An empirical attempt at evaluating stress: A failure discovered through cross-validation

Abstract

Stress remains popular as a psychological construct. Different aspects of stress are emphasized depending upon the environmental issue, target population, and measure used. Existing measures are often confounded between causes of stress and effects of stress and also may emphasize a particular perspective on stress. Here we evaluate the empirical method of item selection as an alternative for developing a stress scale, using salivary cortisol levels as the empirical criterion. Items were adapted from measures of perceived stress, daily hassles, and life events as used in two studies of stress that measured salivary cortisol. Correlations with cortisol levels led to the retention of 75 items of the pool of 535, which were administered to a third sample of 28 medical students. The 75-item scale did not correlate with cortisol levels. Of 15 individual items that did, six correlated in the opposite direction to that predicted. Results illustrate the dangers of empirical item selection methods.