When Yes might really mean No: how opinion polls cannot reflect ‘soft’ votes

World of Politics with Harry McGee – harrymcgee@gmail.com

There are many stock phrases that could describe what happened last Friday but none are as apt as the lovely Irish phrase: ‘Tháinig sé aniar aduaidh orainn’ (It [a sudden gale] came upon us from the north west).

I had a feeling for a long time that the vote was going to be No but I must say I was thrown by the opinion polls, particularly our last one which showed a sizeable gap between the sides. In previous referendums, late swings had been captured by the opinion poll. And even though there was a gap favouring the Yes side, everybody knew that the momentum was behind the other side and the pendulum was swinging from Yes to No.

This time it was different. The sample was taken a week before the poll. There was no major intervening event, a St Paul on the road to Damascus moment where voters suddenly saw the light. Sure, there were the TV debates where Micheál Martin did well and it was certain that No campaigners (especially the impressive Democracy Matters) were far more visible and active in the last few days of the campaign.

But did that explain such a dramatic surge, a mobilisation of the 21 per cent of the ‘don’t knows’ to close the 17 point gap between Yes and Nos.

I’m sorry but I just don’t buy it.

Sure, a lot of people made up their minds – and read the literature – in the last three days and truly engaged.

But those who were recorded as Yes’s in so many polls were not really Yes’s. It was too early for them. Citizens who were contacted more than a week out had not really engaged with the issues, had not thought about how they were going to vote. Most would have been vaguely aware of the arguments and would have perhaps remembered the superficial ones of €20 million savings and fewer politicians pushed by Fine Gael. And so they may have said they were voting Yes, when their votes were soft to say the least.

So how many of the 44 per cent of the Yes were only soft Yes’s, transient yes opters who would either not vote on the day or vote No. I’d say at least half.

On the other hand – and in marked contrast – I would warrant that every single one of the 27 per cent of respondents who said they were voting No meant it. They were conviction voters, committed voters who were not going to waver.

Opinion polls are not exact scientific instruments. They only give a broad brushstroke indication, a lie of the land. It’s how you interpret them that’s important. For example, the first thing you have to realise is that while many respondents express a preference, only a portion will vote. Yet, 92 per cent of respondents in The Irish Times poll said they would vote. In real life, less than half of that number voted. In a poll like this when there is a choice between the establishment view and an alternative view, you can take it that the percentage of anti-establishment support will be closer to the true level.

For more, read this week’s Connacht Tribune.