1).
First, a cross tabulations of bivariate tables is constructed to examine the relationship between gender and supporting of same-sex marriage. The variable ‘sex’ is introduced as the independent variable, it is placed on top of the table and is calculated in columns. The recoded ‘cpsf18′(gays and lesbians should be allowed to get married) is the dependent variable which is place at the left side of the table and calculated as rows. The column percentage is run to compare the percentage between female and male. Additionally, the chi-square measure is run to examine the association and level of significance of the results.
Table I. The cross tabulation of ‘gender’ and ‘cpsf18 recode’
There are altogether 91.2 % (3331) valid cases with 1612 male respondents and 1719 female respondents and 8.8 % (320) missing cases. From the contingency table, there are 19.2% of male strongly agree that ‘gays and lesbians should be allowed to get married’, 28.9% somewhat agree, 12.7 % somewhat disagree and 39.3% strongly disagree. On the other hand, there are 26.1% of female strongly agree with same-sex marriage, 33.9% somewhat agree, 11.6 % somewhat disagree and 28.4% somewhat disagree.
By comparison, female are more likely to agree with same sex marriage and male are more likely to disagree with same sex marriage. There is 5.9% more female strongly agree and 5% more female somewhat agree with same sex marriage than male. While looking at the opposition side, 10.9% more male strongly disagree, and 1.1% more male somewhat disagree than female with the statement ‘gays and lesbians should be allowed to married’.
The contingency table is a nominal by ordinal table. Thus, the appropriate measure for nominal-level measures are chi-square, lambda and Cramer’s V.
Chi-square testes are often used nonparametric tests for nominal and ordinal-level relationship, it can be used to determine the acceptance or the rejection of the null hypothesis. The chi-square test indicates that the relationship has been found to be statistically significant (0.000) and thereby reject the null hypothesis and conclude that the two variables are related.
Chi-square tests do not tell the strength of the relationship. Therefore, it is commonly refer to Lambda or Cramer’s V for nominal-level measures. Lambda is a proportional reduction in error (PRE) measure and uses the mode to make predictions. Cramer’s V is based on the test statistic of chi-square so it is preferably refer to Cramer’s V in this study. Hence, the table indicates that female are more likely to agree with same sex marriage with a weak relationship of v = 0.127 and this is a statistically significant relationship. Note that since Cramer’s V is not a PRE measure, it is impossible to make statements about the improvements in its predictive accuracy,
As a result, the study shows that female are more likely to support same-sex marriage with a weak and significant relationship.
In the second part of the analysis, a multivariate linear regression is run to determine some others possible predictors that could explain individual differences in supporting of same-sex marriage. Many trials are conducted in order to avoid multiculinearity of variables and have a maximum amount of variation explained by the set of predictors.
There are altogether eight independent variables entered to be the predictors of the dependent variable ‘gay and lesbians should be allowed to get married’ (labeled as ‘same-sex marriage’).2 The first set of the variables includes gender, religiosity, marital status and occupation. The second set of variables includes four personality traits – conservative, tolerance, traditionalism and acceptance.
The measurement of religiosity is based on the following question “cpsm 10 b. In you life, would you say religion is very important, somewhat important, not very important or not important at all”.3 Marital status is base on “cpsm 2. Are you presently married, living with a partner, divorced, separated, widowed or have you ever been married?”4 Occupation is based on “Are you presently self-employed, working for pay, are you employed, retired, a student or a homemaker?”5
The measurements of the four personality traits are based on the following questions and all the answers ranked from strongly agree, agree, disagree to strongly disagree. Conservative – “mbsa1. We have gone too far in pushing equal rights in this country”. Tolerance – “mbsa2. We should be more tolerant of people who choose to live according to their own standards, even if they are very different from our own”. Traditionalism – “mbsa7. Newer lifestyles are contributing to the breakdown of society”. Acceptance – “mbsa8. The world is always changing and we should adapt our view of moral behaviors to these changes”.6 (see Appendix II for the frequency table and Appendix III for the frequency tables of the recoded variables).
First, all the independent variables are recoded into precise form for the purpose of this study with the insufficient or unavailable data system missing. The nominal variables are grouped into two categories (0, 1) whereas the scale of all the ordinal variables are recoded into the same direction – from negative to positive (strongly disagree to strongly agree).
Second, the correlation matrix of all the variables is examined to avoid multicollinearity among variables. In other word, there could be variables that are measuring the same dimension of the study. Multicollinearity occurs when the correlation (r) between two variables equal or higher than .8. It is important in multiple regression to ensure that all independent variables are independent of each to get a relatively robust result. In this case, all the independent variables are examined carefully making sure that all the independent variables are not correlated. (See appendix IV).
The following tables are the result of multivariate linear regression with the abovementioned nine variables as independent variable and ‘same-sex marriage’ as dependent variable.
This table indicates the correlation coefficient of the multiple regression. The strength of the relationship is R=0.530 which is a strong relationship. The R square is the ‘proportion variance accounted for’, it indicates the amount of variation in the dependent variable explained by the abovementioned nine independent variables. In this study, as in many other studies, adjusted R square is usually preferred because it removes the differences caused by the change in unit of measurements. In this case, 27.6 % of the variance in the dependent variable is explained by the abovementioned none variables.
This table provides important information about the significance level of the relationship between the independent variables and the dependent variable. to be exact, the significance level is .000, it is less than 1% to get an error, and thus it is very significant with 99% confidence level. In fact, Anova (analysis of variance) is one of the most useful statistical techniques in the social sciences. Analysis of variance provides a systematic method for comparing the relative strength of between group and within group differences. However, this is not the main concern in this study.
Table II. Table of Coefficients of multiple regression
The above table demonstrates B (the slope) of each independent variable and their standard error. The slope of the regression line for each independent variable represents the amount of change in the dependent variable for each unit change in the independent variable. The standard error is similar to the standard deviation of mean, smaller standard error means smaller the error in predicting the dependent variable on the basis of the independent variables. Thus, the standard error cannot be larger than the B.
While looking at the gender, the slope of regression line (supporting of same sex marriage ranked from strongly disagree, somewhat disagree, somewhat agree to strongly agree) will increase by .412 for every one unit change in gender (from 0 male to 1 female). For occupation, the supporting of same sex marriage will increase by .260 for every one unit change from ‘not working’ to ‘working’. For religiosity, the supporting of same sex marriage will increase by .181 for every one unit change from ‘not important at all’, ‘not very important’, ‘somewhat important’ to ‘very important’. Consequently, every one unit change in ‘acceptance’ and ‘tolerance’ (from strongly disagree, disagree, agree to strongly agree) is associated with an increase of .186 and .135 in supporting of same-sex marriage respectively. Thus, female correspondents who are working, think that religion is important with acceptance and tolerance personality traits are more likely to support same-sex marriage.
On the contrary, the slope of the regression line (supporting of same-sex marriage) will decrease by .288 and .232 for every unit change in ‘conservative’ and traditionalism’ (from strongly disagree, disagree, agree to strongly agree). Thus, correspondents with conservative and traditionalism personality traits are more likely to oppose same-sex marriage. All these results are significant at the 95 % confidence level. However, ‘marital status’ is not a significant variable and thereby excluded from the regression analysis and equation. Thus, being either single or married will not change a respondents’ perception towards same-sex marriage.
In addition, the table also shows the standardized coefficients of each variable. In a multiple regression, all the variables are measured on different scales. Thus, the standard coefficients Beta are used to compare the relative impact of each independent variable on the dependent variable. From the table, ‘conservative’ is the strongest predictor (being conservative are more likely to oppose same sex marriage with Beta = -.213), followed by gender (female are more likely to support same sex marriage with Beta = .174 and traditionalism (being traditional are more likely to oppose same sex marriage with Beta= -.172).
In sum, the results from the multiple regression secure the hypothesis that female are more likely to support same sex marriage. Other variables that associated with associated with supporting of same sex marriage include ‘religiosity’, ‘acceptance’, ‘tolerance’ and ‘occupation’ whereas ‘conservative’ and ‘traditionalism’ are associated with opposition of same sex marriage.