This is why blogs are better. I’ve known that John Taylor wasn’t the victim of racism for ages because I read Next Left. What are these Journalists doing printing such dross when the truth has been out there for months and months?

That’s interesting, because I was living in Cheltenham in 1992, (I voted LD), and I seem to remember a fair few people for whom his skin colour was an issue. I’m not saying that him being black lost the Tories the seat, but I think it probably cost him more than a few votes.

@1

Agreed. I’ve long been aware that darkies and dolly birds can’t be trusted thanks to Guido Fawkes. Blogs are definitely better.

GaftheHorse@2

Everybody should be able to agree that endlessly calling this a “Tory safe seat” is simply wrong, since there is an agreed definition: a seat with a majority over 10%. That description settles the question of why the seat was lost. Parties do lose safe seats very occasionally (eg Tatton 1997), the reasons are usually very clear and they don’t match the pattern of other seats. They don’t tend to then lose them at five successive elections either.

So “Taylor failed to hold the marginal seat of Cheltenham, after a row in which some local Conservatives objected to the selection of a black candidate” would be accurate. It would signal that the causality is at least open. (And clearly Taylor did face some ugly racism in the Conservative association; though by definition from a minority given that he had just won the candidate selection vote).

It is a further (and less definitive) argument to say that the evidence for the significance of race in the general election looks very weak.

It seems unlikely that large numbers of Tory voters stayed at home instead of voting for a black candidate, as turnout was up to 81%.

Did voters switch parties to prevent a black candidate winning? Clearly, one can’t definitively prove a counter-factual. What one has to do is look for evidence to estimate how a hypothetical white Tory candidate would have got on. So let’s invent a parallel universe candidate called John Taylor-White (with a similar professional background to the real John Taylor).

Those writing the false “safe seat” story must expect our hypothetical JT-W to hold the seat very comfortably (surely by more than 5%). That depends on there having been 3000+ “racist switcher” voters in Cheltenham, or as many as 1 in 10 LD voters in 1992 having that motivation. This is very implausible, given the comparative results.

Alternatively, another scenario where race was decisive at the margin might see our hypothetical candidate JT-W win by a tiny majority instead of losing by 1668.
That simply requires more than 850 people who would have voted for our hypothetical Tory candidate John Taylor-White to have in fact voted LibDem over Taylor because he was black (though only after netting off any voters who would have switched to or voted LibDem but then didn’t to support Taylor over the attacks on him). That would make 3% of the overall LD vote “racist switchers”.

The media accounts show many people intuitively think this must be exactly what happened: it is not an impossible scenario in principle. The mystery would be why that would seem to leave us with a rather better Tory performance in Cheltenham than in other seats which appear comparable, instead of a broadly similar one.

That isn’t to claim that no voters might have had a racial motivation. However, the hypothesis of a medium or strong race effect capable of being decisive must surely depend on evidence the Tories did particularly badly in Cheltenham compared to similar seats with white candidates. (The evidence of broader quantitative study of the 2001 election found that party massively trumps local candidate characteristics in UK elections, though there can be weak race effects for first time candidates which disappear for incumbent MPs: somebody else may have the link or being able to quantify the effect).

If we had seean 8% swing in Cheltenham compared to 5% in Gloucester and Bath, there would be good evidence for the proposition. When the swings in those seats were so similar, there doesn’t seem to be any good evidence. If somebody has different evidence which suggests the Cheltenham result was unusually poor, I am open to persuasion that there was a weak or medium-sized race effect.

@4: Sunder
I wasn’t disagreeing with you at all, (I’ve just reread my post and I can see that the tone of it could be misconstrued). I was genuinely interested to see that the evidence seems to point to racism not having had an effect on the election result. I’d always assumed it had due to some of the comments I heard when John Taylor was selected, people commenting that they’d no longer vote Tory etc. Cheltenham looks very nice when you see it’s Georgian streets, but some of the newer housing estates on the outskirts were pockets of poverty and it wasn’t uncommon to hear openly racist comments. Certainly this was the case in 1992, but I moved away a few years later so I can’t comment on what it’s like now.
Possibly some of the people who made these comments went and voted Tory anyway. I voted LD but would probably have voted for John Taylor had he not been standing for the Tories. He seemed like a decent enough bloke whenever I saw him being interviewed locally, (although given he’s been fiddling his expenses (allegedly) maybe I was wrong on that as well).
Thanks for the John Taylor-White thought experiment. I’m always happy to have my views changed by good solid evidence

Thanks: the response wasn’t supposed to be sharp or critical either.

Institinctively, before looking for the data, I would not myself find the idea of (net) 300-500 voters out of 60,000 being willing to vote for “John Taylor-White” (Cons) but not John Taylor (Cons) intuitively implausible, ie that somewhere around 1-2% of voters might switch on grounds of candidate race but probably not more. (Like you, I would have sympathised with Taylor in 1992 but not switched my vote to him, but there may be some offsetting votes, which might help to explain the lack of any observable effect)

What the 1992 comparative results suggest – unless I am missing something – is that any effect was probably even weaker than that, so that the 850-1000 voters needed to be decisive in this case seems way out of bounds. On the whole, I would like you expect the 1992 effects to be stronger than a 2010 effect, yet Taylor would probably lose again because of party affiliation, not race.