> -----Original Message----- > From: Stephen Black [mailto:[EMAIL PROTECTED] > Sent: Wednesday, August 27, 2003 11:31 AM > To: Teaching in the Psychological Sciences > Subject: Re: IQ > > On 27 Aug 2003, Allen Esterson wrote: > > > > It was not the correspondent's opinion I was concerned about, rather I > > was hoping for comments on Steven Rose's assertion that IQ tests > > "serve the ideological commitment of testers". Or, to put it in less > > ideological terms than Rose's, that such tests are flawed by class, > > race, and gender biases, by being culture-bound, or [as another Times > > correspondent, cited below, claims] by the conflation of "learned > > knowledge" with intelligence. > > Now that Allen has clarified his interest, let me have a crack at > answering it. I would say that the claim that conventional IQ tests > are biased by being culture-bound is essentially correct. But that > doesn't necessarily make them invalid. IQ tests were developed for a > specific purpose: to predict specific kinds of performance in a > particular kind of environment, notably school achievement and, > perhaps, later success in life as measured by socio-economic status, > occupation, and other such indicators, in an urban, industrialized, > Western and even white society. For a relatively brief test > administered at an early age, it does remarkably well at such > predictions. But it doesn't claim to predict other kinds of > performance in other kinds of environment. It may be unfair to a kid > from the ghetto compared with a middle-class one to ask whether he > knows what a "sonata" is or to explain why cheques must be signed. > But it turns out that for functioning successfully in an urban, > industrialized, Western society, kids who know such stuff do better > than those who don't. Make the test culture-fair and it loses its > predictive power. > > So IQ tests are undoubtedly biased but are still valuable for their > narrowly-defined purpose. But use them to predict something else-- > say, survival in the high arctic or on the streets of inner-city > Detroit-- and knowing about sonatas and cheques is unlikely to have > much predictive power. > > Stephen > ______________________________________________________________ > Stephen L. Black, Ph.D. tel: (819) 822-9600 ext 2470 > Department of Psychology fax: (819) 822-9661 > Bishop's University e-mail: [EMAIL PROTECTED] > Lennoxville, QC J1M 1Z7 > Canada > > Dept web page at http://www.ubishops.ca/ccc/div/soc/psy > TIPS discussion list for psychology teachers at > http://www.frostburg.edu/dept/psyc/southerly/tips > _________________________________________________________ > > > --- > You are currently subscribed to tips as: [EMAIL PROTECTED] > To unsubscribe send a blank email to [EMAIL PROTECTED]
I'll go a step farther than Stephen. The term "bias" has a specific meaning in this context - that the regression of a criterion measure (e.g. grades in school, or success in life) on IQ scores (or other standardized test) yields different slopes and/or y-intercepts for different sub-groups. Here's an example. Let's use SAT (test) and grades (criterion). For a large heterogeneous group assume the relationship is linear, with GPA = 2 + .001875*SAT, and this is used to predict GPA and admit/reject candidates. But unbeknownst to the test givers the relationship for blue eyed people is really GPA = .0025*SAT, while for brown eyed people it is really GPA = 2+ .00125*SAT (intercepts are identical but slopes differ). If the SAT is used for the heterogeneous group it is biased _against_ blue eyed people. A blue eyed person, with SAT 500 would be predicted to have a (2 + .01875*500 =) 2.94 GPA. But this under-predicts their actual GPA which would be (2 + .0025*500 =) 3.25. By the same token the SAT would be biased _in favor_ of brown eyed people because it over-predicts their actual GPA (2.94 versus 2.625). To my knowledge, current IQ or major achievement/aptitude tests do not exhibit either slope or intercept bias for the customary classification schemes (gender, race, etcetera). People are often confused by this, because they use the term "bias" to describe differences in _group means_ rather then the predictive relationship between the test and the criterion. If we rank order the group means, we see that black Americans are lower than American Indians, who are lower than whites, who are lower than Oriental Americans and some Jewish groups (reflecting, by way, the relative rates of success in western society for these groups). But, the regression between the test and the criterions, by themselves, do not differ enough for these groups to justify calling current standardized tests biased against any group (if anything, IQ tests slightly over predict criterion for black Americans, but that's another story). It is likely our current tests are not biased because of the rather intense scrutiny they currently undergo. As to Stephen's point that the current IQ tests would not predict success in Detroit or the artic, this is an open question. Meta-analyses of IQ reveal surprisingly strong correlations with virtually every profession studied, from the professional to the menial. It is possible, but unlikely, that the fundamental role intelligence plays in such a broad range of professions would not be operating in the day-to-day life or inner cities or the artic. I realize the standard answer to these questions is that "IQ predicts within our society" (with the caveat "only in our society" mentally added), but I have a hard time finding evidence or logic for the "only" part. Unfortunately, I am not familiar with Stephen Rose's writings. Whatever his criticisms, they must be squared with the fundamental fact that intelligence test scores (specifically, the 'g' component, not the stuff we memorize in school for the tests) correlates strongly with real-world success across a broad spectrum. Sorry if I repeated things earlier in the history of this thread. ============================================ John W. Kulig Professor of Psychology Plymouth State College Plymouth NH 03264 ============================================ "Live simply that others may simply live" Contemporary saying. --- You are currently subscribed to tips as: [EMAIL PROTECTED] To unsubscribe send a blank email to [EMAIL PROTECTED]
