The polling of political opinion is now intensive and highly scientific in this country. A national opinion poll can be taken as an accurate measure of support levels at a moment in time within a margin of error of plus or minus 2 per cent.
This is provided it is based on a sample size of at least 1,000 electors and is conducted by one of the leading polling agencies.
Similarly, constituency’s polls if based on a sample of 500 electors and conducted by one of these leading agencies can be relied on as an accurate picture of where candidates stand in terms of first preference support at a particular time.
There are always exceptions and outliers, but the above can be taken as a general rule. In recent elections opinion polls have been very accurate in this country.
However, the conducting of the polling itself is only the first stages in how polls come to impact on the coverage of elections and on electoral races.
The second and more significant stage is the reporting of the poll data by the media. This is far from scientific.
Headlines
It is journalists and not statisticians who shape the coverage of and the commentary on polls. It is journalists and their editors who decide what findings in the polls to emphasise and even what questions are asked in the first place. It is also they who decide what headlines and graphics accompany the stories about polls.
Usually the headlines or the characterisation of the findings are consistent with the poll data itself. That is not always the case, however.
Sometimes newspapers or broadcast organisations take short cuts in analysis in order to develop more dramatic headlines.
The task of seeking to divine national political implications or constituency-by-constituency impact of any particular poll is a dangerous exercise. Nowadays when there is such a large volume of published polling the safest course is to look at national opinion polls over several months, or to group several polls during the fast-paced weeks of an election campaign. Too often, however, media rush to judgment on the basis of small shifts in a single poll.
The greatest errors made by media in reporting on polls occur in relation to constituency polling. Sometimes media seek to rely on these polls to suggest outcomes beyond the first preference vote even though the sample size decreases dramatically and becomes unreliable beyond the first count.
Another difficulty often overlooked in covering constituency polling is that asking people who they will vote for, even at this point before the election, runs the risk of overstating the support for incumbents or others who enjoy a high profile.
Manipulation
An even more worrying trend is that some media slavishly report the details of polls summarised to them by political sources which the journalists have not themselves seen or being able to analyse. This leaves reporters open to manipulation by party spin-doctors seeking to talk up their party’s chances, seeking to give their candidates credibility they don’t deserve, or seeking to manage expectations.
The recent saga over the leaking of an internal Labour Party analysis of private constituency polls touches on some of these issues. The leaking of this internal analysis last week triggered a row within the Labour parliamentary party over who was responsibility for the leak. At this week's parliamentary party the party's deputy leader, Alan Kelly, strongly and emotionally denied that he or any of his staff were responsible for the leak.
One thing which is clear, however, is that whoever leaked the analysis saw it in the Labour Party’s interest to do so. That may seem strange but so precarious is Labour’s position that some in the party believe that showing it to be competitive to hold a dozen or so of its seats will assist in countering suggestions that it is in free-fall. Recently another pundit told me that he was specifically briefed by senior Labour figures on private opinion polling precisely for this purpose.
Private polling
Interestingly, the recent pessimistic – or some might say realistic – internal Labour analysis of its private polling contrasts sharply with the bullish view of the same polls taken by
Kelly
himself last July.
When Miriam O’Callaghan on her summer chat show put it to him that Labour could be as low as 6 or 7 per cent at the next election, Kelly pointed out that he was the party’s director of elections and that it had “done a lot of constituencies polls and in every single one of them we are winning a seat”.
It was a statement that lacked credibility then and lacks even more credibility now.
At lease, however, we could assess it as having coming from Kelly himself.
Beware politicians bearing news of opinion polls.
Beware also of journalists basing stories on unseen opinion polls.
These axioms will be particularly important over the next three months.