Polls and Predictions

Tweet I'm always wary about believing the results of opinion polls, even when they look like good news, such as yesterday's. I find polls most useful for the trends they reveal over a period rather than for any specific predictions made on the back of s single poll. All parties commission polls from time to time, and whatever the publicity given to a 'good' result, it's the trends which all of us are really looking at.

Polls taken at an all-Wales level are simply not accurate enough to give a valid prediction at constituency level, because of the size of the sample involved. And that's particularly true in Wales where voting patterns are very different across the country.

Some have rubbished yesterday's poll because it was commissioned by Plaid. Superficially, that sounds like an easy way out; but provided a poll is carried out professionally with unbiased questions by a reputable polling company, and without those being polled being aware of the identity of the paymaster, then there is no reason why polls commissioned by any party should be any less – or any more – reliable than polls commissioned by, say, the BBC.

The real concern of some the other parties is likely to be that polls which they have commissioned are showing them the same trends.

So leaving aside the suggested results at constituency level, what the poll seems to me to be highlighting is that opinion in Wales is not moving in the same direction as it is in England. The Labour vote is holding up much better in Wales, and the Tories are making significantly less progress here than across the border. And the trend towards increasing support for Plaid is continuing. No real surprise in any of that is there?