December 21, 2008

When will psychologists ever learn?

Artifactual relationships, absent sampling and invalid measuring instruments still abound in research into the psychology of politics

Rather to my surprise I recently read in the popular press a rather good article which points out that "phobias" and prejudice are very different. It was written by a young psychology professor named Nicholas Haslam at the University of Melbourne. I have myself for some time been protesting the misapplication of the word "phobia" to just about anything that Leftists disagree with: Homophobia, Islamophobia, xenophobia, etc. As the article will cease to be available on the newspaper site after a while, I have reposted it on POLITICAL CORRECTNESS WATCH today.

To keep his article within the bounds of political correctness, however, Haslam also had to say that "Prejudice flourishes among people who are cold, callous, inflexible, closed-minded and conventional". So you are still a pretty bad egg if you distrust Muslims or regard homosexuality as wrong or unhealthy.

What he said there is a conventional belief among psychologists but the evidence for its truth is very weak. In the 60 years that psychologists have been subjecting such theories to experimental test, just about the only proof for such theories that they have found has been derived from handing out to their students a bunch of questionnaires and seeing if the students who didn't like (say) blacks also expressed views that psychologists regard as close-minded etc. And from what their students say, psychologists generalize to all mankind.

From almost any point of view that is a ludicrous procedure. Not only do they base their research on a non-sample -- meaning that no generalizations can be drawn from it anyhow -- but college students are even a group of people who are KNOWN to be unrepresentative of the general population in all sorts of ways. And even if students were representative, relying on what they say would be most incautious. Students are very good at giving their professors the answers that they think their professors want. So many of the answers given will not be what the students really think.

So for all of those reasons, I was only slightly surprised when, in one of the earliest pieces of research I ever did, I found a correlation of .808 (a very high correlation) between two variables among students but when I repeated the survey on a more representative population, the correlation dropped to around .10, which is negligible. For the rest of my research career. I did almost all my research on proper samples of the general population and almost always ended up getting very different results from my student-using colleagues.

So I was curious to see if Haslam was just mouthing conventional and unsubstantiated platitudes or if he really had some basis for his generalizations. He replied that he was relying on a big review article on the subject by Sibley and Duckitt in Personality and Social Psychology Review titled "Personality and Prejudice: A Meta-Analysis and Theoretical Review". In a very restrained academic way it ploughs the old furrow that racially prejudiced people are sick in the head and anti-racists are just wonderful lovely people. Negative racial views are very common (if rarely acknowledged publicly these days) so there must be a lot of sick people around.

I have debated in the journal literature with Duckitt before so expected him to be less naive and assumption-prone than are most writers in the field -- and so it was. He shows a rare and commendable awareness that alleged correlations between attitudes and personality can arise because the alleged measures of personality are in fact measures of attitudes, for instance. He does not take that awareness as far as he might, however. He seems, for instance, to take correlations between racism and social dominance quite seriously despite the fact that the social dominance questionnaire contains such items as "Inferior groups should stay in their place", "Superior groups should dominate inferior groups" and "Some groups of people are just more worthy than others". Races are of course groups so is it any surprise that such statements correlate with other expressions of racism? All Duckitt has shown is that some expressions of racism correlate with one-another. He has shown nothing about personality at all.

So the finding of a relationship between social dominance and racism is what is called in science a "methodological artifact" -- generally a source of shame among serious scientists, but something that has long been common in this research field. Duckitt himself points out some other examples of it. Psychological research is in general still a profoundly amateur enterprise.

I might mention that the folly that I have just pointed out is not really the fault of Duckitt. His article is simply a summary of what other researchers have found and none of them seemed to see any problem with their measure of social dominance either. So it is psychologists as a whole that my criticism principally applies to. I never cease to be staggered by how blind psychology academics can be. They must never look at the questions they ask people.

Duckitt DOES show an awareness of the sampling problem I have mentioned but does not seem to take it seriously. He claims he has some real samples in his data but he does not identify them and combines them with the student data. There is no repetition of all his analyses on student and non-student data. So the generalizability of his findings is simply unknown.

But the problems with the Sibley & Duckitt article do not end there. Duckitt says very little about the measures of racism that he uses. He concedes that they were only poorly comparable and that some were more narrowly focused than others but he seems to take no account of that in his major analyses. Yet this is a vital point. In my research, I repeatedly found some shared variance betweeen attitudes to different racial groups but not much (about 20% on average). In other words, there were many people who didn't like (say) blacks but who did respect (say) Jews. So in most of the general population, there is essentially no such thing as racism. If there were, knowing a person's attitude to one minority would tell you all you need to know about that person's view of all minorities. But it is not so. Undoubtedly, there are some individuals who dislike all outgroups but that is not generally so. So the concept of racism is close to being an irrelevant concept. The concept it embodies is misleading. Many white people may be wary of blacks but have no firm views on race in general. So Duckitt makes a basic assumption that has very little correspondence with reality. There is ample room for attitudes to different races to have different correlates but Duckitt treats them as all the same. He has thoroughly scrambled Humpty Dumpty.

What I have just challenged is what psychologists call the "validity" of the racism measures. Do they index what they purport to measure? And the pervasive Leftist orientation among psychologists seems to make them very poor at composing valid questionnaires to measure racism, conservatism. authoritarianism etc. If you want to find out what people really believe you have to present them with statements that express that and not statements that are utterly loony. But psychologists tend to think that anything to do with conservatism etc is loony so it is common for them to compose questionnaires that contain way-out statements rather than normal expressions of conservatism etc.

And that shows on the rare occasions when the validity of such a questionnaire becomes testable. Do answers to a psychology questionnaire about conservatism predict a conservative vote in national elections for instance? From the McClosky and Adorno questionnaires to the Altemeyer questionnaire, they dont, or do so very weakly. So some of the measures of conservatism most frequently used by psychologists are demonstrably not valid.

And that IS the fault of the person who devised the questionnaire. I am more a libertarian than a conservative but I do have some conservative sympathies and the questionaire measures of conservatism that I compose correlate with general population vote up to the level of .50, which is not high in any absolute sense but which is very high by the standard of what is normally found in psychological research. It is certainly much higher than what is found in general population samples with the McClosky, Adorno and Altemeyer measures that other psychologists use. And the difference is that my conservatism questionnaires contain examples of what conservatives really say rather than what psychologists think they say. And it is amazing how profoundly wrong the conventional psychological conception of conservatism can be. See here. When psychologists research conservatism, they usually research a caricature of it.

And what is true of conservatism measures used by psychologists is also true of measures of prejudice. And so validation of such measures against real-life behaviour is rarely attempted. Does a "racist" person according to psychologists actually tend to vote for political candidates who are critical of affirmative action or uncontrolled Hispanic immigration, for instance? Psychologists normally seem game to test that only among their students.

So you see why I gave up psychological research around 1990. I felt that I was in a dialogue with mere game-players rather than serious scientists. What they say reflects their prejudices, not the results of any serious research.

The games psychologists play can be dangerous however. The sort of utterance that I quoted from Haslam above has the tendency to dehumanize those it describes and that view of "racists" and others has certainly passed from psychologists into the popular culture. Note here where a NY film critic quite literally questions the humanity of "racists". When one notes how many people -- even critics of illegal immigration -- are routinely denounced by Leftists as "racists", we see that such dehumanization could hit a lot of people. How ironic that the Leftist psychologists who would denounce the dehumanization practiced by the likes of Hitler go on to do a pretty neat job of dehumanization themselves. And just as Hitler based his dehumanizations on fake science, so do modern-day academic psychologists.

Substantiation for the various points I have made above about research findings can be found here.



Go to John Ray's Main academic menu
Go to Menu of longer writings
Go to John Ray's basic home page
Go to John Ray's pictorial Home Page
Go to Selected pictures from John Ray's blogs