Political polling may not reflect the whole picture

A clear shift away from Fianna Fáil emerges from a careful reading of opinion polls, despite recent small gains, writes GARRET…

A clear shift away from Fianna Fáil emerges from a careful reading of opinion polls, despite recent small gains, writes GARRET FITZGERALD

TWO WEEKS ago, Noel Whelan in his column sought to clarify some aspects of political polling in the Irish context. I entirely agree with his comments in that article, but would like to develop some aspects of this matter further.

Political polling has become widely accepted in Ireland. This has been helped by the fact that Irish polling companies apply guidelines laid down by the Marketing Society of Ireland and Esomar, the world organisation that seeks to enable better research into markets, consumers and societies. Moreover the samples they employ for political polling have been quite large – normally about 1,000. (I have the impression that in neighbouring Britain smaller samples of about 600 are often used, which would naturally affect the reliability of their results.)

Different sampling methods have been used by different polling companies. Some have used random sampling of the population and others have employed quota sampling – adjusted to ensure that the sample they use reflects accurately a number of key demographic features of our population. Some polls are undertaken by calling to each sampled home, and others by telephone.

READ MORE

In the case of MRBI, for many years past the crude data from sampling has been adjusted to allow for a historical pattern of overstatement of support for Fianna Fáil as against Fine Gael and Labour. The reasons for that overstatement have not been clear, but may reflect the past salience of that party due to its near monopoly of government leadership over more than two decades. Experience over many recent elections has until recently validated these MRBI adjustments, which have consistently yielded poll results very close to actual election outcomes. But, because of the recent radical change in attitudes towards Fianna Fáil, this long-established adjustment is probably no longer appropriate.

There has been no IMS poll during the past year, but during the past year the data from each of the other two polling companies – Ipsos/MRBI (who poll for The Irish Times) and Red C (who poll for The Sunday Business Post) – in respect of the three main parties have, with one marginal exception, fallen consistently within a range where the margin of error is 2 per cent or less. In other words for a party accorded 25 per cent support in a poll, there is a 95 per cent probability that its actual support will lie between 23 and 27 per cent. For parties with about 10 per cent support this “confidence limit” would be just one percentage point either way.

Now, in the past six months MRBI (now IPSOS/MRBI) has carried out three polls while Red C has undertaken five. The two poll systems yield somewhat different sets of figures for the three main parties – a point that I shall address in a moment. But I want first to draw attention to the very significant fact that, with a single possible exception, all of the figures in each of the two sets of poll results recorded by these two companies in the past six months have fallen within the confidence limit appropriate to their share of the vote.

Indeed, going back to the start of last year each of these two poll systems suggest that by the end of 2008 voters had decided which party they would vote for in a general election – and throughout the past 12 months there is only very limited and uncertain evidence of any change of mind on the part of the electorate.

Small shifts of a couple of percentage points in party support have no statistical significance – a fact that most of the media clearly fail to grasp, as they celebrate in dramatic headlines what have – at any rate until the most recent Red C poll – been merely minor fluctuations in party support, falling well within the confidence limits of the polling system.

However, it is possible to interpret recent data as showing Fianna Fáil to have gained marginally since late last year. The last Red C poll showed a four-point increase in Fianna Fáil’s share of the vote to 27 per cent, and the last three MRBI polls (two in September before the Lisbon referendum) showed an apparent increase in Fianna Fáil support from a quite exceptional non-adjusted figure of 21 per cent up to 24 per cent during September 2009 and to 26 per cent last month.

However, with the possible exception of that Fianna Fáil 21 per cent non-adjusted figure early in September, all the data, including the most recent figures, fall within the confidence limits for each poll series, so it is not yet clear that this recent apparent small shift in Fianna Fáil is significant. We must await further data to confirm or negate the existence of such a trend.

While both the MRBI poll, when stripped of its probably no longer appropriate adjustment factor, and the Red C poll yield almost identical data for Fianna Fáil (24 + or – 2 per cent) using different polling techniques, they differ quite sharply – and I have to say puzzlingly – in respect of the shares of the Opposition vote that they accord to each of the two main Opposition parties.

Taking all these factors into account I would judge that the general pattern of party support today may be something like this: Fianna Fáil 25 per cent; Fine Gael/Labour 52 per cent; Sinn Féin 9 per cent; Green Party 4 per cent; Independents 10 per cent.