POLITICS

Pollsters Say Marriage Survey Won't Be Another Donald Trump Moment

The marriage equality survey will come back with a 'Yes', experts say.

09/10/2017 5:37 PM AEDT
Getty Images/iStockphoto

If you believe the polls, the marriage postal survey will end with a resounding tick of approval for same-sex marriage -- up to 70 percent, according to some research.

In recent weeks, however, those on both sides of the debate have tried to damp down expectations of such an overwhelming result, citing the voluntary nature of the poll as a threat of skewing polling methods, inexplicably maintaining the 'silent majority' opposes marriage equality, and citing the so-called 'Donald Trump effect' or 'shy Tory' where people tell pollsters one thing but vote the other way.

Experts from two of Australia's biggest polling companies say their methods are sound, and expect the results of the survey -- to be announced in November -- to closely reflect the numbers in their reports. They say the result won't be another like Trump's shock election win.

"The vast majority of people think gay couples are OK, that they should be able to adopt and those things. These are majority views and we know they are because we've been surveying them for years," Michele Levine, CEO of Roy Morgan Research, told HuffPost Australia.

While polls show support for same-sex marriage is the majority view, opponents say it is not quite so clear-cut. For one, pollsters in Australia are nearly always gauging opinion in polls where voting is compulsory, whereas the postal survey is voluntary.

This could mean that, despite the wider public supporting the change, not all of those may actually vote, for a number of reasons including the postal nature of the vote which has previously seen younger people -- those most supportive of marriage equality -- not respond as highly as older people.

As Herald Sun political editor James Campbell said in a Sky News interview last months, "Australian pollsters have almost no experience in estimating turnout, they don't have to."

Jessica Elgood, director of Ipsos polling, said pollsters had considered this factor when publishing their findings. Ipsos conducts random polling, calling around 1000 random landlines and mobiles, then getting demographic information from respondents to weight their results to reflect the national population.

"The approach we took was that we asked people how likely they are to vote, to get a response rate. We have nothing comparable [to this type of vote], this is the big unknown for this," she said.

"What was most interesting is that [response rates] are consistently high across all generations. It divides across age, the older are more against or reticent on the reform, the young are more supportive. Are young people less likely to respond? With voluntary voting, they are less likely, and also the method of a postal survey is that young people are less likely to respond. You have to consider certain groups are less to reply, older people far more likely to send it back, young people less likely."

Levine said it was "dicey" to have voluntary voting in a poll like this.

"These things are very fraught. If there's a skew, it will be to people who care deeply about the issue, and in an inflamed environment, plus only getting votes from people who feel deeply about the issue, who knows which way it will go?" she said.

Levine cited the example of simple phone-in polls run by newspapers or TV stations, and in particular a poll in a conservative newspaper in the 1980s on IVF, as how voluntary voting can skew surveys toward people who have a very strong view on the subject.

"We've surveyed people's attitudes toward test tube babies, IVF, since at least the early 1980s and Australia has been in the 80 percent supportive. Then there was a phone-in poll, a NewsCorp publication, asking people to call one number if they agreed and another if they didn't, and there was a massive vote against it," Levine said.

"Because the people who are against it are very deeply against it, it's not reflecting the population's view but only those who bothered to phone in.

"[The marriage survey result] will land somewhere between the IVF poll, which got it completely wrong, and the truth of the matter. Not everybody is going to vote, the ones who don't vote are the ones who don't mind or care too deeply but are probably OK with it."

Others have claimed that many people are telling pollsters that they will be voting yes in the survey, but will actually be voting no, due to societal pressures.

Elgood called it the 'shy Tory' effect, which pollsters saw in 1980s Britain where polling underestimated the support for Margaret Thatcher as people didn't want to admit they were voting for her. This phenomenon has also been blamed, in part, for why polls got Donald Trump's U.S. presidential election result so wrong.

"It's a valid point to raise. We're conscious of that in all questions we ask, but in some situations there is a social suitability of answers," Elgood said.

"I do think it's interesting to consider whether there is this element of a bandwagon. Maybe people feel they should be saying yes, but in the secrecy of the survey, will they do so?"

She said that, due to the more impersonal nature of a phone poll where a respondent doesn't have to directly address the pollster, she was more confident that people were being honest.

"The more distant the survey method, the argument is that you get greater honesty," she said.

Levine agreed.

"There's always the potential for that kind of thing. It's not about people lying to mislead or push the polls in one direction, but as pollsters we've been aware of a prestige effect, like 'I read Vogue not those trashy gossip magazines'. There are some things people are embarrassed to talk about, that's why a Pauline Hanson vote could be under-represented," she said.

"We've studied these things for many years. We try to make sure our interviewers are so well-briefed there's no judgement in their voice. It's a science but it's a little bit of art as well. We're briefed to ask a question in a completely bland way, there are lots of ways to overcome this."

"The bigger issue to me is the turnout, how many actually vote, and if the polls captured the right people who had really voted or are seriously considering voting."​​​​​

More On This Topic