A REANALYSIS OF MILLER AND FRIESEN LIFE-CYCLE DATA

This paper reanalyzes Miller and Friesen's (1984a) life cycle dana by statistically assessing their conclusions regarding the life cycle imperative. Three separate models were tested: a three‐phase model, a four‐phase model and a five‐phase model. Using the del procedure for the prediction analysis of cross‐classification tables, a sensitivity analysis showed that there is support for the life cycle progression hypothesis, but only when firms that stay in the same phase are not treated as errors. When firms that stay the same are increasingly weighted as errors, del values drop significantly. In general, the three‐ and four‐phase models show better support for the life cycle hypothesis than the five‐phase model originally used by Miller and Friesen. Copyright © 1990 John Wiley & Sons, Ltd.