The polling effect of "Hackgate"

There are three voting intention polls tonight. Their topline figures are as follows

YouGov/Sun - CON 37%, LAB 42%, LDEM 9% ICM/Guardian - CON 37%, LAB 36%, LDEM 16% Populus/Times - CON 34%, LAB 39%, LDEM 11%

ICM shows Labour down since before the phone hacking scandal broke, putting the Conservatives ahead for the first time in months. Populus show the Conservatives sharply down. What's the real situation?

First off, it's worth remembering that all polls are subject to a margin of error of plus or minus 3 points - so if Populus happen to have got a sample that's a bit light on loyal Tories, if ICM have a sample that's a bit light on loyal Labour voters, there's your difference. The methodologies used by ICM and Populus are very similar, so sample error is the most obviously explanation for the difference.

ICM continue to show much higher support for the Liberal Democrats than any other company - ComRes, Populus & MORI all tend to have them at about 11%, with YouGov a point or two lower - the reasons for the size of the difference are unclear, but this difference probably also explains why ICM has twice shown a Tory lead this year: their 2010 Lib Dem voters seem to be less likely to have switched to Labour than those in other polls.

So, has "hackgate" impacted on voting intention or not? Well, the biggest advantage of YouGov's daily polling is that we don't need to worry about random sample error to the same extent - if there is a rogue poll, it should become clear the following day. The news that Milly Dowler's phone had been hacked, which effectively brought the phone hacking scandal to a peak, broke on the 4th July. In the month leading up to that YouGov had pretty consistently shown a Labour lead of 5-7 points, with the Conservatives on 36-37% and Labour on 41-43%. In the fortnight since then YouGov has shown Labour leads of 8+ points on five occassions, and the Conservatives

down on 35% on five occassions. While one cannot rule out co-incidence, this obviously suggests a small but genuine knock to the Conservative party's support from hackgate. However, the last three YouGov polls have been back within the normal "pre-hackgate" range, suggesting it may have been a short-term effect that is already fading. We shall see.

The big increase in Lib Dem support in the ICM poll is probably illusionary - they fell by 3 points in the last ICM poll so I suspect the increase this month is mostly just a reversion to the mean. Populus's Lib Dem score a month ago was also conspicuously low. However it's possible that there has been some benefit for the Lib Dems, while their score in YouGov's daily poll today is typical, they hit 11% in a YouGov poll on Sunday for the first time in weeks. I'll reserve judgement on that one.

The other impact of "hackgate" is on perceptions of the leaders. Now, there are two ways of measuring whether an event has changed how people perceive a party or leader. You can ask people if their opinion has gone up or down, which is what Populus did in their survey. This found 39% of people thought worse of Cameron as a result of hackgate, but also found people were more likely to think worse than better of Ed Miliband. However, questions like this tend to give misleading results - people who never liked a politician to start with say it's made their view worse and vice-versa.

The better way, if you are lucky enough to be in the position to do it, is to have asked people what they think before the event happened, ask them again afterwards, and see what the difference is. ICM, YouGov and ComRes have all been in a position to do this, and have found the same pattern - David Cameron's ratings are down, but only marginally, Ed Miliband's ratings are significantly up.

While YouGov found people thought

Cameron had handled the phone hacking affair badly, it hasn't had much affect upon his approval ratings. YouGov have Cameron's approval rating at minus 12, compared to minus 10 before hackgate. ICM have his approval rating at minus 5, unchanged from last month. ComRes have 33% thinking he is a good PM, down from 37% a month ago. In contrast, there clearly has been a positive effect on how people view Ed Miliband. His approval rating from YouGov has risen from minus 34 to minus 21, his rating from ICM has risen from minus 21 to minus 16, with ComRes 27% think Miliband is turning out to be a good leader, up from 18%. Whether these positive ratings remain once the political agenda moves on is, of course, something else that remains to be seen.

So in conclusion - from the polling so far hackgate appears to have had only a small negative effect on the Conservatives in the polls, and one which may already be fading. While people think Cameron has handled it badly, it has had only a minor effect on his broader approval ratings. Ed Miliband meanwhile has seen a significant boost in how the public see him, though it remains to be seen if it lasts.

The other poll tonight is from ComRes for ITV news. The most interesting question in there was whether people agreed or disagreed that James Murdoch should resign - 65% think he should, 9% disagree.