Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


The principal objective of the present research was to examine Mobley's (1982) contention that a dynamic experimental design is necessary in order to advance our understanding of the cognitive and behavioral events which precede turnover. The one aspect of Mobley's contention tested here was the extent to which a repeated measures design contributes to the predictive power of existing models of turnover. A second objective was to conduct a competitive test between the Mobley, Horner and Hollingsworth (1978) and the Price and Mueller (1981b) models of nurse turnover. Four non-profit community general hospitals and two government-owned general hospitals provided the subject pool from which a study sample (n = 527) of registered nurses was randomly selected. A 53-item employee survey, containing the variables in both turnover models, was mailed (Time 1) to each nurse in the study sample. From those nurses who returned the questionnaires, a dynamic paradigm group with complete data (n = 84) was randomly selected to receive additional mailings of the survey two months (Time 2) and four months (Time 3) later. Turnover data was collected at the time of each survey and at the end (Time 4) of the six-month study. Multiple regression procedures yielded traditional static-paradigm R('2)s for each turnover model (Mobley R('2) = .29; Price R('2) = 2.6). The static R('2)s were then compared to dynamic R('2)s that took into account the changes that occurred in model variables over time. Although the dynamic paradigm produced slightly higher R('2)s (Mobley R('2) = .20; Price R('2) = .41), neither significant partial correlations of first differences nor significant extra sums of squares resulted. No significant differences were ascertained when the static R('2)s for each model were compared. It was concluded that although the dynamic research paradigm does record process events, the method presently does not significantly improve the predictive ability of existing turnover models. At this time, dynamic paradigms can contribute most by delineating the sequence and flow of events which precede turnover. Once this is accomplished, their usefulness in prediction and intervention may improve. It was further concluded that both theoretical models considered here were comparable in their modest predictive ability.