Although the No side is facing an uphill struggle to overturn a strong Yes lead of 16 points in the final week before the referendum vote, campaigners will take heart from the tendency of recent referendums to feature late swings.
The most cited example is the marriage referendum, which saw an eight point decline in the Yes campaign in the last 10 days, as measured by the final Irish Times/Ipsos MRBI poll of the campaign.
The last poll estimated the Yes vote as leading by 70-30; the final result was 62 per cent in favour to 38 per cent against.
Other polls recorded similar numbers a week or 10 days out from the vote.
A number of other referendums held by the last Government also saw late swings against the Government’s proposal.
In many cases, these were on issues that failed to excite much public interest or attention, and many voters appear to have followed the “If you don’t know, vote No” dictum.
These referendums included the proposals on judicial pay, on Oireachtas inquiries, on children's rights, on the abolition of the Seanad and on the establishment of the Court of Appeal. All saw late swings against the Government-sponsored Yes position, with the Yes vote declining by more than 10 per cent, and in some cases by much more.
If something similar were to happen in the present referendum campaign, it would certainly throw the result into doubt. At the very least, the precedent gives the No campaigners encouragement that they can close the gap in the final week.
However, what the changes demonstrate is not so much that polls were incorrect, but that the days running up to the actual voting in referendums often see sharp movement in voter sentiment. So if this is to be replicated, then this campaign would have to be like the previous referendums instanced above.
But there are significant reasons for believing that this campaign is different. For a start, people know what the question is, and they are aware of the debate which has been raging for months.
Few voters knew, for example, what the Oireachtas inquiries referendum was about. That reduces the scope for volatility.
Secondly, while the measure is being proposed by the Government, the campaign has been led by civic society groups. That diminishes the chances of a large anti-Government vote - and anyway, this Government is not as unpopular as the last one.
Thirdly, there is also evidence in the data that voters are firming up their opinions well in advance of polling day. An overwhelming 96 per cent of people say that they are absolutely not, extremely unlikely or probably unlikely to change their vote - that suggests, again, that the potential for major late volatility is not huge.
And fourthly, while there are substantial doubts about the 12 weeks proposal, there is a greater consensus in favour of changing Ireland’s abortion laws.
There are two further pieces of data from the poll worth considering. The first is when people are asked how they think their close friends will vote, they say the Yes vote is leading strongly (51 per cent to 24 per cent, with 25 per cent undecided).
In addition, when the so-called “wisdom of crowds” approach is adopted - asking all 1,200 respondents what they think the result will be - they predict a 56-44 result in favour of Yes.
The biggest remaining uncertainty about the vote is turnout. And there is an advantage for the No side here. This is a very divisive referendum, and the biggest dividing line is on age.
Older voters are much more likely to vote against repeal - if the referendum was held only among under 50s, it would romp home; if it was held among over 50s, it would crash and burn. And older voters are usually more likely to turn out.
However, the poll shows a high degree of commitment amongst both cohorts; a big effort to get out the vote will be a high priority for both sides.
Referendums have often proved notoriously unpredictable. But there are some reasons to think this one will be less unpredictable than previous votes.
Turnout will be a big part of the eventual outcome. But for all the warning signs from recent history for the Yes side - and they are real and substantial - the data is clear: the advantage remains with the repealers.
