Are there better ways of predicting likelihood to vote?

A local party ward organisation that I’m familiar with was one of the first to get its records computerised and can access detailed data on individual electors going back for more than two decades. Not only can you see how they responded to canvassers on each occasion that they were contacted since the late 1980s but there’s also a record of whether they voted or not.

For it’s not often appreciated that following an election the marked register, showing who voted, is made available for inspection and smart local party organisations record this data to be keyed into their computers. The result is that they can track at an instance whether someone is a voter or not.

The pattern that I am sure that others are familiar with is that there’s one category, let’s call them the “super voters”, who always turn out in local, national and Euro elections. A second group sometimes vote for their local councillors, but not always, but they invariably turn out in general elections. A third category don’t vote in local elections but are usually there on general election day. A fourth category have a mixed record when it comes to general elections but never both with local elections.. The final group simply never vote.

My contention is that those who turn out in local elections are almost certain to vote at general elections. Past voting action is, surely, a better guide than what people say they will do in the future?

So why don’t pollsters ask whether respondents vote in local election when trying to determine “certainty to vote”. Rather than the current “can you rate on scale of 0-10 the likelihood of you vote” those who say they do vote in council elections should be put into the top category immediately. At the very minimum their responses should be given a greater weight than those who don’t turn out but say they are 100% certain to vote in the next general election.

Identifying voters is important as we have seen in number of discussions on the site in recent weeks on the way that opinion movements amongst declared voters have been less pronounced than amongst those who did not turn out in 2005. A significant part of the very big Tory figures earlier in the year were from non-voters and now the same thing is happening with Labour support.

If pollsters paid more attention to past voting actions, particularly in council elections, then I believe they could make a better assessment of certainty to vote and produce more accurate figures.

Mike Smithson