Thursday, 19 January 2017

Statistical Power

There is a quite interesting article in today's Guardian by Will Davies about, How Statistics Lost their Power. No it is nothing to do with the probability of rejecting the null when the alternative is true, but a rather fanciful discussion of the alleged decline of belief by mass publics in quantitative descriptions of their societies. 

To my mind he mixes a lot of things up in the first section where the story of political arithmetic, trade indicators and the beginnings of the sample survey are all jumbled together. By the way he seems to bring the latter forward  to the 1920s. The big innovations actually happened about 20 years earlier and were associated with people like the Norwegian Kier, but then again being a proponent of elitist quantification I guess that is just the sort of petty minded concern with accuracy that you would expect from me.

What he neglects to say anything about is how social scientists have themselves contributed to a generalized distrust not just in the use of numbers to describe the social world but even in their  very calculation. Things were already bad in the 1970s, so bad that a prominent British  Althusserian   Marxist found it necessary to write a little book defending the use of official statistics! Well, if you were a Marxist you couldn't really admit it was all a social construction, after all the Master himself used the Blue Books pretty uncritically.

But after that it was pretty much downhill following the rise of autoethnography, poetry writing, epistemic communities etc.

The brave Marxist, after writing an auto-critique, took himself off to Australia.

Monday, 16 January 2017

Epistemic Communities

Social media directed me to this paper put out by the NCRM. I  know that there are some members of our profession who are incomprehensible  and pride themselves on it, but the abstract seems to grant some kind of legitimacy  to incomprehensibility. How else should one interpret its advice about "validity within epistemic communities"?

If its now acceptable to admit that there are different criteria (standards?) of validity depending on which tribe you belong to isn't it time to  admit  that in some of the social sciences pretty much anything goes (as long as as you can get enough fellow-travellers on board) and consider seriously the implications of this for journal refereeing, REF panels etc. 

Don't get me wrong, I don't want to ban the social poets or wish them any harm. I just want them to have their own epistemic community, their own sources of funding and their own journals far away from the rest of us where they can adopt whatever criteria of validity they find pleasing without interfering with those of us that want to do social science. Then we'll see how long they can last when they have to stand on their own two feet.

Strangely enough my recreational reading at the moment is Stanislaw Lem's Solaris which is all about mutual incomprehensibility. Synchronicity or what?

Wednesday, 14 December 2016

Perception Gap

Trivia. I've been wasting time taking the Guardian's latest perception gap quiz, you know, the one where you're asked some  questions about the population of your country and you get to see whether you are better informed than the average citizen. So in 8 out of 11 questions I did better, usually much better, than the UK average and was pretty much on the money. In 3 I was a bit off target and slightly worse than the UK average. The summary conclusion from the Guardian's algorithm was:  

I don't know the UK as well as people in the UK :(

Uh? Given my pattern of results this is a bit counter intuitive. I'd love to know how the aggregation worked. Hey, who knows? Maybe a life-time of doing quantitative social science means I don't know shit about my own country. I'd be the last to dismiss the possibility. Or maybe this is just the usual Grauniad bollocks. 

Here's my totally speculative guess. Somebody has implicitly given more weight to questions where the responses are scaled in millions of people than to questions where the responses are scaled in terms of percentages.

Friday, 9 December 2016

Greg Lake RIP

So. Farewell then
Gregory Stuart Lake.

Yes.
You began with frippery,
Stood on a Persian rug
And believed in Father Christmas.

Keith's mum said it
Was pretentious.
Nice.

We were invited
To the show
That never ends.

But now it has.
Ended.

R. W. Emerson 12½




Wednesday, 7 December 2016

May: Si tu dois partir

It suddenly struck me that if the EU's Brexit negotiators have need of a motivating anthem they could do worse than this weird franglais version of an original penned by an anglophone Nobel laureate. Listen out for the broken milk bottle (a pint of course) round about 1:57.

Tuesday, 29 November 2016

Imaginary Friends

It's a peccadillo but  I have to confess that I sometimes listen to old episodes of Desert Island Discs while I'm doing insufferably boring things like cleaning the bathroom. So at the weekend I was listening to Alison Lurie talking to Roy Plomley. 

What I've read of Lurie I  liked, though she came over on the show as uptight and humourless. No matter, it's the kind of setting  that doesn't always bring out the best in people. What surprised me though is what she had to say about Imaginary Friends

Plomley asked her directly whether the novel was based on anything in particular, or words to that effect.  Yes, she answered, it was influenced by Hawthorne's The Blithedale Romance and by Henry James' The Bostonians. OK, I can see the former and I'll take the latter on trust, but I always thought the most important influence in the everyday sense of the word was Festinger, Riecken and Schachter's When Prophecy Fails.  I mean, it's virtually a fictional retelling of the same events.

At least that's what I always used to tell my students when I was trying to fake some knowledge of ethnographic fieldwork. Maybe I was wrong or maybe she was thinking about influence in some deeper literary sense than my literal minded interpretation. Still, very odd.

Wednesday, 23 November 2016

Wednesday morning magic

Two pieces of musical magic for Wednesday morning. First up Chet Baker's Almost Blue. Chet's  doing a Miles pastiche but man can he make that horn sing. And then there is Daniel Kahn's Yiddish version of Leonard Cohen's Hallelujah. Personally I think the lyrics are better than the original. Kahn also does a nice line in Yiddish worker's songs, but I admit that might be a bit of an esoteric taste. L'chaim.