By Douglas K. Detterman, Western Reserve University.
The book «Introduction to Psychology». Authors — R.L. Atkinson, R.S. Atkinson, E.E. Smith, D.J. Boehm, S. Nolen-Hoeksema. Under the general editorship of V.P. Zinchenko. 15th international edition, St. Petersburg, Prime Eurosign, 2007.
Article from chapter 12. Individual differences
How can you know that two kinds of measurements point to the same thing? You calculate a statistic called correlation (which ranges from 0 to 1,0), and the higher the correlation, the more similar these kinds of measurements are. The tests of aptitude, achievement, intelligence, and cognitive ability are so highly correlated with each other that many experts believe that they are all the same tests. The abilities mentioned in test names may have more to do with social acceptability than what the tests actually measure. It won’t be a surprise if I say that these tests are highly correlated, since from the Binet test to the newest computer-assisted Council for Graduate Written Examinations (GRE) test, they are designed to predict learning success .
A test is nothing more than a set of actions that, in theory, allows you to predict future behavior. The simplest thing that tests should predict is the results of running the same test in the future. This quality is known as the reliability of the test. Tests of intelligence, ability, and achievement are the most reliable of all psychological tests. Individually administered intelligence tests and the same tests administered later correlate at levels above 0,9. In 1932, a group test was conducted for all Scottish schoolchildren. Recently, some of these people were asked to take the same test again. The correlation of tests separated by 66 years was 0,74. Compare this to an average reliability of 0,6 for measurements of height, weight, and blood pressure taken in the doctor’s office. For most psychological tests of personality, psychopathology, or motivation, the reliability of two results of the same test in the range of 0,4-0,8 is considered high.
Do the tests predict anything useful? Certainly! Anghoff and Johnson correlated the scores of 20 students on the SAT and GRE. It turned out to be equal to 0,86. Adjusted for gender and specialization, the correlation increased to 0,93 or even higher, while the quality of teaching was of little importance. High college fees won’t get you into graduate school, at least not based on test scores. Other academic criteria, such as grades or class position, can also be predicted by these tests, but less well because grades and class position are themselves less reliable. Intelligence tests are good predictors of success in many areas of real life, from the ability to pilot a jet fighter to the ability to become a good lawyer or musician. Occupations that require a high IQ are also the most socially desirable. Most importantly, mental tests predict academic success, usually guaranteeing success in life.
In that case, why do they get so many negative reviews? First, tests are often misused. While almost all colleges require an admissions test, approximately 80% of colleges accept almost everyone who applies. Why are they doing tests? I think to create the appearance of selection. Even in the case of the 20% of colleges that do select, many factors other than test scores influence admission decisions. If two students score the same on tests, the advantage will be given to the one who has the following characteristics: minor, athlete, attended a good private high school, family income of more than $70, one parent’s college education, continuity (one of the parents went to this same college), parental donations to the college (Bowen & Bok, 1998). Prior to the introduction of tests, only these factors were taken into account when enrolling. I think enrollment in educational institutions would be more fair if it was only based on test scores.
The second reason people often have negative opinions about tests is that there are situations in which tests cannot predict the outcome. When applicants are carefully selected for IQ, they will all have similar IQs, which thus cannot affect the outcome. The American education system is based on the assessment of intelligence, and each stage requires a higher level of academic success. You may have noted that college is harder than high school. Here is the reason for this. The average IQ of college graduates is over 110, and the average IQ of those graduating from professional school or graduate school is over 120. Intelligence will make little contribution to success in graduate school or medical school. Differences in success will be determined by other parameters, such as diligence and character, because everyone is smart. You must be tall to play basketball. The NBA selects players based on their height, and NBA players are much taller than the average person. But the correlation between height and performance in the NBA is almost zero. Saying that intelligence doesn’t matter in graduate school or medical school is as foolish as saying height doesn’t matter in basketball.
Tests like the SAT and GRE actually measure intelligence. They are highly reliable and predict real world behavior better than anything else we know. Tests are often abused and are not the main evaluation of a person. But when used correctly, tests can make selection decisions fairer and more accurate and help students plan for future education.
Why IQ, SAT and GRE don’t measure general intelligence
Thousands of «validity» studies show that general intelligence tests predict a wide range of different behaviors, although not perfectly, but better than any other method known to us. Grades of first-year students are somewhat better predicted by IQ scores than by grades or characteristics obtained by students in high school. Grades obtained by students in their first year of graduate school are also better predicted by IQ scores than university grades and characteristics. But the accuracy of prediction based on IQ (or SAT or GRE) is limited, and scores for many candidates will not be as expected. Test makers argue that even limited predictability can help enrollment officials make better decisions than without tests (Hunt, 1995). See →
Theories of intelligence: results
Despite these differences, all theories of intelligence have a number of common features. All of them try to take into account the biological basis of intelligence, whether it be a basic processing mechanism or a set of multiple intellectual abilities, modules or cognitive potentials. See →