I am currently writing an encylopedia entry on Hubert Blalock. I got into statistics through his excellent textbook as an undergraduate. Reading over his work has made me think about the issues posed by the predominance of multi-variate methods, mostly linear methods, in the quantitative programmes of Sociology and Political Science. Of course such methods dominate in Econometrics which is defined by them. Radstats seldom discusses methods as opposed to engaging with data as descriptions. Anybody any views on this. Blalock, who was plainly a decent man, got on the wrong side of some US radicals not about the real politics - for example he was a fierce defender of civil rights and the implications for Universities - but rather because his kind of approach was seen as some combination of elitist and scientistic. I am trying to locate a pamphlet from the Red Feather institute on Gramscii and Blalock but that is hard to find! Rather than personalize to one man I wonder what people in general think about the methods issues in relation to social scientific understanding and any sort of radical programme.
David Byrne