The turnout in the June 1997 general election was 66 per cent. In other words, 34 per cent (930,000 electors) did not vote; this represented a downturn of more than 10 points since the 1981 election, which was the first in the context of this 166seat Dail.
In the final Irish Times/MRBI opinion poll on May 28th the party support figures, having been adjusted for the undecided, were Fianna Fail 42 per cent; Fine Gael 26 per cent; Labour 11 per cent; PDs 7 per cent; Green Party 4 per cent and Others 10 per cent.
Relative to the actual result, the survey overstated Fianna Fail support and understated that for Fine Gael and Others. The figures for Labour and the Green Party were within the statistical variation of the sample.
However, one crucial point was that in the survey 86 per cent of those interviewed committed themselves to voting for a specified party or candidate in the election, and since 66 per cent actually voted, it followed that 20 per cent of respondents in the survey, having stated in the interview that they would do so, did not vote.
On June 6th, election day, MRBI conducted an in-home survey of 1,400 electors for The Irish Times, with interviewing terminating at 8 p.m., to allow for transmission of the results in time for publication the following day.
The survey first preference figures, having excluded the undecided, analysed by those who had (1) already voted, and (2) those who had yet to vote and definitely intended to, relative to the actual election results, were:
The overall survey figures (column a) overstated support for Fianna Fail by five points, and understated that for Fine Gael, Labour, PDs and Others, by one, two, one and one point respectively.
The pre-voting analysis (those who had not yet voted - column b), overstated Fianna Fail support by eight points, and understated Fine Gael, Labour, PDs and Others by two, three, two and one point respectively.
The post-voting figures (those who had voted - column c) were identical to the election results in five instances, and varied by one point in the remaining two.
These figures were very accurate and were comparable with the results of the exit poll conducted by Lansdowne M.R.
In establishing the extent to which support for the various parties was at variance with the final outcome, the election day survey makes it possible to calculate the relevant weighting factors, which now form the basis for the new MRBI adjustment procedures.
The framework is factual, relevant and up to date and is structured to ensure validity in the final published figures.
Contrary to some suggestions, recent MRBI research has shown that neither direct questions nor projective techniques provide a solution, and the new methodology is certainly the most acceptable technical option in the circumstances.