Interpretive Error in Psychological Testing.

When evaluating the utility of a psychological test for clinical decision making, both the psychometric properties of the test (i.e., the reliability and validity of the instrument) and the ambiguity of the language by which test results are interpreted or communicated need to be considered. Although each has been studied independently, to date the two have not been related. This paper discusses each of these sources of "interpretive error" with the goal being the development of a model that could systematically relate these two sources of error to the process and outcome of test interpretation. In an example using this new model together with optimistic good-hearted assumptions favorable to tests, it was found that the effect of ambiguity was to more than double the probability that those who test positive will be incorrectly classified. It is suggested that the practice of using non-numeric statements of likelihood be stopped. Instead, results of tests should be presented in tabular form with one row for each diagnostic category. Examples of these tables are provided. It is further noted that there is a significant professional need to move toward empirically supported expertise in test interpretation. Contains 24 references. (Author/MKA)