Monday, February 12, 2007

One-dimensional man and political language

It is revealing that in western countries leftists intellectuals usually outnumber conservatives in education, mass media, show business and humanities, while in hard sciences, engineering and economics situation is reversed. In Soviet Union humanitarian or teaching career, as a rule, was chosen by those who felt themselves not smart enough to pass math exams.

Recently Psychology Today published a study comparing psychological features of conservatives and liberals and asserting that conservatives are neurotics possessed by irrational fear and reacting to it by conformism and clinging to authoritarian rule. Dan Beste debunked this analysis and proposed his own view on the essence of conservatism and liberalism in different historical circumstances. He found that the meaning of these terms depends on concrete historical reality and can change dramatically as this reality changes, even can became opposite to what it designated originally.

What Dan Beste really did looks alike what authors of scandalous Psychology Today “study” attempted: make a “saving translation” from the language of political philosophy to the language of personality psychology. But he did it much more thoughtfully and honestly. And so he found that personality characteristics can’t be expressed by a single value on any possible scale, but need many dimensions ­– the more the better. Personality is multi-dimensional “variable”, actually infinite dimensional. That is why serious psychologists, like Carl Jung or Viktor Frankl, prefer multi-dimensional approach to typology of personality. There are, alas, several formidable obstacles in using any of such classifications: first, axes are somewhat arbitrary, they are not really orthogonal, that is characteristics are not independent and usually correlate or anti-correlate, and, second, these correlations are not inherent, they are culturally dependent and wildly vary from one generation to another or from country to country. And, of course, most people hardly can imagine geometry of multi-dimensional space. So these correlation are used to simplify the picture and reduce dimensionality, usually to one. Such reduction can be more or less justifiable in one moment of history and became totally inadequate in another, because old correlations fade out and new emerge. More rigorous approach would be not to specify axes beforehand, but use Pearson’s biometrical statistics: multi-dimensional scaling, factor analysis, method of principal axes, cluster-analysis, and seek real correlations by objective statistical criteria. But modern psychologists, it seems, are too mathematically illiterate for this task.

1 comment:

Buddy Larsen said...

Great post. You're way ahead of anybody on the Psychology Today staff --that's pretty obvious!