Friday, September 22, 2006
....to ruin a perfectly good theory. So says Beth Visser and colleagues in the latest edition of Intelligence. She tested Gardner's theory of Multiple Intelligences (MI), gasp!, empirically. What did she find?
Our analyses of tests measuring the "intelligences" of Gardner's MI theory revealed that many of those tests were substantially intercorrelated, despite representing different domains of Gardner's framework, and also showed strong loadings on a g factor and strong correlations with an external test of general intelligence. These results are difficult to reconcile with the core aspects of MI theory.
More specifically, the researchers combed over quite a few tests to find two or three that match Gardner's own explication of his intelligences. When they gave them to a group of college students, (surprise surprise) they all had a strong correlation with each other, except the tests that measured Bodily-Kinesthetic, Intrapersonal, and Musical "intelligence." I don't think anyone outside of Gardner and his sycophants think that tests that measure Bodily-Kinesthetic skills are actually a subset under intelligence. Likewise, Intrapersonal intelligence is little more than self-esteem/self-concept, which we know doesn't really predict anything; the low musical correlations, however, were a bit of surprise (esp. given Carroll's opus), but in all likelihood, this a solely a function of the tests horrible reliability.
What can we conclude? (a) sans physical ability and self-esteem, Gardner's intelligence domains can best be thought of as second-stratum factors of Carroll's hierarchy, (b) that being said, before any more time, effort, and energy is exerted in MI, Gardner or his colleagues need to provide "falsifiable, testable, MI-based hypotheses that would predict results different from those predicted by existing models of the structure of mental abilities," and (c) any educational curriculum based on MI theory is, currently, a waste of teacher time and student energy.