Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Extensive research has been conducted assessing psychophysiological reactivity to experimental stressors in various populations. However, there is a paucity of empirical investigations concerning the test-retest reliability of these experimental stressors. Establishment of the relative reliability of specific stressor procedures is important so that results can be compared over time and across studies. The purpose of this investigation was to compare test-retest reliability of two mental stressors (i.e. Quiz Electrocardiogram, mental arithmetic) and two physical stressors (i.e. cold pressor, isometric challenge). These stressors were presented in a counterbalanced fashion to forty-eight undergraduate and graduate students who returned two weeks later for the same stressor presentation. The experimental conditions comprised a 2 (sessions) x 2 (baseline/test) x 4 (stressors) within-subject design. The major hypothesis was that physical stressors would have greater univariate and multivariate test-retest reliability since they are a direct function of physical stimuli and result less from cognitive mediation which can vary across sessions. Thus mental stressors were hypothesized to have lower reliability because they directly result from cognitive involvement which may change across testing sessions. The results showed that all four stressors generated significant increases in physiological arousal over baseline. In general, univariate and multivariate test-retest reliability was consistently significant and equivalent across all four stressor conditions. More specifically, univariate reliability as measured by Pearson correlation, was adequate for absolute test values and baseline levels across the physiological variables of skin temperature, skin resistance, vasomotor response, heart rate, systolic and diastolic blood pressure. Forearm electromyogram was the only dependent measure found to be unreliable. Difference scores, which represent change from baseline to test conditions, did not have adequate univariate reliability. Multivariate reliability as indexed by profile of similarity, was found to be adequate across baseline, test and difference scores for all stressors. The hypothesis of differential reliability between physical and mental stressors was not supported. In conclusion, the consistent reliability found across stressors provides an empirical basis for the validity of conclusions drawn from these procedures in psychophysiological research.