A note on the effect of filling out an "anxiety scale" on examination performance

Certain statistically significant relationships were found in an earlier study by the writer (Carrier, 1957) designed to determine whether certain personality measures are related to course examination performance under stress conditions. But the conclusions appeared tenuous because mean score differences between high and low stress experimental groups were generally small. Post hoc speculation suggested several tenable explanations. One of these involved the old problem of the extent to which the investigator's measuring operations themselves may unwantedly influence that which is measured. Since the present author's earlier study attempted to get self-reports of autonomic responses, one effect of the measure designed to get these reports may have been to increase emotional responses where it was especially not wanted, i.e., in the low stress group. Consequently, a partial "replication" was planned to determine whether filling out anxiety scales does in fact depress examination performance. The Ss were 1,074 students at the University of Colorado. Two forms of a 75 item, multiple-choice examination were administered. Both forms contained the same 75 questions, but the order of questions was different in the two forms. One form was designated as the "anxiety" form, the other was called the "neutral" form. The mean examination score for the 553 students who filled out the anxiety scales was 53.33; the mean score for the 521 students who filled out the personal data sheets was 53.36. The standard deviation of the former group was 8.89; that for the latter group was 8.95. Both distributions were normal in shape by visual inspection. These remarkably similar results give no support to the contention that filling out anxiety scales during an examination depresses performance by "reminding" the student of his emotional state.