Referenda, like candidate elections, are highly unpredictable in Ireland. While political parties will always retain a “core” vote which doesn’t move at any General Election, referenda can attract support (or opposition) from all swathes of Irish society.
In trying to asses the prospects for the Marriage Equality referendum it’s crucial to put the polls thus far into context and assess turnout amongst key age groups.
What the final polls say
Other polls may emerge ahead of Friday’s vote, but at the time of writing we have four opinion polls; one from each of the main polling companies (Red C, Millward Brown, Behaviour & Attitudes and Ipsos MRBI).
Evident in these polls is a huge variance in the vote. Excluding undecided voters, the YES vote hovers at around 70% while the core YES vote has a range of between 53% and 69%. The sampling dates on these polls covers anything between two weeks (Millward Brown) to two days (Ipsos MRBI).
In seeking to combine these final polls I have added “weights” to those polls which were taken most recently, and have produced the following average:
YES – 60%
NO – 25%
Undecided – 15%
The assignment of undecided voters is a tricky business. Red C assigns one third of undecided voters to the YES camp, with B&A assigning 27%. The remaining polling companies lie somewhere inbetween. I am going to be generous to the NO side and give them 80% of the undecided voters. This means a 3% + 12% split in our undecided voters (15%) on polling day.
YES – 63%
NO – 37%
If we were to follow a split of 1/3 YES and 2/3 NO, the result would be:
YES – 65%
NO – 35%
Will these numbers change by Friday?
Of course. Referenda in Ireland are notorious for tending towards the 50% mark. Polls in the recent Children’s Referendum were stuck in the 60s and 70s for most of the campaign, yet only 58% supported the amendment on the day.
The point of comparison that is regularly referenced in this campaign is the 1995 Referendum on Divorce.
The drop in support for Divorce was incredibly strong, with most of the decline taking place during the campaign itself. To understand this drop, I have created an overlay of the Divorce campaign onto a timeline of polls during the 2015 Marriage Referendum campaign. For the sake of consistency I have included the polling company’s own estimates on the allocation of undecided voters.
The YES lead in this campaign has been solidly above the YES lead in 1995 and the decline in the YES vote has not been as strong. In short, there would need to be a cataclysmic revelation in the campaign over the next two days for this vote to fail. Or, the polls are completely wrong.
The “shy NO” vote
The recent UK General Election has led to an investigation into poor polling accuracy. Many fear the “shy Tory” vote which emerged on May 7th may repeat itself with a “shy NO” vote on May 22nd here. So what if it does?
David Cameron’s Conservative party won 37% of the vote, when polls placed the party at around 34%. If the YES vote is off by 3%, it will not be enough to defeat the referendum. The “shy NO” vote in the Divorce referendum was only 1.4%.
Turnout and age groups
Overall turnout will be a key indicator in this election. We will know turnout figures throughout Friday’s vote with a high turnout likely to favour the YES side. But why do we say this? Put simply it is down to demographics.
The above survey by the CSO must be taken with a pinch of salt. No age group has 90% turnout. A considerable proportion of respondents in this survey did not actually vote in 2011 or 2002. However the graph does capture the traditional apathy towards elections by young people. It also debunks the myth that turnout moves “linearly” throughout age groups (i.e that 65+ voters have the highest turnout, with all preceding age groups having less).
If this survey is accurate, it shows that this referendum will not be won by those in their 20s or defeated by those in their 70s, but decided mainly by those in their 40s and 50s. This generation grew up in the 1980s alongside the AIDS epidemic and in the early 1990s when homosexuality was decriminalised. So how are they voting?
Red C has provided an age breakdown of the ‘No’/’Don’t Know’ voters in their most recent poll. This assumes that every single undecided voter will decide to vote NO on Friday. Despite this allocation, 62% of those aged 54-65 intend on voting YES with 71% of those aged 35-44 intending to vote YES.
So, turnout is often high for those aged 35-65, and a significant majority of these voters are backing a YES vote, but are there enough of them? Surely those who are aged 65 to 100 or higher will eclipse the vote of younger people?
No, once again conventional wisdom does not hold. You would be forgiven for assuming that the 65+ age group was the largest.
The mobilisation of students in this referendum is often highlighted in the campaign, and the sign-up/turnout efforts will have a marginal effect, but in reality the 25-34 age group will have the larger impact. Their likelihood to vote is higher than students, they are only marginally less likely to support a YES vote and most importantly, there’s lots of them!
The narrative of this referendum (like most others of a socially divisive nature) is that of students versus pensioners. Granted, both groups capture the cores of the YES and NO camps, but neither group has sufficient numbers in their own right to swing this election. The quiet middle-aged middle-ground of middle-Ireland will decide this referendum, and so far their vote is nothing but a YES.
Result on the day
With a strong break of undecided voters to the NO camp in the closing days, we’re unlikely to see the 70% YES that has transpired in most polls so far. However little evidence has been produced to suggest that a shy NO vote exists, other than a juxtaposition to our UK neighbours. Determination will drive turnout on both sides, and we’re likely to see brisk numbers at polling stations with the YES vote benefiting most from such a scenario.
My heart wants a YES vote in the 65-70% range, but my head says somewhere between 60% and 65% YES.