Sunday, June 12, 2016

The psychology of conservatives

Some psychometric observations

Powerline has up a fun story which is rather long and quote-heavy so I will not reproduce it in full.  But here is the opening paragraph:

"Hoo-wee, the New York Times will really have to extend itself to top the boner and mother-of-all-corrections at the American Journal of Political Science. This is the journal that published a finding much beloved of liberals a few years back that purported to find scientific evidence that conservatives are more likely to exhibit traits associated with psychoticism, such as authoritarianism and tough-mindedness, and that the supposed “authoritarian” personality of conservatives might even have a genetic basis (and therefore be treatable someday?). Settle in with a cup or glass of your favorite beverage, and get ready to enjoy one of the most epic academic face plants ever"

It turns out that the authors got their statistics back to front.  What they said was true of conservatives was actually true of liberals and vice versa.  The article was titled: "Correlation not Causation: The Relationship between Personality Traits and Political Ideologies"

If you look closely at the findings, however, it doesn't much matter either way.  The only correlations of some note were between politics and Eysenck's P scale -- correlations explaining around 12% of the variance. Correlations of that magnitude could be of interest but in this case, the problem actually is the P scale.  It has poor reliability and originates as a set of "leftover" items in a factor analysis yielding two main factors.  In my research I have found internal reliabilities for it of as low as .40, which is way below the normal criteria for measuring anything.  What correlations with it mean is therefore hard to say.  You would have to do an item-by-item analysis to find out what was really going on.  As a scale, however, it is of doubtful meaning and proves nothing certain.

And given the known psychometric weakness of the P scale, the authors were quite remiss not to report any reliability data for the scale in their use of it.

The authors also reported various correlations with neuroticism but such correlations are all over the place.  Sometimes conservatives are found to be slightly more neurotic and sometimes it is liberals. And mostly it is neither. Again, however, there is a large problem with the measuring instrument used.  The Eysenck N scale is one-way-worded, which can greatly distort the score on it.  When I used a scale to neuter that problem, I found (in a community sample) no correlation between neuroticism and politics.

Finally the authors report some slight correlations with social desirability responding.  What these mean is also obscure.  Social desirability scales can collapse into complete  unreliability sometimes -- but, far from giving us reliabilty data, the authors do not even tell us which of the several available scales they use.  I will not therefore attempt any interpretation of their findings

In short, both in its initial and revised form, the report is pure junk.  Hatemi and Eaves are generally good scholars but they seem to have fallen into a trap common among psychologists -- treating a measuring instrument as a black box -- without making any enquiry about what is going on inside it.

So all we have is another episode in the long tradition of doing junk research to examine the psychology of politics.  That Leftism is the politics of hate is just too unpleasant for most people to face.

Posted by John J. Ray (M.A.; Ph.D.).

No comments:


eXTReMe Tracker