Back in the summer I wrote a piece titled “Why Voting Intention Polls Matter” largely about the Brown boost in the polls but making the point about the real importance of opinion polling in politics today.

It isn’t about predicting what the next election will be, because apart from those taken immediately before an election, polls can’t do that. They aren’t really important because they measure public opinion either – they can do so, you can tell if people’s gut instinct is in favour of inheritance tax or against environmental taxes or whatever, but it takes quite a lot of analysis to get anything concrete out of it. Most newspaper polls are written with a view of getting a good news story out of the results, rather than actually aiding understanding. You’d have to look at questions asked various different ways, in different polls, looking at issues from different angles, before you really got a good understanding of what the public think about an issue.

Polls are important because they set the political weather. They are the means by which we know if a party is doing well or badly, if a leader is on the up or on the down. In that sense, they have overwhelming power to set the agenda, to decide if interviewers begin an interview by asking why a party is doing so badly, to decide whether media reports talk about a party leader fighting back or enjoying a honeymoon, to determine whether a party’s policies are reported as something that might actually happen in the future, or just a doomed suggestion.

Gordon Brown would always have pondered the possiblity of a snap election, but the only thing that turned it into something that was actively considered, the only reason why some of the young Turks around him were pushing for an early election was that the polls told him Labour had a double figure lead. If there hadn’t been polls commissioned that week, or if they had shown only a narrow Labour lead, would an early election ever have become the likelihood it briefly seemed?

The only reason Gordon Brown eventually decided not to call an election, whatever he may say, is surely that opinion polls told him he might well lose it. Again, had there been no opinion polls that week, or had they showed Labour still with a double point lead, isn’t it likely that we’d be in the middle of an election campaign now?

And Menzies Campell, it seems clear that the reason he had to step down as Liberal Democrat leader is at least partially to do with the recent polls showing the party down at 11% or 12%. If the polls still showed them at 18%, or if there were no opinion polls to point out how low they had sunk, wouldn’t he still be there? That is the power of the polls.

Are the polls too important, too influential? What would the alternate be, people will always want to try and measure public opinion. Were polls banned tomorrow people would look at local council by-election results (and many already do), or Parliamentary by-elections, or silly voodoo polls on the telly or whatever. You would just get less accurate measurements of public opinion. And, at the end of the day, polls are but a way of measuring public opinion – the best way we have. What encouraged an election, cancelled an election and removed Ming Campbell was public opinion.

CrosbyTextorPepper (the company of the Conservatives’ 2005 election guru Lynton Crosby) have published the results of a poll of the 112 Conservative target seats they need to win to gain an overall majority (I believe they used my notional figures, so the list of seats polled is here. The poll was conducted between the 27th September and the 7th October, so straddling the Conservative party conference.

In 2005 these seats voted LAB 36%, CON 36%, LDEM 21% (that alone seems strange, how can the Conservatives be equal with Labour, yet have lost all these seats? It;’s because tactical voting will mean the non-Conservative vote is concentrated around the party that can best beat the Conservatives in that seat.) The figures now are LAB 40%(+4), CON 35%(-1), LDEM 16%(-5) – so the Conservatives wouldn’t have gained any seats from Labour at all.

However, as I said earlier, this poll straddled the Conservative conference, and the majority of it was done when Labour were enjoying huge great 10 and 11 point leads in the national polls. A swing towards Labour should come as no surprise. The figures also include breaks for those respondents interviewed before the Conservative party conference, and those interviewed after it had begun. Pre Conservative party conference, when Labour had double digit leads nationwide, the figures in the marginals were LAB 42%(+6), CON 34%(-2), LDEM 17%(-4) – so the swing in the marginals was pretty comparable to the swing nationwide. In the responses gathered during and after the Conservative conference the figures were LAB 39%(+3), CON 38%(+2), LDEM 15%(-6) – almost no Con-Lab swing at all since the last election.

It’s important to note that the number of people interviewed after the Conservative party conference was only 443, so there will be a large margin of error on these figures. However, for what it’s worth, had there been an election and these figures been repeated at the ballot box, Gordon Brown would have easily retained his majority.

In terms of methodology, it’s not quite clear how the poll was done – the long time period and the fact that data was stratified by postcode implies that it was done using face-to-face quota sampling. It was weighted by demographics of the seats concerned and used only “likely voters”. Interestingly, it also prompted respondents with the name of the candidates for each party in their seat, something that MORI often do in the last couple of polls before an election but which is normally only possible in face-to-face polls. My guess is that at this stage it probably favours the incumbent, since some people will know that X is their local MP, but (sorry to PPCs reading this, but it’s true) only hardcore anoraks will know that Y is their Conservative candidate. That said, that is accurate, the same largely applies come the actual election.

We don’t know what, if any, political weighting was applied, or how likely voters were identified, so it is hard to compare the figures to other pollsters – it shows Labour performing better in marginals than ICM’s poll did, but is that because of different methodology in terms of weighting, is it because it covers far more marginals (the Conservatives could be doing well in the bottom 50 that ICM polled, but much less well in more the distant prosects that CrosbyTextor included, or because of the different timescale or just because of that big margin of error. We don’t know.

To an extent, now there isn’t an election for at least two years it doesn’t matter. More interesting, given the fact that it was conducted over the period of the Conservative conference, is the opportunity to see how perceptions of the Conservatives actually changed during that period. We know they went up in the polls, but this is a chance to see why.

On page 11 of the results is a list of attributes people said applied to David Cameron and the Conservatives, ordered by how much they changed during the party conference. There aren’t many surprises there, but it does sum up what changed – the Conservatives went up nearly everywhere, but gains were smaller on things like the environment and immigration and the economy (still Labour’s strong card). The really noticable leaps were in the proportion of people who thought the Conservatives had strong leadership, had a strong and united team and had well thought out policies – these all nearly doubled. The largest increase of all was the proportion of people who thought they were performing well and communicating clearly, doubling from 18% to 36%.

The Conservative recovery has been mostly put down to inheritance tax, a policy which polls have universally said is popular and which Labour have rapidly sought to neutralise. The proportion of people who thought the Conservatives would cut taxes did rise, but not quite so dramatically – up from 38% to 49%. A fair old chunk of people thought that the Conservatives would cut taxes anyway, its not the tax cutting per se that changed, but that amongst other things hte inheritance tax proposals represented actual concrete policies, put across clearly. What changed was that the Tories managed to communicate a clear message, look capable and united and put forward coherent policies.

A new Populus poll for the Times has topline voting intentions of CON 38%(+2), LAB 40%(+1), LDEM 12%(-3). The poll was conducted between Friday and Sunday, and the vast majority of the fieldwork was conducted before Gordon Brown announced there would not be an early election, so unfortunately it doesn’t shed any light on how people have reacted to the non-election.

Labour are just ahead in this poll, but like YouGov’s at the weekend it shows both the main parties up in the high 30s and low 40s, with the Lib Dems way down. We’ve come to expect lower levels of Liberal Democrat support in YouGov polls, figures this low in one of the phone pollsters are something new.

There is also a shift in the overwhelming advantage Labour had on economic competence in a crisis last month – at the height of the Northern Rock crisis 56% of people said they would trust Gordon Brown and Alistair Darling rather than David Cameron and George Osborne come economic troubles, against only 18% the other way round. The figures have now shifted to 43% preferring Brown/Darling and 28% prefering Cameron/Osborne. Still a big gap, but it’s also a big shift. For the really interesting figures though we’ll have to wait for some polls conducted after the non-election announcement.