We should get rid of these big standardized tests to select students into college. There are many good reasons; let’s just review a few.
First, these tests are not very valid. They correlate anywhere from 0.20 to 0.30 with grades in college. This means that on a 0-10 scale, where 0= no correlation and 10= perfect correlation, the correlation of test scores to grades is about 2-3. Not very impressive. Of course, this matches our common sense. Can you really expect to predict college success with a 4-hour test? It is much better to rely on the student’s history of performance in high school as well as other measures of intellectual attainment and motivation.
Second, these tests unfairly discriminate against women and minorities. Women score lower than men on the math or quantitative part of the ACT, SAT and GRE even though we know that women are as intelligent as men. African Americans score substantially lower than Caucasians on the tests, there is good evidence these tests are culturally biased against minorities. It is not fair to use biased tests as the predominant means of selecting people into colleges and universities.
Third, research by Stanford professor Claude Steele has shown that these sorts of tests activate what is called a stereotype threat. What this means is that since minorities have learned they are supposed to score less well on these tests, they perform according to the stereotype. Why do they do this? Steele argues that the tests prompt a view that they are less intelligent, so minority students are upset and demoralized.
The SAT and ACT exams were developed precisely because high school performance doesn’t tell you that much about how a student will do in college. In the United States, there are dramatic differences between secondary schools, even within the same city. Fore example, in Florida’s Miami-Dade County, 86 percent of students at Coral Reef Senior High school achieved the minimum proficiency level in a state wide math test, compared to only 3 percent of students at Miami Douglas MacArthur Senior High school. So, goes a 3.0 GPA at Coral Reef mean the same thing as a 3.0 GPA at Douglas MacArthur? Of course not. And this is why high school grades are poor predictors of college success It is impossible for universities to compare applicants across different high schools based on their GPA.
The SAT and ACT tests have been subject to massive validation efforts by some of the best industrial psychologists in the world. No, the validity isn’t perfect, but it’s better than anything else researchers have been able to identify, including high school achievement. The reason is that these tests are carefully developed and validated, and all applicants take the same basic test. So, we don’t have the apples and oranges problem we do with high school grades. By the same token, the GRE also works well. The authors of an analysis of 1,753 independent samples of the GRE (on 82,659 graduate students) concluded. The GRE is valid across disciplines for a variety of important criterion measures and not just 1st year graduate grade point average as is often assumed.
In terms of bias, the research shows that, although African Americans do score lower on the test, the test results predict college success as well for African Americans as they do for Caucasians. So, the tests are not biased. Women do tend to score lower on the math part, but that is offset by the fact that they tend to score higher on the verbal part.
Finally, evidence indicates that test preparation courses do help somewhat. But the benefit is modest, and evidence suggests that the existence of such courses has done nothing to underline the validity of the SAT, ASCT, or GRE.