The Kellner conundrum

Michael Portillo was widely ridiculed on Thursday night for suggesting that the Tories' 44% showing wasn't good enough, and that they "should be doing better" if they wanted to win the general election. At the time, I thought the rubbishing of this claim was entirely justified - after all, the Conservatives' lead of 44% to 24% looks remarkably similar to the 47%-25% lead Labour enjoyed in 1995 in the run-up to securing a record parliamentary majority of 179 in 1997. Anthony King (not my favourite psephologist, a subject I may return to another day) confirmed that a 20 point Tory lead in local elections has few historical precedents.

But now Peter Kellner of the polling company YouGov has muddied the waters by suggesting that, in real terms, Gordon Brown's 24% is somehow less bad than John Major's 25% in 1995. This is because, he claims, Labour tend to under-perform in local elections when compared to general elections, while the Liberal Democrats tend to over-perform. This has left me deeply confused. The point about the Lib Dems is undoubtedly true, but surely it's the incumbent government - whether Labour or Tory - that tends to see its vote depressed? Is there really a separate phenomenon that hurts Labour regardless of whether they find themselves in government or opposition? If I ever have a spare month or two, I might try to wade through the figures and work it out.