Could polls be wrong about outcome of Ireland's gay marriage referendum?
|
Version 0 of 1. On Friday, Ireland could become the first country in the world to legalise same-sex marriage through a popular vote. According to the most recent polls Irish voters are expected to overwhelmingly vote in favour of the amendment to the country’s constitution. Three sets of figures released over the weekend give the yes vote a lead of about 40 points once undecided voters are excluded. An Ipsos MRBI survey for the Irish Times has yes on 70%, Millward Brown for the Sunday Independent has it on 69% and a Red C poll for the Sunday Business Post gives it 73%. The yes vote lead seems unassailable – the polls would need to be dramatically wrong for the no vote to win. Although an error of that scale is improbable, there are several factors of uncertainty to consider, and that could mean the result ends up closer than these figures imply. Yes lead remains commanding but it has narrowed in recent weeks Taking into account the “don’t knows”, the Millward Brown numbers are yes 53%, no 24% and 23% don’t know. In March, the figures were 66%, 21% and 13% respectively. The Ipsos figures portray a similar trend. Its most recent poll has those saying yes on 58% (down six points since March), the nos on 25% (up two points) and the undecideds and those who say they won’t vote on 17% (up five points). But the fall in support for the yes camp is nowhere near as sharp as the drop in support for legalising divorce in Ireland in the 1995 referendum. Back then, the yes side had a 44-point lead in the weeks before the vote. The referendum eventually passed by 0.6 percentage points. In fact, in this case, the drop appears to be driven more by an increase in the number of undecided voters than in a rise for the no side of the argument. Polls before referendums in Ireland have been wrong before Polling before a vote in 2013 to abolish the senate pointed to a clear majority for the yes vote. Ipsos MRBI had those in favour on 62% compared with 38% for retaining the upper house of the Irish parliament. Millward Brown expected a similar outcome: 64% in favour, while 36% against, according to its figures. In the event, no (to abolishing the senate) won with 52% of the votes. It is is also important to point out that the proportion of undecided voters were also far higher then – 29% in the Ipsos poll and 42% in Millward Brown’s. However, it wasn’t the first time that polling on an issue that triggers strong emotions on both sides of the debate was off the mark. In Switzerland, which holds regular referendums and where polling is generally accurate, a vote to introduce immigration quotas passed by 50.3% in February 2014, whereas polling was pointing to a small majority for those against the measure. And in 2009, a ban on the construction of minarets was passed with 57.5% of the vote. Then, too, polls had the no vote in the lead before the referendum. In all these debates, there may be greater shyness on the side of the debate that feels their view is less publicly acceptable. In fact, in the Ipsos poll on equal marriage, when those surveyed were asked if they felt they could not openly discuss the issue, 28% of no voters fell into this category compared with only 9% of yes supporters. Turnout may prove decisive in Friday’s vote When polls get it wrong, one of the possible explanations is usually a difference between the expected turnout implied by polling and actual turnout on the day. It is one of the several reasons put forward for the recent polling failures in the UK general election. In the Ipsos equal marriage survey, 71% of 18- to 24-year-olds say they intend to vote yes and 15% no. By contrast, among the over-65s, 52% intend to vote no and 34% yes. If there is a low turnout on Friday then the outcome could be closer than implied in polls because older age groups are generally more likely to vote. But support for yes is also strongest among the middle-class and urban voters, and weaker in rural constituencies (where only a minority intend to vote yes) and past referendums turnout among the former two groups tends to be higher compared with the latter. Ireland will most probably vote to introduce equal marriage. The yes side holds a commanding lead, and almost eight in 10 voters say they have made up their mind. And even once all factors of uncertainty are considered, the no vote would need, as the Irish Times puts it, a seismic shift to overturn that lead. But, if the no vote does win, it wouldn’t be the first time the polls are wrong. The Ipsos poll was carried out among among a representative sample of 1,200 voters, aged 18 and over, in face-to-face interviews at 120 sampling points in all constituencies. The margin of error is +/-2.8%. The Millward Brown figures are from a face-to-face poll of 994 respondents. It was taken at 64 sampling points nationwide from 2-15 May. The margin of error is 3.1% . |