Thursday, November 6, 2008

Controversy over the Construct of Intelligence

The study of intelligences has come across large differences between races and IQ scores. IQ scores of intelligence are set on a bell curve where 100 is the average with a standard deviation of 15. Asians on average, fall one standard deviation above whites whereas, blacks fall one standard deviation below whites on the bell curve. These racial differences are not due to test bias but reflects mental capacity differences that g (general intelligence factor) embodies (Gottfredson, 2002). Although there are differences between several racial groups it is the gap between blacks and whites that focuses on disparate impact. Disparate impact does occur when selection involves the use of intelligence testing. The controversy is between equity and efficiency.
Those in favor of equity suggest several changes that could reduce disparate impact. Goldstein et al. (2002) suggest that non-cognitive ability tests illustrate the reduction of group differences especially when used in combination with other methods of assessment (i.e., personality, samples). Researchers suggest that the validation of such devices would gain equity and efficiency. Other suggestions include banding intelligence and other assessment scores into groups. People with similar scores could be grouped into a test score band. Banding although sometimes looked down upon, would reduce adverse impact. A third suggestion includes altering the testing method from written intelligence tests to oral tests and paper/ pencil tests to providing sample work. Preliminary research suggests that such changes reduce group differences but are less predictive of overall job performance and academic success (Murphy, 2002). There is a final suggestion and that is to eliminate intelligence tests from the selection pool completely to eliminate the effects of adverse impact.
Those supporting efficiency states that selection procedures will not benefit from slashing the use of intelligence testing. Intelligences are one of the best predictors of job performance to date. Although there are other good predictors (i.e., personality) these are best used in collaboration with intelligence tests. Such suggestions will minimize the test validity (Gottfredson, 2002). Personnel decisions could affect employee opportunities (Murphy, 2002). In the short-term such eliminations may decrease the pressures disparate impact places on researcher but it is not a long term solution. In response to altering assessment methods researchers argue that such changes are not assessing cognitive ability and thus are not assessing intelligence. Such changes would reduce the validity and predictive ability of intelligence tests. Supporters of efficiency although not against equality do not necessarily believe that equality should be an issue in selection. This rational is derived from looking at the purpose of selection systems and that is to select the best employee for the position. If intelligence measure show the there are group difference not due to test bias than why would researchers and personnel change the system to select a less qualified applicant.
As presented there remains a large divide between those who want to reduce adverse impact now and those who what to select the best performers. Future research must examine the long term consequences of change selection methods to establish a valid assessment tool. At the end of the day corporations must make the decision whether they will select with equity or efficiency in mind.

No comments: