Advantages of computer administered testing by Paul Kline
I am reproducing what Paul Kline, (1937-1999) a psychologist at the University of Exeter, wrote on computer administered testing.
1. The first and greatest advantage in computer-administered testing is the capability it provides for almost immediate feedback of results. In the applied setting this is of considerable value. For example, in all counseling and therapeutic applications of tests immediate access to results is an enormous advantage. The test can be discussed while the whole thing is fresh in the subject’s mind. In selection where the number of candidates is small and there is personal contact between the selector and candidates again immediate results are valuable. Where a huge number of subjects are tested, as in selection procedures for the armed services or civil service, this facility is not so important. In general, however, there can be no doubt that immediate feedback is a valuable characteristic of computer testing.
2. A second advantage of the computer-administered test is that the administration is always the same. Variance due to testers is, therefore, eliminated. It should be pointed out that there is a possibility that a different computer screen from that on which the test was standardized might produce a different result so that strictly identical computers should be used to those for which the test was specified. Of course, traditional tests can be administered by tape recorder to ensure identity of administration, but this always seems, in my experience, somewhat artificial, when the human tester is present, whereas the impersonal computer is more acceptable.
3. Where there is a shortage of skilled personnel, the prewritten interpretations can be valuable. These are used by many personnel managers in commercial organizations where the psychological and psychometric knowledge of the test users is low, or even non-existent.
4. Obviously where valid items are used which can only be presented on computer, the computer is simply irreplaceable. However, except in certain specialized applications, computer items have not yet replaced those used in traditional tests.
It is possible to develop computer programs that will produce items of a given difficulty level where the items parameters are simple. This can be done for elementary cognitive tasks, as defined by Carrol (1983). These, as the title suggests, measure the simple cognitive processes which constitute general intelligence (according to cognitive theorists such as Carroll, 1983, and Hunt 1978). Semantic encoding and spatial orientation are typical examples.
Irvine et al. (1990) and Dan et al. (1993) have produced a set of tests in which items of given difficulty level are generated by the computer, and have discussed some of their underlying rationale. Their arguments and claims will now be scrutinized.
5. The fact that, in computer-administered tests, the data for each item are automatically stored makes the statistical analysis of the test effortless. It saves the necessity, and the possible errors, of punching in all the data.
6. One impressive application of computerized tests is in testing the severely handicapped. If, for, example, subjects have very poor motor control, special keyboards with large keys can be used. Wilson et all (1982) discusses some useful input devices for this purpose. For those whose vision is poor large script can be used on the computer screen. Of course, such tests are special tests and they could not be used for comparison with normal subjects. Nevertheless valuable clinical information can be obtained from such special computer testing.
