The manipulation of depressive self-report in college students

Recently, the use of self-report assessment devices in depression research has come under heavy criticism. Such instruments are commonly employed to split samples into 'depressed' and 'nondepressed' groups. Among the problems associated with this practice is the issue of faking or impression managment by subjects. In this study, a typical self-report depression inventory was administered to subjects in a simulated selection process in order to investigate: (1) the susceptibility of such measures to response bias, and (2) the tendency for college students to partake in impression management during a selection process. This depression inventory was composed of two subscales for measuring cognitive and somatic depressive symptomatology. Each subscale served as a dependent variable in a separate 2x2 with a control group ANOVA design. The independent variables were manipulated in the test instructions for each group. One independent variable consisted of implying to subjects that either high scores or low scores were the criteria for selection to participate in a future research project. The other independent variable was the inclusion or exclusion of a stigmatizing aspect to the test instructions suggesting that subjects selected would be considered emotionally unstable. In all experimental conditions, subjects were told that if selected they would receive a monetary reward for their participation in the future project. In the control condition subjects were instructed to complete the inventory anonymously for the purposes of gathering normative data. It was predicted that for both sets of analyses there would be a significant interaction between the independent variables such that subjects would score significantly differently in high and low directions only when stigmatization was not included in the test instructions. Further, it was predicted that the mean of the control group would fall between the means of these groups. The results of the somatic subscale strongly supported these hypotheses. On the cognitive subscale, while the interaction between the independent variables was not statistically significant, partial support for the hypothoses was provided in that differences betweeen the means of the experimental groups appeared only in the absence of stigmatization. These results raise questions about the validity of research utilizing selfreport assessment devices to classify college students as 'depressed' and 'nondepressed' while extending the concept of impression management to the self-report of depressive symptomatology.