Predicting support for marriage equality in Australia

Abstract

Objective

The marriage equality debate is becoming increasingly prominent in Australian political and media discourse. Moreover, as policy that would legislate marriage equality continues to be debated in political circles, public opinion on the topic appears to be becoming increasingly divided. This article presents a cross‐sectional study exploring predictors of support for marriage equality.

Method

A sample of 137 Australians (66% females) responded to a series of demographic items, a measure of attitudes towards gay men and lesbian women, and a forced‐choice question asking whether or not they would support marriage between two people—regardless of their gender.

Results

The results revealed that support for marriage equality in this sample was predicted by religious affiliation, political orientation, and sexual prejudice. More specifically, individuals in support of marriage equality were more likely to be non‐religious, politically liberal, and have more positive attitudes towards gay men (attitudes towards lesbian women were unrelated to support for marriage equality). Age and gender did not predict support for marriage equality.

Conclusions

Support for marriage equality in Australia can be predicted by social attitudes and demographic variables. These findings are discussed in terms of the implications of maintaining the legal status quo, which mandates marriage inequality.

Key words:

• discrimination
• equality
• marriage equality
• plebiscite
• same‐sex marriage

What is already known about this topic?

1. A range of demographic predictors reliably predict sexual prejudice (or attitudes towards gay men and lesbian women).

2. Gay men and lesbian women are the victims of subtle forms of prejudice in contemporary Australian culture, including the denial of marriage equality.

3. Recent poll data reveals increasingly divided support for marriage equality, although no Australian data has explored predictors.

What this topic adds?

1. A cross‐sectional study revealed a majority support for marriage equality.

2. Positive attitudes towards gay men predicted support for marriage equality—attitudes towards lesbians were unrelated.

3. After controlling for sexual prejudice levels, politically liberal views and being unaffiliated with a religion were the demographic variables that predicted support for marriage equality.

The Australian marriage equality debate

The marriage equality debate (sometimes under‐inclusively referred to as the same‐sex marriage debate) has gained considerable attention in Australian political and social discourse. The history of same‐sex marriage legalisation in Australia has been brief and tumultuous. In 2009, the Australian Marriage Bill (The Parliament of the Commonwealth of Australia, 2009) was amended to redefine marriage as the union of a man to a woman by adding the wording ‘to the exclusion of all others’ to the bill (Attorney‐Generals’ Department, 2009, subsection 5[1]). In amending this law, the Australian Government actively removed legislative ambiguity, and in doing so continued and enforced the denial of an individual to marry a same‐sex partner. In 2013, same‐sex marriage was legalised in the Australian Capital Territory (The Marriage Equality (Same Sex) Act 2013, Australian Capital Territory Government, 2001). However, 5 days later, the federal High Court unanimously voted to revoke the Act in its entirety, voiding the 31 marriages that had already taken place (Marszalek, 2013). For many (e.g., Byrne, 2013), this is considered a clear example of institutionalised discrimination and provides support for arguments that prejudices based on sexual orientation are socially acceptable (Pereira, Monteiro, & Camino, 2009).

Recently, the Australian government proposed a marriage equality plebiscite. This nationwide vote would ask the Australian public whether they support a change in the law that would allow same‐sex couples to marry, with a simple national majority determining the non‐binding result (Australian Electoral Commission, 2015). The Australian Government had budgeted $170 million to run the plebiscite (Ryan & Brandis, 2016); others estimate costs upwards of $500 million, once campaigning and lost productivity figures are accounted for (Keany, 2016). Financial issues are of some concern, yet of more interest to psychologists are concerns pertaining to the impact on a vulnerable social group.

Undisputedly, public ‘no’ campaigns would add increased strain on the mental health and wellbeing of the non‐heterosexual community. Australian data have shown that when non‐heterosexual individuals are exposed to articles and advertisements opposing same‐sex marriage they are significantly more likely to report feelings of negativity, loneliness, and depression (Barlow, Dane, Techakesari, & Stork‐Brett, 2012). These findings allude to the potential emotional impacts that the proposed plebiscite may have on the Australian lesbian, gay, bisexual, transgender, intersex, and queer (LGBTIQ) community. It is unclear whether or not this plebiscite will eventuate. However, the fact that it has been scheduled and cancelled and is repeatedly being debated in the Australian parliament (Ryan & Brandis, 2016) is evidence of the ongoing and oscillating nature of the discourse on marriage equality. Thus, in this article we aim to examine predictors of support for, or opposition to, marriage equality in the Australian context.

Sexual prejudice and attitudes towards marriage equality

Surprisingly little is known about the predictors of support for marriage equality, and indeed no Australian literature yet exists on this topic. Attitudes towards same‐sex marriage have been linked (in Western samples) to the demographic variables of gender (e.g., Parrott, 2009), religious affiliation (e.g., Harrison & Michelson, 2015), political orientation (e.g., Brown & Henriquez, 2008), and age (e.g., Becker, 2012), and while this knowledge is important, it is worth noting that these variables have not yet been considered simultaneously. The consideration of these factors as simultaneous predictors in the same statistical model would provide evidence for the relative strength of each of these predictors, and subsequently afford a more holistic picture of the demographic factors related to support for marriage equality. Interestingly, but not surprisingly, each of these variables also predict sexual prejudice more broadly (negative attitudes towards individuals because of their sexual orientation; see Herek & McLemore, 2013). Below, we will briefly review the literature on demographic correlates of sexual prejudice, in order to formulate predictions for this article.

One of the most consistent predictors of sexual prejudice is gender; men typically report more sexual prejudice than women (Herek, 2002; LaMar & Kite, 1998), and this effect is stronger towards gay men than towards lesbians (Herek, 2002; Herek & Capitanio, 1999). This may be because male homosexuality deviates from traditional gender roles more so than female homosexuality (Basow & Johnson, 2000), or because deviation from masculine gender roles by men is seen as a violation of social privilege (Kerns & Fine, 1994). If this argument explains gender differences in sexual prejudice, it could also be used to justify predictions of gender differences in same‐sex marriage support (see Parrott, 2009).

A second consistent predictor of sexual prejudice is religion, which has been documented as a strong predictor of attitudes towards gay men (ATG) and lesbians (ATL), as well as same‐sex marriage beliefs. Religiously affiliated individuals often see non‐heterosexual relationships as a violation of religious values (Harrison & Michelson, 2015). As such, reported levels of sexual prejudice are often the most negative in Christian samples (Jonathan, 2008; Laythe, Finkel, Bringle, & Kirkpatrick, 2002), as well as individuals belonging to other denominations (e.g., Muslims; Adamczyk & Pitt, 2009; Anderson & Koc, 2015). Research has shown that the relationship between religion and support for marriage equality is often negative. Among those who are religiously affiliated, the attributional nature of non‐heterosexual relationships accounts for a considerable proportion of variability in support for marriage equality. Those belonging to a religion are more likely to view these relationships as a choice and easily changeable, which in turn results in less favourable endorsements, in comparison to those not religiously affiliated (i.e., Atheists, agnostics, etc.; Whitehead, 2010).

Research tends to show that older people report higher levels of sexual prejudice than younger people (Kelley, 2001; Lewis, 2003). This is likely an artefact of a liberal cohort effect, which is a trend describing each generation as more socially tolerant than its predecessor (e.g., Crockett & Voas, 2003). An inverse relationship between respondent's age and support for marriage equality has previously been documented by Becker (2012) who found that younger individuals (i.e., millennials) report greater tolerance and acceptance of same‐sex marriage than older individuals (i.e., baby boomers). Moreover, younger and middle aged individuals are more likely to actively express these views in comparison to individuals over the age of 50 (Becker & Scheufele, 2011).

Finally, political orientation has been found to predict support for marriage equality. Those subscribing to a conservative political ideology typically tend to display greater levels of sexual prejudice (Herek & Capitanio, 1996) than those subscribing to a more liberal political ideology. Right‐wing authoritarianism, a construct which has been found to strongly correlate with political conservatism, is also a significant predictor of sexual prejudice (Hunsberger, 1996; Stones, 2006). More recently, research has revealed that highly conservative individuals are more likely to report negative attitudes towards marriage equality in comparison to their less conservative counterparts (Brown & Henriquez, 2008; Olson, Cadge, & Harrison, 2006). Taken together, a review of the literature has revealed that gender, religious affiliation, political orientation, and age predict both sexual prejudice and attitudes towards same‐sex marriage, however the majority of this research has been conducted outside of Australia and these predictive power of these variables have not yet been considered in unison.

The current study

Attitudes towards marriage equality in Australia are becoming more divided; while proportions of individuals supporting marriage equality is increasing, proponents on each side of the debate are becoming more convicted in their positions (Essential Media Communications [EMC], 2016), and debates continue to dominate political and media rhetoric on the topic. Surprisingly, no local research has yet explored predictors of support for marriage equality. The major aim for this article is to establish the demographic predictors of support for marriage equality. Given that we can assume sexual prejudice would predict a lack of support for marriage equality, we are interested in the demographic variables that account for variance in marriage equality beyond that accounted for by ATG and ATL. Thus, the article presents a correlational study in which we predict support for marriage equality from demographic factors and sexual prejudice. Specifically, ATG and ATL, and four variables known to predict sexual prejudice (gender, religious affiliation, age, and political orientation) can be used as predictors of support for (or against) marriage equality, and the predictive strength of each of these variables can be established. Specifically, we predict that lower levels of sexual prejudice, and each of these prejudice‐relevant predictors (i.e., being female, non‐religious, young, and politically liberal) will predict marriage equality support in an Australian sample.

METHOD

Participants

The sample comprised 137 Australians (age range: 18–61 years, M = 25.42, SD = 10.11, 91 females) who were recruited online using advertisements on social media. Participation was incentivised by chance to win one of five gift cards (valued at $AUD 25 each).¹ The majority of the sample had a secondary school education (70.6%), with some having an undergraduate degree (23.5%) or a postgraduate degree (4.4%), and the remainder not having completed secondary school (1.5%). On average, the sample reported a neutral political orientation (M = 3.36, SD = 0.92 by responding to a scale ranging from 1 [politically conservative] to 7 [politically liberal/progressive]). The majority of the sample identified as heterosexual (83.9%; 4.4% identified as gay, 8% identified as bisexual, and 3.7% did not disclose this information),² and approximately half of the sample reported being religious (Christian: 50.4%; Jewish, Hindu, and Muslim: 1.5% each), with the remainder being non‐religious (Atheist: 21.9%; agnostic: 13.1%; ‘do not believe’: 8.8%).

Measures

Sexual prejudice was measured using the Attitudes Towards Lesbians and Gay Men Scale (Herek, 1984), which comprises 10 items measuring negative ATL (e.g., ‘Female homosexuality is a sin.’; α = .77) and a further 10 items measuring negative ATG (e.g., ‘I think gay men are disgusting.’; α = .82). Participants endorse these statements on a scale ranging from 1 (strongly disagree) to 7 (strongly agree). After reverse scoring appropriate items, average scores are made for each subscale, with a higher score reflecting the explicit endorsement of sexual prejudice towards lesbians and gay men, respectively.

Support for marriage equality was measured with a single item. Specifically, participants were asked ‘We would like to ask you your personal opinion (remember there is no right or wrong answer, we are interested in your honest opinion)—Do you support Australia allowing marriage between two people, regardless of their gender?’ and responded with either 0 (no) or 1 (yes).

Procedure

Participants read the consent form online, and after agreeing to participate were forwarded to the website hosting the survey. After responding to demographic questions (i.e., age, gender, sexual orientation, highest achieved level of education, political orientation, and religious affiliation), participants responded to the single item questioning them on their support for marriage equality, and then finally the sexual prejudice measure. Participants were then debriefed to the aims of the project, and thanked for their time.

RESULTS

Descriptive and correlational findings

Descriptive statistics and correlation coefficients are presented in Table 1. Prior to interpreting results, data were screened for assumption violations. Two cases on the ATL (zs = 3.95, 3.30) and a further three cases on the ATG scales (zs = 3.26, 3.26, 3.13) were treated as outliers (i.e., replaced with M − 2 × SD; Tabachnick & Fidell, 2007a). Participants’ age and attitudes towards both lesbians and gay men were positively skewed; issues of normality were successfully corrected using logarithmic transformations. On average, sexual prejudice scores were low (significantly lower than the statistical midpoint of the scale, t(114) = −13.94, p < .001). The majority of the sample reported that they would vote in favour of marriage equality (78%; statistically more than would be expected by chance, χ ²(1) = 35.93, p < .001, and close to estimates expected from the general community based on poll data which suggests that 74% of individuals are not against marriage equality; EMC, 2016).

Table 1. Correlations of sexual prejudice, support for marriage equality, and demographic variables (N = 137)

	1	2	3	4	5	6
1. ATL	—	—	—	—	—	—
2. ATG	.68**	—	—	—	—	—
3. Gender	.33**	.30**	—	—	—	—
4. Religion	.24^*	.38**	.04	—	—	—
5. Politics	−.33**	−.44**	−.20^*	−.37**	—	—
6. Age	.14	.24^*	.04	.04	−.08	—
7. Marriage equality	−.43**	−.56**	−.22^*	−.39**	.43**	−.23^*

Note. Significant predictors and their statistics are presented in boldface. Higher politics scores indicate liberal political beliefs; positive correlations with gender indicate a point‐biserial correlation with being male; positive correlations with religion indicate a point‐biserial correlation with being affiliated with a religion; positive correlations with marriage equality indicate a point‐biserial correlation with support for marriage equality. ATL = attitudes towards lesbians; ATG = attitudes towards gay men.

* p < .05

** p < .001.

Correlation analyses were conducted between support for marriage equality, the sexual prejudice measures, and demographic variables. A bivariate correlation analysis revealed that both sexual prejudice and support for marriage equality were related in the expected direction to political orientation, religious affiliation, and gender. Specifically, decreases in sexual prejudice and support for marriage equality were correlated with political progressiveness, being non‐religious, and identifying as female. Support for marriage equality was also related to age (specifically to being younger). Finally, sexual prejudice (negative attitudes towards both gay men and lesbians) was correlated with less support for marriage equality.

Inferential statistics

Participant group‐differences

Pearson's chi‐square tests revealed both gender and religious affiliation differences in support for marriage equality. Non‐religious individuals were statistically more likely to support marriage equality than religious individuals, χ ²(1) = 20.67, p < .001, and females were more likely to support marriage equality than males, χ ²(1) = 6.31, p = .017. Paired samples t‐tests revealed that attitudes were more negative towards gay men (M = 2.30, SD = 1.44) than lesbians (M = 2.10, SD = 1.24, t(136) = −2.46, p = .015 Cohen's d = .15).

Predicting marriage equality support

We conducted a hierarchical logistic regression in order to predict support for marriage equality from sexual prejudice (Step 1) and prejudice‐relevant demographic variables (Step 2; see Table 2). A test of a model with sexual prejudice predictors (i.e., ATL and ATG) against a model with only the constant was statistically significant. This finding suggests that levels of sexual prejudice can be used to reliably distinguish between individuals who support marriage equality from those who do not, χ ²(2) = 56.37, p < .001, and that this model is a good fit H–L statistic p = .130. Prediction success of the model was 89.30% (interestingly, our model predicts people who intended to vote for marriage equality with a very high accuracy [94.30%], but was much less accurate for those intending to vote against [68.00%]).

Table 2. Coefficients of the model predicting support for marriage equality from sexual prejudice or relevant demographic factors (N = 137)

	B	SE b		95% CI for odds ratio		Wald	p
			Lower	Odds	Upper
Step 1
ATL	−0.04	0.85	0.03	0.96	36.14	0.00	.983
ATG	−9.67	2.52	0.00	0.00	0.01	14.70	<.001
Step 2
ATL	4.30	2.85	0.27	73.30	19699.84	2.26	.132
ATG	−14.19	4.62	0.00	0.00	0.06	9.42	.002
Gender	−1.50	0.91	0.04	0.22	1.35	2.67	.102
Religion	−4.14	1.51	0.00	0.02	0.31	7.48	.006
Political orientation	2.49	0.86	2.23	12.11	65.77	8.34	.004
Age	−1.93	2.75	0.00	0.15	31.99	0.49	.483

Note. Significant predictors and their statistics are presented in boldface. Step 1: constant = 5.56 (SE = 1.03), model: χ ²(2) = 56.37, p < .001; Step 2: constant = 5.57, SE = 3.96, model: χ ²(6) = 83.81, p < .001. ATG = attitudes towards gay men; ATL = attitudes towards lesbians; CI = confidence interval.

The inclusion of prejudice‐relevant demographic variables in the second step improved the model, χ ²(6) = 83.81, p < .001, H–L statistic p = .97. Prediction success of the final model was 93.10% (interestingly, our model could predict people who intended to vote for marriage equality with a very high accuracy (96.20%), but was less accurate for those intending to vote against (80.00%). Inspection of the odds ratio suggests that individuals from this sample who reported support for marriage equality are more likely to be non‐religious, politically liberal, and have positive ATG (but not lesbians). Conversely, it could be interpreted that individuals from this sample who reported opposition to marriage equality are more likely to be religious, politically conservative, and have negative ATG (but not ATL).

DISCUSSION

This article presents a study exploring the demographic predictors of support for marriage equality above and beyond ATG and ATL. In a relatively egalitarian sample (typically reporting low levels of sexual prejudice and high levels of support for marriage equality), we found partial support for our hypotheses. Specifically, we found that support for marriage equality was related to religious (non)affiliation, political liberal ideology, and lower levels of sexual prejudice (but only when the targets were gay men). Previous research has also found that politically liberal individuals and people who self‐identify as non‐religious are more likely to endorse marriage equality (Brown & Henriquez, 2008; Olson et al., 2006; Whitehead, 2010). For example, matching our findings, Lewis and Gossett (2008) used six waves of US public poll data that assessed support for marriage equality, and found that liberals (i.e., Democrats) and the less religious had become more accepting of gay men and lesbian women, and more supportive of marriage equality, over the previous two decades. Conversely, levels of opposition to marriage equality remained unchanged in conservatives (i.e., Republicans) and Protestants.

We did not find support for the parts of our hypotheses which posited that gender or age differences would predict support for marriage equality. One likely explanation for this is, that in the specific statistical model presented in this article, the variance was accounted for by the other variables. Indeed, given that all variables entered into the logistic regression model correlated with the dependent variable, and that gender differences existed, it can still be argued that these variables are related to support for marriage equality, just that they do not help explain this support beyond the role of religious affiliation and political orientation.

Lower levels of sexual prejudice were related to marriage equality support, but only when the attitude‐targets were gay men, and not when they were lesbian women. This was in contrast to previous research by Herek (2009) who found that positive attitudes towards both gay men and lesbian women predicted support for marriage equality. This anomaly could be explained by several factors. First, Herek's research was conducted in the US and some 7 years prior to the current research, which is short in absolute terms, however changes in the history of marriage equality in Australia have all happened within this time frame. Second, ATL are typically less negative than ATG (Herek, 2002; Herek & McLemore, 2013), and perhaps this effect is strengthening over time. Finally, a speculative explanation is that Australian attitudes towards two women getting married might be qualitatively different to (and potentially more acceptable than) attitudes towards two men getting married. The potential for this final explanation cannot be commented on by the data in this article, as marriage equality for both genders were assessed in the same measure of the dependent variable.

Implications, limitations, and future directions

One direct application of the findings presented in this article is that the identified demographic predictors of support for marriage equality also allow the prediction of opposition to marriage equality. By extension, this will also facilitate the prediction of specific social and political groups in Australia who are likely to oppose marriage equality. For example, some conservative religious groups (e.g., the Australian Christian Lobby) and political groups (e.g., the One Nation Party) publically oppose marriage equality. It is likely that other religious and politically conservative individuals and social groups are also against marriage equality, but do not make this position known in a public arena due to evolving social norms around expectations to present as non‐prejudiced or egalitarian (Plant & Devine, 1998; Smith & Matějů, 2012). Thus, given the negative health implications of opposition to marriage equality for the LGBTIQ community, this research can assist in targeting those who oppose marriage equality in order to raise their awareness of these implications.

The current study has some limitations that warrant consideration when interpreting these results. Some limitations pertain to the statistical analyses presented within. Specifically, the data set could be considered somewhat problematic because of the disparity between the proportions of individuals in the sample supporting marriage equality (78%) relative to those who do not (22%). Although this disproportion is large, it is reflective of recent survey poll data on this topic (EMC, 2016), and logistic regression is statistically capable of dealing with this disparity (Tabachnick & Fidell, 2007b).

Other limitations pertain to the sample; it arguably is not large enough or diverse enough to allow the findings to be easily generalisable. The sample size is large enough for the statistical analyses used, and is relatively representative (distributions of education, religious affiliation, etc.). However, the sample is relatively young (a variable known to predict political liberalism, and specifically acceptance of LGBTIQ individuals and support for marriage equality), and the data were collected only in Victoria. Future studies in this field could consider using a larger sample that is selected using stratified random sampling techniques (i.e., classifying the population on relevant background variables, and randomly selecting an appropriate proportion of respondents from each stratum). Future research could also consider the role of other demographic variables (e.g., education, location, etc.), ideological factors (i.e., individual differences in authoritarianism, personality, etc.), and the role of social desirable responding in both how participants respond to the measure of sexual prejudice or their level of support for marriage equality.

We would like to highlight two further points for consideration when interpreting these findings. First, this is not meant to be a poll to provide an index of support for marriage equality in Australia, and indeed the limited and non‐representative sample means that the reported level of support by the sample is unlikely to match levels of support in the wider community. Instead, this is supposed to simply provide evidence for the predictors of marriage equality support, and these findings should not be over‐interpreted, nor over‐generalised beyond what is appropriate for this sample. Replication and extension of these findings to other individual differences is warranted. Second, we would like to acknowledge that the marriage equality debate encompasses legalised inequalities that impact social groups beyond just gay men and lesbian women. In this article we used a quantification of sexual prejudice that specifically comprised attitudes towards gay men and lesbian women. While it is outside the scope of this article, this debate also impacts the lives of a range of other individuals identifying as non‐heterosexual, including bisexual and transgender individuals.

Concluding remarks

Public opinion on marriage equality in Australia is divided. In this article, we present findings revealing that support for marriage equality is largely predicted by positive attitudes towards gay men, and also being politically liberal and not affiliated with religion. In closing, this article presents evidence for the demographic predictors of Australians’ support for, or opposition to, marriage equality and contributes to the under‐developed literature on this topic. We close by suggesting that more research is needed to understand other individual differences that are associated with opinions on this topic to facilitate a meaningful and respectful progression on the marriage equality debate.

Notes

1. If students were enrolled in a psychology degree at the hosting institution, they could instead request they receive partial course credit instead of being included in the draw for gift cards.

2. Analyses were run on both the full sample and a sub‐sample comprising only heterosexual participants. There were no substantial differences between these analyses, so the full data set was used in all analyses presented within.