Economics and similar, for the sleep-deprived

Wednesday, May 16, 2007

 
The Resistible Rise ...

Time for an MBA post, I think, providing some of the analytical guts of my argument in this Guardian piece. But first, a digression.

Perhaps the best book ever written about the blogosphere is "The Status Seekers by Vance Packard, and the fact that it was written in the 1950s and isn't about blogs doesn't change this fact. Packard's description of social climbing, the importance of clubs and cliques and the position of dining clubs as facilities for "status lenders" can all be mapped onto the social dynamics of weblogs by anyone with a brain in his head. Give it a read, it's an excellent book (although I would point out that the excerpts available on that website aren't actually the bits I'm thinking about here, helas).

Of course, "D-Squared Digest" is definitely part of the "old money" of the blog world. I have very little traffic to direct, a technorati rank in the upper trillions and a frequency and quality of posting which is a shadow of its former days. Nevertheless, I think that this blog has a certain "tone" which remains from the days when its great grandfather was listed high on the Eschaton blogroll. Atrios never links here - good God why would he, what's to link to - but he probably remembers who I am. If you mentioned to Kevin Drum "hey I saw something good on D-Squared Digest last month", he would probably say "oh I remember that blog it used to be good about five years ago". Although this blog's true status in terms of incoming links is "genteel poverty", subsisting off the pension-like annuity of a) having invented the term "Shorter" (which the Sadly No! guys really don't need to keep crediting to me but thanks anyway) and b) once asking whether the Bush administration had any policies at all which they hadn't fucked up (thanks Brad), I like to think that my outgoing links have a tone, a sort of je ne sais quoi which the mere Google Pagerank of the arrivistes and nouveaus can never hope to replicate.

All of which is by way of a link to Dan Hardie Digest. Dan has a long history with this website as a commenter; he was the first person I ever banned from the comments section, and he is currently odds-on favourite for being the second. But nevertheless, he has written something reasonably intelligent about my Guardian article and I thought a response was merited.

To be honest, I think Dan is arguing at a 45 degree angle to most of what I meant to write. As it happens, I do think that local government is pretty unimportant and may explain why I think this some day (basically because of the role of council officers on the one side and direct local democracy in the form of single issue campaigns on the other side but let's not get into this now). However, my statement that local elections were trivial, unimportant and only attended by weirdoes was meant to be a positive, sociological description, not a normative statement. The fact that mostly weirdoes and hacks vote in local elections is not up for debate; it's something you can read off the turnout figures. The argument that if the BNP were to gain any material power at the local level they could do a lot of harm with it, however, is something of a hypothetical case for me; my whole argument is that this is not going to happen.

The article was called "The Mythical Rise of the BNP" for a reason; I thought at the time that too many people were taking it as a given that the BNP were on the rise, and thinking on from this premise in order to draw conclusions about what to do about it. It was my view at the time that this premise was flawed; that there was at least as good reason to believe that the BNP's political support was stagnating, that they already had as much power as they were likely to get (ie none) and that therefore there was no need to do anything in particular to stop them. Basically, the majority view appeared to be that the BNP were on a rising trend, and my view was that they were about to plateau (which in real terms probably means about to reverse, as political movements like this are very sensitive to loss of momentum). For the reason why, we need to go back to the dot com years. And then still further back to 1962.

It was in 1962 that Everett Rogers first published the S-Curve model of the diffusion of innovations and brought the phrase "early adopter" into the language. The S-Curve model is a beautiful piece of practical sociology, and is eminently implementable; Frank Bass was the first to fit a diffusion-type model to data and had some stunning successes in forecasting things like the size of the market for colour televisions. Before his death, me and Chris Lightfoot were swapping a few emails about what one might be able to do in fitting a Generalised Bass Model" to all sorts of social and political data, to model the development of media scares and moral panics.

It's in the same spirit that I thought it made sense to apply something like Rogers' model to the growth of the BNP. As you can see from the original Guardian piece, my underlying model of the way the BNP had grown over the last few years was to assume that there was an innovation (a new "product", if you like) called "switching political allegiance from the Conservative Party to the BNP". Following Rogers' model, this product would be adopted by innovators (2.5% of the total potential market), early adopters (13.5%), early majority (34%), late majority (34%) and laggards (16%), until everyone who was ever likely to make this switch had already done it.

By some pretty slapdash casual empiricism (but hey, at least I was using a model, which puts me a couple of steps ahead of the curve) matching the characteristics of BNP voters quoted in the newspapers to the stylised psychological profiles of the different categories of adopter, I guesstimated that as of their 2006 triumphs, the BNP were gaining their votes from the "late majority" category - they seemed to be targeting their message at people who would only be comfortable voting BNP if they thought that others were too (the impact of Margaret Hodge's catastrophic mistake in Barking was really influential on me here). As you can see from the curve, once you are appealing to this bit of the market, you are well past the inflection point, and the rate of growth in your support is just about to slow down as sharply as it had recently been accelerating (note that the "S-Curve" in the Wikipedia article would be describing the total BNP vote in this context, while the Gaussian curve on the "valuebasedmanagement" site is describing the number of "new" BNP voters. On this basis, I decided to take the punt that I was right on two assumptions:

a) that the source of the new BNP voters was not defecting Labour voters but working class Tories
b) that the inflection point had been reached and the flow of new supporters from this pool was about to slow down.

I've explained b) above; a) was more of a gut feel than anything else. People who vote in local elections are politically involved. Politically involved Labour supporters aren't going to go over to the BNP. The low-involvement, class-based component of the Labour Party might be more susceptible, but they don't turn out for local elections, and the BNP hadn't done anything like as well in the general election. The fact that BNP success was a low-turnout phenomenon made me sceptical that it was coming from the Labour Party. The new BNP voters didn't feel like mainstream citizens driven to the extremes by Labour's terrible record (and actually Hazel Blears is correct to say that on a lot of doorstep local issues Labour really doesn't have such a terrible record; the John Cruddas case that the party has deserted its working class base is a lot stronger on assertion than fact). They looked, to be frank, like weirdoes. I also noticed that a lot more people said they were "thinking of voting BNP", when specifically pressed to give an answer by journalists, than said that they "would vote BNP". Normally this would make you think of the known "Tory shame bias" in opinion polls, but the BNP really wasn't registering in actual polls, so I suspected that the journalists were getting the answer they'd solicited and any shame factor would be outweighed by the lack of depth of the BNP's support.

(Parenthetically, I'll note that the occasionally made claim that the Communist electorate in France defected directly to the FN is something of an urban myth; the evidence is really not that great that this is that case. It relies on the behaviour of a few Paris arondissements which went from red to black in 1988 (during a communist meltdown) and in Robert Hue's loss of Pas-de-Calais and similar departmentes in 1995. James Shields does not find much evidence of direct switches from PC to FN, suggesting instead that Le Pen picked up a lot of first-time voters in poor neighbourhoods because of the declining role of the Communists as a tribune party)

Contra this assumption, Dan says that an implication of my thesis that BNP voters are disaffected Tories ought to be that the geographical profile of BNP support ought to be the same as that of the Conservative vote. I think that this is fallacious statistical reasoning. Here's an analogy; every shooting involves a gun, so we should expect the geographical profile of shootings in the USA to match the geographical profile of gun ownership? It doesn't. It doesn't because there's a confounding factor which is that the majority of shootings take place in cities, while the majority of guns are owned in rural areas[1].

Similarly, I think that there are probably (at least) two confounding factors which make it impossible to argue that the general pattern of Conservative support has any connection to that part of it which is susceptible to BNP influence. I think we have to realise that the majority of the Tory Party doesn't support the BNP, is not likely to ever support the BNP and abhors what they stand for. There are and have always been soft-fascists within the Tory Party, but they have never been the mainstream and they are regarded as weirdoes there too. The first of these confounding factors that comes to my mind is that fascists have an unusually (IMO, pathologically) high degree of the "authoritarian personality" (or McClosky’s conservatism, or Eysenck’s T-scale or for that matter the first principal component of Chris Lightfoot's Political Quiz, they measure the same thing). And second, class itself.

To start with the first, I'd note that it's quite clear why the majority of Tory voters vote Tory; they do so because it's in their economic interest to do so. There is no particular mystery of why someone living in Sevenoaks might vote Tory. On the other hand, someone living in Blackburn, with a manual job or in receipt of disability allowance, who votes Tory? That's quite a weird thing to do, and a common reason why some people do weird things is that they are weird people. And if you're a weird person who votes Tory, then my guess is that the BNP are going to at least want to have you in their direct mail database.

So are there no middle class weirdoes then? Well, not as many; it's been known since the 1950s that high authoritarian scores are associated with low education and poor impulse control, both of which are characteristics associated with low income (and I don't think that there are confounding factors here)[2]. So my guess is that the sociodemographics of bigots skew below the median income. But furthermore there is the second factor to consider, which is that middle class authoritarian personalities within the Tory electorate are not a particularly happy hunting ground for the BNP because they have already been scooped out by UKIP[3]. UKIP's policies don't overlap with the BNP's all that much, but if you look at Chris's survey, you can see that they are very much fishing in the same tank as regards the personalities of their supporters.

So in other words, the answer is no, I don't think my thesis of the BNP-Tory transition implies any particular geographical profile of BNP support in the way suggested. The BNP-susceptible element are part of the tail of the distribution of the working-class Tory vote (which in itself is not particularly representative of the overall Tory vote) and you don't expect relationships between the tails of the distribution like this to be visible in the broad aggregates.

I can tie my points together with a quick thought experiment. When UKIP emerged as a political force in the 1999 European election (and then improved on its vote in 2004), did the fact that around 10% of the electorate had voted for an anti-Europe party mean that Euroscepticism was now a powerful force in the land, that had to be respected and have its concerns addressed? Or was it actually a sign that the Europhobe rump of the Tory party had given up on mainstream politics, and that the case for Britain in Europe had been definitively and irreversibly won? The Tory party managed to get this one wrong and spent seven years in the wilderness as a result. The answer is the second; the 1999 and 2004 showings were signs of weakness in Euroscepticism, not strength. UKIP exists as a party because the Europhobes have given up on mainstream politics; by leaving the Tory party they implicitly acknowledge that they have no hope of achieving their aim. And furthermore, the electoral successes they have achieved have been entirely artefacts of special situations in low-turnout elections; their overall support is flat to declining. My thesis is that the "rise of the BNP" is every bit as fictional as the "rise of UKIP" was.

And of course, though neither me nor Dan knew this at the time we wrote, the evidence appears to be that the 2007 local elections were "disappointing for the BNP. At present, I think the evidence provides pretty good support for my view that the "rise of the BNP" is not "very real"; it's mythical.

[1] This is not an invitation for anyone to have a big old argument about the statistics relating to gun crime in the USA. If you don’t like this, then pretend I said “Freedonia” instead of the US and treat it as a constructed example of what a confounding factor is.
[2] I'm sure that if you were in the mood, you could twist "bigots tend to be poorly educated and therefore lower down the income scale" to "Davies says the working class are stupid and bigots!", but I think it's pretty clear that this isn't what I said. "Most Xs are Ys" does not imply "Most Ys are Xs", although it does mean that if you're shopping for Xs, you'll do better going to Y than to Z. (Chris Dillow has a good tabular representation of the Bayesian calculation involved, though note that while he's using "60% of the population are stupid" as his example, I'm using "10% of the population are bigots" and this changes the qualitative conclusions markedly.
[3] UKIP has of course had a number of problems with overlaps between its membership and the BNP, but I don't remember any details except that the current leadership of UKIP are quite aggressive in defending their reputation.
31 comments this item posted by the management 5/16/2007 10:24:00 AM