After the same-sex marriage vote, what next? Public opinion and policy responsiveness

We care about public opinion and policy congruency for normative and practical reasons. For reasons of values, if nothing else, some form of policy representation is necessary for democratic government.

However, this link between voters’ preferences and policy also has practical implications. As Downs argued back in 1957, elected officials are induced to respond to voters’ preferences if they wish to be (re)elected (see: Antony Downs in 1957 and many others).

In my paper with Andrea Carson and Yannick Dufresne (forthcoming in the Australian Journal of Political Science, available here), we examined whether Australian federal legislators’ policy positions are congruent with public opinion.

We used same-sex marriage as a case study. We hypothesised that parties would generally represent the preferences of their constituents, but that this representative behavior would be imperfect for two reasons.

First, because we believe Australia’s major political parties are interest aggregators, representing the policy interests of candidates, party activists and donors, as well as their constituents (discussed in another paper of mine published in the AusJPS, and another post on this blog).

Second, because we expect to see a form of status quo bias, with groups (lobby, interest and party factions) use their political influence to veto change in political parties’ policy formation; with opposition being easier than advocacy for large-scale policy change (see for instance Gilens 2012)

Given the characteristics of same-sex marriage debate, it highlights the impediments to perfect representation.

Same sex marriage as a case study

Since 2007, Australian public opinion polls indicated most Australian citizens support same-sex marriage (with polls indicating support above 50% since 2007, and hovering between 57-68% since 2010). Despite changing social norms, the federal Marriage Act has still not been amended and the parliament remains divided on this issue.

Even more so, the centre-right federal Coalition is divided. Within the parties that make up the Coalition, most MPs oppose same-sex marriage, although a significant minority within the Liberal Party (the senior Coalition partner) are strong supporters. The issue also in part a proxy for ongoing competition between moderates and conservatives, and over the party leadership and its future.

To resolve this impasse, the current Coalition government – led by high-profile supporter for same-sex marriage, Malcolm Turnbull – organised for the Australian Bureau of Statistics to conduct a voluntary, non-binding “survey” on same-sex marriage (after attempts to pass legislation directing the Australian Electoral Commission to conduct a plebiscite were twice blocked in the Senate).

At 10am on Wednesday 16 November in Canberra, the Chief Statistician of the Australian Bureau of Statistics (ABS) announced that 61.6 per cent of voters supported a change to the Marriage Act.

The process was designed to inform parliamentarians’ support and translate into legislation before the end of the year. Given that a number of MPs positions are contrary to how their constituents voted,  do we expect federal parliamentarians to vote consistent with their constituents’ preferences on this issue, though?

In our paper, we used a mixed methods approach to test the relationship between public opinion in a parliamentarian’s electorate, and their position on this issue between 2012 and 2016.

During this time the first private members bill (2012) was introduced into the lower house prompting members to reveal their public positions. Given the Coalition were compelled to vote against the Bill, it was unremarkable it was defeated 98 to 42. We tracked shifts in MP’s positions (including new MPs) over four years.

Estimating public opinion

Our study examined the policy responsiveness of legislators to public opinion using unique large-scale public opinion data sourced from the 2013 Vote Compass voter advice application (which provided usable data on 601,550 voters), ABS census data, and election outcomes data from the Australian Electoral Commission (AEC). We also undertook an analysis of newspaper databases, MPs’ and activists’ websites and Hansard to verify and construct a database of MPs’ positions on same-sex marriage.

From the Vote Compass data, we used responses to the statement ‘Marriage should only be between a man and a woman’ as the dependent variable in the first part of our analysis, coded as a binary proposition (yes, only between a man or woman, opposing same-sex marriage; no or neutral, not opposing same-sex marriage). We used multilevel regression with post-stratification (MRP; for more information on the methodology of MRP see this paper) to ensure that our public opinion estimates were based on a representative model of the electorate.

Consistent with commercial surveys at the time, and the result of the postal “survey”, our model estimated 33 per cent of the electorate opposed same-sex marriage (with 66 per cent supportive or undecided).

The congruence between public opinion and policy responses 

We modelled the congruence between public opinion and the positions taken by lower house parliamentarians MPs.  Using opposition to same-sex marriage as a function of our estimates of public opinion in the divisions they represent (made with the Vote Compass data). We start this analysis using the recorded positions taken by parliamentarians in their vote for the Marriage Amendments Bill 2012 (for the breakdown see here).

Given Coalition members were bound by party policy, there are limits to what we can learn from these data. We update this with a second analysis of MPs’ stated opposition and support by early 2016,[1] obtained from legislators’ statements in media between 2012 and 2016, which we used to build a database on their positions.

Congruence in 2012

We estimate the congruence between public opinion and the positions taken by members of the Australian House of Representatives by fitting a logistic regression to parliamentarians’ votes on the Marriage Amendments Bill 2012.

The first figure shows the results from the models fit to the positions taken by Labor MPs. These show the probability MPs would vote against the legalisation of same-sex marriage as a function of public opinion (opposition) in their electorate. Probabilities are represented by the curves plotted in these graphs (the Labor curve and CI contain noise due to simulation variability). We include the actual positions taken by legislators compared to estimated public opinion in their electorate, represented by a separate point for each MP. I did not fit this model to Coalition parliamentarians due to their binding party vote (with all their MPs voting against the bill), but have shown them in the figure to show the distribution of public opinion for their electorates.

As might be expected by an advocate of the normative value of representation in a democracy, the probability a Labor parliamentarian would oppose a change in legislation concerning same-sex marriage increased as the estimated opposition in their electorate did (or conversely, a Labor MP representing a more supportive electorate was more likely to vote for the legalisation of same-sex marriage in parliament).

Probability of MP support for Same Sex Marriage 2012

Probability of MP support for Same Sex Marriage

Figure 1: Policy congruence on same-sex marriage in the house of representatives, 2012. From Carson, Ratcliff and Dufresne (forthcoming)Each curve represents the probability a member of the house of representatives would vote against the second reading of the Marriage Amendment Bill 2012, which would have legalised same-sex marriage, conditional on opposition to same-sex marriage in their electorate. Points represent the actual position of each legislator compared to estimated public opinion. Curves are simulated probabilities of opposing the legislation. Shaded areas 95% confidence intervals.

It’s possible that legislators representing more marginal divisions – more at risk of losing their place in parliament – may be more responsive to public opinion. We tested this by fitting a second specification that included the margin of each MPs vote at the 2013 election,[2] and a third where both estimated public opinion and margin are included, and allowed to interact. The results of this third specification are shown in figure 2, which shows only small differences in responsiveness to public opinion between Labor MPs irrespective of whether they represent safe or marginal electorates.

Labor MPs show evidence of supporting changes to the Marriage Act when there are higher rates of support to same-sex marriage in their electorates, even after we control for their margin.

There is also evidence of status quo bias in parliament. Opposition below 40 per cent was required in a Labor electorate before chances were greater than even that its representative would support this legislation. As expected, the probability Labor MPs would oppose the legalisation of same-sex marriage continued to decline as opposition in their electorate did, down to less than 25 per cent as estimated divisional opposition declined below 30 per cent. It is impossible to draw strong conclusions from the behaviour of Coalition MPs, because of the enforced bloc vote.

Labor MPs

Marginality and policy congruence for Labor MPs on same-sex marriage in the house of representatives, 2012

Figure 2: Marginality and policy congruence for Labor MPs on same-sex marriage in the house of representatives, 2012. From Carson, Ratcliff and Dufresne (forthcoming)Each curve represents the probability a Labor MP would vote against the second reading of the Marriage Amendment Bill 2012, conditional on the estimated opposition to same-sex marriage in each legislator’s electorate, and their margin at the 2013 election. “Marginal seats” are those with margins of zero, “average” with margins of 10 per cent, and “safe seats” 20 per cent.

However, we do learn two things from the unified opposition of the centre-right parties. First, a majority of Coalition parliamentarians were willing to bind themselves and their colleagues on this issue (via the party room). Second, no backbenchers crossed the floor and voted for the legislation (which is allowed by the Coalition parties).

Beyond the 2012 parliamentary vote

Since the 2012 parliamentary vote on the Marriage Act, almost all Members of the federal parliament have publicly restated their position on same-sex marriage, with a number shifting to support it.

We do this by fitting a second model estimating how the House of Representatives would vote if given the chance in 2016. The results from these models are shown in figure 3. These indicate there was representational behaviour from MPs of both parties, but that each displayed different patterns.

Labor parliamentarians were less likely to oppose same-sex marriage than their Coalition counterparts, even when they both represent electorates with the same estimated public opinion on the issue. Coalition MPs leaned towards opposition in all but the most supportive of electorates. Labor parliamentarians also continued to exhibit greater congruence with the preferences of their constituents (although because so few Labor MPs opposed same-sex marriage by 2016, there is a higher degree of uncertainty here); with their opposition to same-sex marriage predicted to shift rapidly as public opposition declined.

Policy congruence on same-sex marriage in the house of representatives between 2012 and 2016

Policy congruence on same-sex marriage in the house of representatives between 2012 and 2016.

Figure 3: Policy congruence on same-sex marriage in the house of representatives between 2012 and 2016. From Carson, Ratcliff and Dufresne (forthcoming)Note: Each curve represents the probability a member of the house of representatives would oppose same-sex marriage, based on the public position they had taken between 2012 and 2016, conditional on the estimated opposition to same-sex marriage in each legislator’s electorate. Points represent the actual position of each legislator compared to estimated public opinion.

Most Coalition parliamentarians continued to show a strong ideological preference towards opposing same-sex marriage (even in a hypothetical free vote context). Those representing divisions where estimated opposition was 40 per cent (60 per cent public support or no position) were predicted to have approximately a 95 per cent chance of opposing amendments to marriage legislation. It was only once same-sex marriage opposition declined below an estimated 30 per did the probability of a Coalition MP opposing changes to marriage laws drop below 50 per cent (and then only by a small margin).

These results also indicate parliamentarians continued to exhibit status quo bias. If half the voting public in one electorate opposed same-sex marriage, it was almost certain their MP (from either major party) would too. Labor MPs representing an electorate with approximately 50 per cent opposition (or, 50 percent supporting or indifferent) would have had an 80 per cent probability of opposing same-sex marriage. Coalition MPs representing constituents with these high levels of estimated opposition had greater than 95 per cent probabilities of opposing changes to the marriage laws. However, time matters. Overall MP preferences for the status quo had weakened by 2016 compared with 2012.

When we examined how the safety of an MPs seat in parliament influenced their position on same-sex marriage by 2016 (figure 4) we find a similar pattern for both parties regardless of marginality:  greater electorate opposition results in greater MP opposition. While we should not over-interpret these findings, overall Labor parliamentarians are more responsive in marginal electorates; whereas Coalition MPs become more representative to voters’ preferences in safer divisions.

Marginality and policy congruence on same-sex marriage in the house of representatives between 2012 and 2016

Marginality and policy congruence on same-sex marriage in the house of representatives between 2012 and 2016

Figure 4: Marginality and policy congruence on same-sex marriage in the house of representatives between 2012 and 2016. From Carson, Ratcliff and Dufresne (forthcoming)Each curve represents the probability a member of the Australian house of representatives would oppose same-sex marriage, based on the public position they had taken between 2012 and 2016, conditional on the estimated opposition to same-sex marriage in each legislator’s electorate, and their margin at the 2013 election. Solid lines represent legislators in divisions with margins of zero (marginal seats), dashed lines with margins of 10 per cent, and dotted lines 20 per cent (safe seats).

 What does this mean for parliament’s response to the Australian Marriage Law Postal Survey?

We find parliamentarians in Australia represent the preferences of their constituents; albeit imperfectly. Congruence has gradually increased between the estimated preferences of an electorate and its representative’s public position since 2012. This suggests the enduring, stable majority public support for same-sex marriage over the past decade in Australia is slowly overwhelming the status quo bias and other policy interests that oppose it.

What does this mean for the outcome of a vote on a marriage bill?

We might still expect some status quo bias. Opposition to same-sex marriage was in the majority in just 17 electorates. However, it would not be surprising if opposition in parliament was higher than this, with some Coalition MPs in particular voting (or abstaining) even if their electorates voted in support. An example of this

In the other direction, some Labor MPs whose electorates were strongly in opposition (such as Blaxland in Western Sydney, which is help by Labor’s Jason Clare and where support for same-sex marriage was only 25 per cent) might vote for a change to the Marriage Act.

Based on our results and statements from MPs, it is likely most will vote to represent the preferences of their constituents and change in the Marriage Act to provide legal recognition for same-sex marriages.

Shaun Ratcliff is a Lecturer at the United States Studies Centre, University of Sydney, Andrea Carson is a lecturer at the University of Melbourne and Yannik Dufresne postdoctoral fellow in political science at Laval University in Quebec

[1] The difference in two-party vote between the parliamentarian who won and their main opponent.

[2] With the exception of NSW, WA and the ACT where it ends in mid-2015, when electoral boundaries were redrawn, potentially disrupting the comparability of earlier and later data, with some MPs expecting to represent a different electorate after boundaries were redrawn, potentially changing their public position on same-sex marriage.