A few thoughts on Eastleigh

I wrote most of what I had to say about Eastleigh on Tuesday night: by elections are very unusual events and you can't tell anything about public opinion from them that you couldn't get a much better handle upon from national polling. It won't stop acres of press being written about it today! Suffice to say, the result in Eastleigh does not show the Lib Dems retaining their support in their own seats (their drop in support was completely in line with national polling), it does not necessarily show anything about patterns and extent of tactical voting (since this is a by-election and they are extremely unusual in terms of campaign intensity and having no direct impact on who actually governs), it does not necessarily show Labour face problems in the south (it's perfectly normal for a party with no hope of winning to see its support squeezed in a by-election), it does to some extent confirm growing UKIP support... but we knew about that from national polling anyway.

Equally, as I said yesterday, this doesn't mean the result is unimportant or irrelevant, quite the opposite. A victory for the Lib Dems is vital good news for Clegg and the Lib Dems will hope it helps them move on from the Rennard crisis. There was speculation prior to the by-election that losing it on top of the Rennard scandal would put Clegg's leadership in peril... now we shall never know. For the Conservatives it is much worse news in terms of the morale of the Parliamentary party. Fractious already, we now have to see if they hold it together or go into complete panic. For UKIP it is obviously terrific news, building into a narrative of growing support - expect to see another round of good publicity possibly translating into increased support in the polls.

And on the subject of the polls, the final polls by Lord Ashcroft and Populus were pretty accurate in terms of Con, Lab and Lib Dem support... but significantly underestimated UKIP support. As ever it is possible that people simply changed their minds between fieldwork and poling day, especially since momentum did appear to be with UKIP,

but as I said when the Populus poll was published I am less than convinced about the utility of reallocating dont knows in by-election polls. There is good evidence that people still saying don't know on the eve of a general election are disproportionately likely to end up backing the party they did last time, but I'm not certain we can assume that the same applies in by-elections. Certainly in this case the Populus and Ashcroft polls were both more accurate before don't knows were reallocated.