The other day I was asked whether I thought that modern polling still had any weaknesses and what would we be saying about their final predictions in comparison with the outcome on the Friday after the next general election.

Are there any segments of the electorate who have a markedly different political profile from the norm who might be being over or under-represented in current polling approaches?

The obvious group is the over 55s for we know that older voters are more likely to be on the electoral register, more likely to turnout to vote and more likely to veer towards the Tories.

Late last year I linked to a fascinating academic paper by Scott Davidson, then of Loughborough University now of De Montfort, which was published in 2006 called “The Grey Battleground”. The above table is from the paper.

Davidson’s thesis is that the “baby-boomers”, those born in the 1945 to the early 1950s, are just coming through the system and because of the greater propensity to vote the over 55s could well comprise of more than 45% of those who take part in the next general election. Yet in most polls the headline polling figures are based on the voting intentions of much smaller proportions.

Thus with YouGov, the only pollster which does not weight by certainty to vote, the over 55s segment generally makes up about 36 – 37% of those in the sample on which the headline figures are calculated. With ICM, Populus and ComRes the proportions are generally in the 39-40% region.

This all contrasts with MORI, which only includes the views of those who say they are 100% certain to vote, where the 55+ group can exceed 50% of the total on which the final calculation is based – a point that in its own way supports the Davidson argument.

My headline is deliberately provocative – but if the polling numbers at the next election are out of kilter compared with the outcome then my guess is that the elderly could be a factor.