An Empirical Comparison of Several Within-Subjects Forms of Expectancy Theory.

Abstract

Expectancy theory is formulated as a choice model, so when the theory is predicting effort the assumption is that individuals choose that level of effort which has for them the highest perceived probabilities of resulting in their valued outcomes. The purpose of this study was to provide the first test of this original choice model of expectancy theory in predicting effort. This means that an expectancy motivational force score was generated for each of several effort levels for each subject and that the subject was assumed to choose that level of effort which had the highest activational force score. In contrast, nearly all the research to date has utilized computed motivational force scores for only the single alternative of high effort--a single-alternative model. A few studies have computed motivational force scores for only the single alternative of high effort--a single-alternative model. A few studies have computed the motivational force score for high and low effort an subtracted the score for low effort from the score for high effort, resulting in a form of marginal utility--a difference model. Because of theoretical accuracy, an ipsative orientation, and mitigation of response bias, the choice model was predicted to be somewhat superior to the difference model and greatly superior to the single-alternative model. In addition, the intra-subjects correlation technique was predicted for all three models to be superior to the inter-subjects correlation technique. The correlational research design used data from 74 university students who were participating in a time allocation feedback study. Effort was defined as reported hours spent and preferred hours spent in six activities. Expectancies for selected outcomes were asked for one, two, three, four, and five hours of effort for each of the six activities, enabling the computation of a motivational force score for each level of effort. The hypotheses were tested by comparing correlations. The results showed the choice model and the difference model to be superior to the single-alternative model. The data also suggested, contrary to the hypothesis, that for the intra-subjects correlation technique the difference model was superior to the choice model. The intra-subjects correlation technique was found to be more effective than the inter-subjects method, though less so for the single-alternative model. An analysis of the results indicated that the difference model was more effective than expected because (1) it had an ipsative orientation, (2) marginal benefits were crucial, and (3) it benefitted from spurious correlation, which in a sense combined both low and high effort scores as predictors. Changes were suggested for reformulating and testing the theory in a more mathematically rigorous form, possibly dealing with effort budgets instead of effort levels. It was concluded that the choice model was not the most effective model, but that instead some combination of (1) the single-alternative model (using both high and low effort alternatives), (2) the difference model, and (3) the intra-subjects correlation technique provided the most effective and parsimonious test of expectancy theory.