A Markov chains model of medical specialty choice†

Excerpts from this paper were presented at the 1972 annual meeting of the American Sociological Association held in New Orleans, Louisiana. Preparation of this paper was supported in part by Research Training Grant 5T01HD00164 of the National Institute of Child Health and Human Development. Analyses were made possible through grants received from the Duke University Computer Center and the University of Manitoba Computer Center. The writer is indebted to Jack J. Preiss for provision of the data employed here and for suggestions “along the way.” Among the others whose suggestions proved useful, William M. O'Fallon deserves special mention for his statistical and mathematical advice. The suggestions of Richard H. Conviser, Alfred Hunter, David D. McFarland, Charles H. Proctor, Nancy B. Tuma, and the JMS reviewers are gratefully acknowledged.

Data from 118 members of three consecutive cohorts of students at a private Southern medical school are employed to develop a discrete parameter stationary Markov chains model of the medical specialty choice process. This model was then tested on independent subsamples from the same population, generating a series of predicted specialty choices which did not differ from observed choices by more than chance expectations. Suggestions for future research, beyond the present demonstration of the utility of a simple Markovian model of a choice process, are offered.

Notes

Excerpts from this paper were presented at the 1972 annual meeting of the American Sociological Association held in New Orleans, Louisiana. Preparation of this paper was supported in part by Research Training Grant 5T01HD00164 of the National Institute of Child Health and Human Development. Analyses were made possible through grants received from the Duke University Computer Center and the University of Manitoba Computer Center. The writer is indebted to Jack J. Preiss for provision of the data employed here and for suggestions “along the way.” Among the others whose suggestions proved useful, William M. O'Fallon deserves special mention for his statistical and mathematical advice. The suggestions of Richard H. Conviser, Alfred Hunter, David D. McFarland, Charles H. Proctor, Nancy B. Tuma, and the JMS reviewers are gratefully acknowledged.