A test built to correlate with prejudice will correlate more than an unfiltered one.

Evidence

First of all, notice that the argument here amounts to ‘‘Less is better.” The D Scale allegedly measures general authoritarianism better because it does not correlate as well with ethnocentrism as the F Scale does. But that does not show the D Scale measures anything better. Second, the D Scale may have done a little poorer job simply because it is a more disjointed, psychometrically weaker instrument (“Worse is better”?). Third, the F Scale was built from the ground up to correlate with the Ethnocentrism Scale. Items that did not correlate with prejudice were dropped from the F Scale item pool. If the Dogmatism Scale can do almost as well, without sending its potential statements through such a sieve, it could be argued that Plant’s data indicate that the domain of the D Scale is really more related to ethnocentrism than “fascist authoritarianism” is. Fourth, the strongest correlations in the study existed between the F and D Scales, about .72, which nearly equals the usua! level of reliability of the two tests. So the D Scale can hardly be measuring something substantially different from the F Scale. And finally, since the Ethnocentrism Scale also consists entirely of protrait items, we have a pretty good idea of what mostly bound these three scales together. We do not know what the relationship among these constructs would be if response sets were controlled, except that they would be much, much lower (Campbell, Siegman, and Rees, 1967; RWA, pp. 26-35). — Bob Altemeyer, The Authoritarian Specter, 2581

Categorization

• XREF: Connects to the broader methodological debate about whether the F Scale vs. Dogmatism Scale rivalry was partly an artifact of item construction; pairs with Altemeyer’s treatment of response sets and acquiescence in earlier RWA chapters.
• DEFINE: Clarifies the ‘protrait items’ problem — scales composed entirely of agree-direction items can be bound together by acquiescence response set rather than shared content.
• SEED: An essay angle: how the construction process of a measurement instrument can manufacture the apparent superiority of one theory over a rival.