Test fairness

from Wikipedia, the free encyclopedia

Test fairness is a term from psychological diagnostics and test theory. Test fairness is a quality criterion for tests. The test board of the Federation of German Psychological Associations defines the term unfairness as the extent of systematic discrimination of certain test persons on the basis of their ethnic, socio-cultural or gender-specific group membership.

History

Test fairness has been a relevant topic in psychology at least since R. B. Cattell , as he for the first time conceived a so-called Culture Fair Intelligence Test for the culture and language-free measurement of intelligence. In the course of the crisis in psychological diagnostics in the 1970s, the theory of domination was increasingly criticized for the fact that psychological diagnostics merely represented the interests of capital and thus contributed to the oppression of the ruled. With regard to the educational participation of educationally disadvantaged groups, the fairness of standardized tests has been and is particularly discussed in places where they have been used more and more for a long time, such as in the United States. Psychological selection tests thus find themselves in a dilemma between replicating social conditions and, on the other hand, the possibility of actually recording those who are most capable for a job regardless of social status and gender through culture-free tests .

As a rule, however, typical differences between groups of applicants in psychological selection procedures remain stable, as can be shown using the example of African-American applicants in the Scholastic Aptitude Test . This immutability of the results has led proponents such as the philosopher Ralf Dahrendorf to the fact that a regulation in the sense of quota fairness - the so-called affirmative action - was introduced in the USA . The downside of this regulation is, of course, that due to a quota system, those who belong to the worse of the better group are disadvantaged and, because of their group membership, are disadvantaged compared to someone with a poorer result. In this case, social and individual justice are in irresolvable contradiction.

Areas of application of test fairness

Test fairness is particularly important where selection decisions are made based on the test result ( high-stake tests ). High demands on test fairness are therefore more likely to be made in psychological performance tests (traffic psychological idiot test , study aptitude tests, occupational aptitude tests) than in personality tests. If personality tests are used to select applicants, one pole of the personality dimension is usually preferred (e.g. extraversion instead of introversion for a waiter). De iure would therefore be the fairness of the extraversion test z. B. Check in terms of gender. In fact , such fairness tests are rarely carried out, all at most when media attention is drawn to a fairness deficit that results from a selection test (e.g. disadvantage of African-American applicants in the Scholastic Aptitude Test ).

The particular difficulty in assessing the fairness of a test is that there can be well-founded differences between different groups of applicants. For example, women might actually be better suited to various professions in the field of copywriting because of their higher verbal skills. An advertising copy test should therefore also be able to capture this difference between men and women. For this reason, the model of equal success probabilities applies in psychological diagnostics. This means that different groups of applicants do not have to be represented aliquot in the sense of fairness of quotas among the selected, but that the different groups of applicants among the selected have the same probability of success with regard to the target criterion (e.g. professional or academic success). This model makes it clear that such tests of fairness can possibly take years, as the probability of success on the target criterion can often only be recorded years later. Another problem with this model is that a third intervening variable could affect both measures (aptitude test and outcome measure) that has nothing to do with aptitude (e.g. hesitation in answering questions).

Test fairness is particularly often examined in relation to

  • gender
  • socio-economic status
  • Education vs. Educational distance
  • Migration background
  • linguistic background

Methods of test fairness improvement

With regard to improving the fairness of psychological tests, there are currently no clear guidelines on how to proceed systematically. In principle, there are several options:

  • Subsequent correction of the test result: with the help of the DIF analysis , individual item results can be corrected subsequently.
  • Develop new tasks that discriminate less: on the basis of theoretical considerations, but also through trial and error, tasks that discriminate little can be determined (e.g. tests on mental rotation discriminate women more than other 3D tests even after training).
  • Improvement of the preparation of different groups of applicants. B. Specific summer enrichment programs for low-education applicants for study aptitude tests to be used, which are intended to balance out the environmental differences between the groups of applicants.

Models

There are different models of what fairness means:

  • Cleary model
  • constant ratio model
  • conditional probability model
  • equal probability model

Individual evidence

  1. ^ Test Board (of the Federation of German Psychological Associations) (1986). Communication, Diagnostica , 32, 358-360.
  2. ↑ Spiel, C., Litzenberger, M. & Haiden, D. (2007). Educational and psychological aspects of selection procedures. In C. Badelt, W. Wegscheider & H. Wulz (eds.) University entrance in Austria (pp. 253–327). Graz: Graz University Press.
  3. ↑ Linn, MC & Petersen, AC (1985). Emergence and Characterization of Sex-Differences in Spatial Abilities: A Meta-Analysis. Child Development, 56,6,1479-1498
  4. ↑ Hesser, A., Cregler, LL & Lewis, L. (1998). Prediciting the admission into medical school of African American college students who have participated in summer academic enrichment programs. Academic Medicine, 73, 2, 187-191.
  5. ↑ a b c d Lothar Schmidt-Atzert, Manfred Amelang: Psychological Diagnostics (textbook with online materials) . Springer, 2012, ISBN 978-3-642-17001-0 , pp. 170–171 ( limited preview in Google Book search).