How did the polls get it so wrong in the Seanad referendum?

Post by Kevin Cunningham
PhD Candidate in Political Science at Trinity College Dublin

pollingThe Millward Brown/Independent poll conducted between the 13th and 25th of September suggested a 64 per cent to 36 per cent margin in favour of abolition. The Red C/Sunday Business Post poll conducted between the 9th and 11th of September indicated a 59% to 41% margin. The night before the election, Paddy Power offered 1/10 on a Yes vote compared to a 5 to 1 for a No vote.

The error in these polls was a consequence of insufficiently accounting for the likelihood of an individual to turn out. Although polling firms follow well-recognised guidelines, the guidelines are not necessarily equipped for elections where less than half of the electorate are likely to turn out. Red C, for example, asked respondents to rate from 1 to 10 the likelihood that they will vote. They discard only those who rated their likelihood as being between 1 and 3. This is eminently sensible for a General Election, however by including people who rated themselves as having a 4 out of 10 or 5 out of 10 chance of voting the poll includes large numbers of people with too low a likelihood of voting. It would appear that those more likely to vote were those in the ‘No’ camp.

With this in mind, some proponents of the ‘Yes’ campaign have complained that a lower turnout may have been to blame for the result. The suggestion is that a higher turnout would have produced a different result. Indeed, it certainly seems that people less likely to vote favoured abolition and perhaps if forced to, would have voted for the motion.

However, it is far from certain that an increase in turnout would have affected the result in this way. Extensive research indicates that an elector’s propensity to vote is driven by the extent to which they are informed about the underlying issue. All things being equal, informed individuals are more likely to vote than uninformed ones. Thus, apart from the relative importance of the issue, increases in turnout are driven by more extensive information campaigns which would perhaps only serve to increase turnout without really having any effect on the relative proportions in the result. In this way, given what we know about the composition of non-voters compared to voters, it would appear that if mobilized to vote, a significant proportion of non-voters would be likely to change their allegiance from a ‘Yes’ (as indicated in polls) to a ‘No’ vote (as indicated by the actual election).

20 thoughts on “How did the polls get it so wrong in the Seanad referendum?

  1. I agree with most of this. Certainly the odds offered by bookies were far too generous, given the polls and their flaws. There’s a self-serving piece in the Irish Times suggesting something happened in the last week, but there was no Seán Gallagher moment. It’s a bit like Nice I and Nice II; the people who were passionate voted No in both, but in Nice II the less passionate also voted and it passed.

    I’m not sure it follows then that if people were more informed, and then motivated to vote they would have switched sides. 1/3 of the country will vote no to anything, and given we see majorities in Donegal and inner city Dublin, I suspect they were motivated to vote more by a willingness to kick the government, and less by an allegiance to the Seanad. If the referendum had been held on the day of the local/ EP elections, the people who don’t feel attached to the Seanad, but don’t care that much about the issue may have voted, and voted for abolition. The campaigns didn’t give very strong reasons to vote Yes, whereas the effective Democracy Matters campaign successfully (if ironically) linked saving the Seanad to saving Irish democracy.

  2. Apart from a lower turn-out in “stand-alone” referendums, do the polling companies and/or the political scientists use different models to analyse the outcomes compared to general elections or even local/European elections?
    Given the big disparity between what was predicted and what happened, can the analytical frameworks be refined? If not, fine. But if so, what else is needed? How much would it cost? Who would pay?

    • They need larger sample sizes. If you’ve to throw away more of your sample (because they are not going to vote), then you need a larger sample size than 1000 people. While the +/- 3% thing is somewhat of a nonsense, that same nonsense is arguably larger where you have to throw away almost 60% of the sample.

      • Given that 35% voted against the Court of Appeal when there was no campaign against it, I think we can fairly say that 1/3 of voters were kicking the government. And when the Seanad referendum fell by a margin of 3% then we might say that the ‘kick the government’ factor was indeed decisive.

  3. Interesting post. I would be particularly interested in reading the research on propensity to vote being linked to a voter feeling informed. Do you have a link?

    • Hey Bart,

      Its not about the voter ”feeling informed”, but rather that voters act differently if they are informed. A classic is Larry Bartel’s 1996 ‘Uninformed Votes: Information Effects in Presidential Elections’, it broadly corroborates with King and Gelman’s ‘Reinforcement’ effects of campaigns.

  4. @Eoin,

    A lot of people were confused by the Court of Appeal proposal and, accordingly, voted against. I attended a public meeting held locally on the Monday before the vote and a barrister gave a useful and informative presentation of the issues involved, which motivated me to to vote in favour of establishing the Court of Appeal because all things being equal, the proposal may not be the perfect solution but the burgeoning problem is so serious that a ‘yes’ vote seemed the most reasonable response.

    • Paul Anthony McDermott when interviewed about the establishment of this extra court said it absolutely did nothing to fix a broken system. So more money wasted as we get yet another Irish solution to an Irish problem. As a matter of fact, in his opinion, it made things worse, he said that he could not make sense out of some of the florid and verbose legal jargon in the legislation which ran sentence after sentence without any punctuation, resembling those planning applications posted up down at the bottom of O’Connell Street made deliberately unintelligible. So what are the rest of us non legal experts to make of this? If some one of his experience and stature is perplexed? Obviously he voted “no”.

  5. On the Seanad referendum, it’s of interest that the higher the turnout in indivdual constituencies, the higher the trend towards a ‘no’ vote e.g. my own constituency area, Dublin West. Of 35 or so referendums since the Constitution was itself ratified in 1937 (on a conspicuously low turnout!) 11 have been lost and 24 passed. Of that 11, a couple ( EU referendums on Nice and Lisbon) cancel themselves out. Three others were concerned with the introduction of FPTP voting system. Then there was the Oireacthas Committees proposal about which a great deal of concern was expressed in regard to its wording and conferring unwarranted powers on the Government by default. And now the Seanad. The turnout for most referendums, where there is not an additional election on the same day or where there is not some hugely contentious social issue at stake, such as divorce or abortion, has traditionally been low. But it is also apparent that Irish people jealously guard their powers under the Constitution from measures that are perceived as being solely in the interest of political parties and that ALL referendums seeking to dilute those powers over the decades have been defeated. In this referendum, too, since the ‘far left’ were all on the side of abolition and since they tend to make up a significant proportion of ‘no’ voters to EU referendums etc., the conclusion that the ‘no’ vote on this occassion comprised serial no voters may not be warranted.

    • “Three others were concerned with the introduction of FPTP voting system.”
      Are you sure about this? I thought there were only two on the voting system ie in 1959 and 1968.
      Perhaps you anticipate another?
      Not that it matters to your main point about the dilution of powers, within the Republic. Note that referenda involving the transfer of powers to the EU have been passed eg. trade, monetary policy.

      • Of course, you’re right! There were only two efforts to change the voting system to FPTP, but there was a 1968 referendum on constituency boundaries which was rejected by the people. Sorry for the confusion.

  6. With respect, the polls did not get it wrong, the people (democracy) got it right.
    I often recall the 1992 polls in Britain predicting the fall of the new Major government and a Kinnock led Labour majority, this did not happen.
    Equally, early polls here in 2007 gave victory to FG/ILP, again, wholly wrong.
    Finally, unfortunate to see the maintenance of the Republic of Ireland status quo with not only lack of representation for six county citizens but it seems they have fallen off the nation on the map used by Kevin.

    • @Mark
      “with not only lack of representation for six county citizens ”

      I am puzzled by the continuous drive to give citizens who do not reside here a vote and similarly, no equal drive to give other citizens who do live here a vote. There may be good and valid reasons for what seems to be an unquestioned form of discrimination.

      I suggest that to maintain social cohesion and legitimacy of government, it seems sensible that those with the right to vote in elections in this Republic should be subject to the outcome of that vote, in terms of laws passed by the Oireachtas and other measures enacted by the Governments and local authorities which govern after elections.
      On this basis, why should
      1) Some long time residents, who are not Irish citizens, have the right to vote in elections here –
      2) While other long time residents in this state, but who are also not citizens of the Republic, do not have the right to vote in elections here –
      when they are subject to the outcomes of these elections?

      On the same basis, why should
      3) Irish citizens who no longer reside here (eg. emigrants) be granted the right to vote in any election here
      4) Irish citizens who have never lived here, be granted the right to vote –
      or even some form of representation in our elected assemblies when they are not subject to the outcomes?

      Have i missed something?

      • Hello again Donal, you might have missed one major point, I do reside ‘here’, here too I pay my taxes, directly to PAYE in Dublin city center yet I have little say in how my taxes are spent.
        As to your other points raised, I am not in government so am probably the wrong person to ask, since I cannot influence whom can vote at elections.

  7. Kevin,
    “Extensive research indicates that an elector’s propensity to vote is driven by the extent to which they are informed about the underlying issue”

    I am currently writing my thesis on Irish print media’s coverage of the referendum, just wondering if you could point me in the direction of some of this research (particularly if it’s related to Ireland) as it may prove useful.


    • Hello there,

      There are a few studies confirming this effect. To pick one, Lassen (2005) published a very solid piece of work in the AJPS titled ‘The effect of information on voter turnout: Evidence from a natural experiment’ (Lassen, 2005). It is available at:

      Click to access lassen.pdf

      He confirms ‘A sizeable and statistically significant causal effect of being informed on the propensity to vote.’ See also: Coupé and Noury, 2004, p. 261; Larcinese, 2000;Wattenberg et al., 2000;

      Feel free to get in touch


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s