Prediction analysis of cross classifications /
Hildebrand, David K.,
Prediction analysis of cross classifications / - New York : Wiley, c1977. - xv, 311 p. :ill,
Includes indexes.
Bibliography: p. 294-299.
Without acknowledging the paradigm difference between testing theory and predicting events, researchers in the field of management and organization continue to use the DEL-technique as a promising technique to evaluate theory based on cross-classification data analysis. We address the purpose and interpretation of the DEL-measure within the theory-testing and events-predicting paradigm. We argue that DEL, a proportionate reduction in error measure, is not to be interpreted in terms of the proportionate error reduction of knowing a prediction rule over not knowing it. In addition, a significant DEL-value is not to be interpreted as a dependence-measure of acceptance of a hypothesis as the only and best relationship between two categorical variables, just as a non-significant DEL-value cannot be interpreted as a measure of independence. Furthermore, an alternative proportionate reduction in error measure generates unequivocally interpretable results compared to the DEL-technique.
0471395757
Prediction theory.
QA279.2 / .H54
519.5'4 / HIL
Prediction analysis of cross classifications / - New York : Wiley, c1977. - xv, 311 p. :ill,
Includes indexes.
Bibliography: p. 294-299.
Without acknowledging the paradigm difference between testing theory and predicting events, researchers in the field of management and organization continue to use the DEL-technique as a promising technique to evaluate theory based on cross-classification data analysis. We address the purpose and interpretation of the DEL-measure within the theory-testing and events-predicting paradigm. We argue that DEL, a proportionate reduction in error measure, is not to be interpreted in terms of the proportionate error reduction of knowing a prediction rule over not knowing it. In addition, a significant DEL-value is not to be interpreted as a dependence-measure of acceptance of a hypothesis as the only and best relationship between two categorical variables, just as a non-significant DEL-value cannot be interpreted as a measure of independence. Furthermore, an alternative proportionate reduction in error measure generates unequivocally interpretable results compared to the DEL-technique.
0471395757
Prediction theory.
QA279.2 / .H54
519.5'4 / HIL