“Do psychological tests discriminate against certain groups?”
· Standardized tests have cited several weaknesses which include the inability of a single instrument to accommodate the diversity of test takers, the inadequacy of representative sample items, the relationship of knowledge tested and the items content validity, and finally the general lack of theoretical consensus concerning the nature of cognitive processes involved in assessed intellectual skills.
· Bias in the statistical averages of different racial, cultural, and socioeconomic subgroups when circumstances cause one or more of the constituent subgroups to perform differently from others
· People who are left-brain dominant prefer to work independently and enjoy competitiveness and artistry. The right hemisphere specializes in simultaneous information processing and holistic thinking. People who predominantly utilize the right hemisphere are more concrete, visual and tactile learners; therefore one group will do better at certain tests versus the other groups.
· Psychologists have validated tests for employment purposes including cognitive ability tests, personality tests, honesty and integrity tests, and interest inventories to screen applicants. These tests raise concerns about validity, privacy, and discrimination.
· Item difficulty, depending on education, culture, socio-economic background, etc., a person may or may not know a certain question. For example: Who was Boliver Scagnasty? Who was Martin Luther King?
· There is Legislation that protects minorities against discriminating psychological testing (ex. EEOC published guidelines)
· No difference in outcome when 50 inner city kids where given a standard version (white, middle-class language) and alternate version (African-American dialect). Therefore, language cannot be used as an argument.
· Experts took out all questions that were seen as unfair (about 16% of test) and administered it to two groups (minority and majority). There was no difference in questions they thought were biased for the majority group.
· Regression plot: When a group score lower than the other does, it’s important to see if the test can predict the performance equally well for both groups. If it can, then it doesn’t discriminate.
· The increase in various employment litigation issues such as negligent hiring an employer may successfully insulate itself from costly litigation by rejecting an applicant based upon his/her psychological profile.
· Research done by Farnham-Diggory (1970) and Frederickson (1986) showed inherent bias in achievement tests that discriminate between individuals.
· Research found that different cultural and racial groups have preferred learning styles and therefore different dynamics can hurt or help them.
· Some psychological tests can specifically screen out an individual, especially with people that have psychiatric disabilities that may or may not have anything to do with the position they are seeking at the job they are applying for.
· A poor student may make a guess and get it right, whereas a good student may be suspicious of an easy question and take the harder pathway and get it wrong – which makes for not effective test results. The SAT test for example gives different tests, so is that fair? Everyone does not have the same chance to do as well on this test.
Psyc 542 (Tues.)
March 21, 2000
Do Psychological Tests Discriminate?
The fact that certain ethnic groups score lower on average on some psychological tests has caused some people to believe that it is because the tests themselves are biased toward specific groups of individuals. What is important in understanding if psychological tests discriminate is determining whether the tests are as valid for one group as they are for another. There is legislation protecting minorities against this. For example, in 1978 the Equal Employment Opportunity Commission published what they called the “Uniform Guidelines on Employee Selection Procedures”. These guidelines apply to tests of intelligence as well as vocational tests. Not only do they forbid discrimination, but they also assure that, “the government will view any screening procedure, including the use of psychological tests as having an adverse impact if it systematically rejects substantially higher proportions of minority than nonminority applicants” (Kaplan/Sacuzzo, p. 537).
Specifically in the past, critics have accused the intelligence tests of using language that is part of a white, middle-class background. This would imply that children of other backgrounds are set up for failure. However, there has been research done that proves that the language of the test is not responsible for the differing outcomes of different groups. In 1971, Quay gave the Stanford-Binet test to one hundred inner-city children. Half received the regular version, while the other half received a version in African-American dialect. There was only a difference of less than one point in the scores, which shows that the language did not play the part the critics have claimed it does.
Another criticism is that some standardized tests have particular items that are considered not to be fair. In a study designed to determine if this makes any difference in the scoring, 16% of the questions on an elementary reading test were “thrown out” after they were thought to be biased toward the majority group. However, the differences among minority and non-minority groups were no different than the originals. Flaugher (1978) argues that a so-called “biased” test is not supposed to ask about only the things a test taker might know, but about many different bits of information.
What some have considered as a problem is that the SAT yields lower scores for minorities than for other groups. When a test is administered to two groups of individuals, and they score significantly different, it is important to use statistical analysis such as a regression plot. If group 1 scored lower than group 2, it is important to know if the test predicted the performance equally well for both groups. If the relationship between the test score, and the performance on the criterion is the same, there will be a single regression slope, which predicts performance equally well for the two groups, even though their means are not the same. What this shows, is that there is not enough evidence for test bias.