The proposed study will develop and demonstrate an empirical methodology for assessing the roles of insurance type, health-related patient characteristics, and physician habits on the amount of medical treatment provided to a patient. The study will focus on practice patterns of physicians in a primary care, ambulatory setting. The empirical model is explicitly derived from an underlying economic optimization model. The statistical methods used here will allow measurement of the separate influences of physician habits, insurance type, and patient health characteristics on the amount of treatment provided. Also, a separate channel of influence, by which the mix of insurance and patient characteristics treated by a doctor affect the habits of that doctor, will also be quantified. The model introduces the concepts of """"""""strong practice style variation"""""""" and """"""""weak practice style variation"""""""", and a method for calculation of upper bounds of these two measures will be introduced and implemented. The study will contribute to the literature on practice variations by providing an econometric methodology for measuring the contribution of practice style variation to the total amount of variation in medical treatment for a given medical condition. The study will also provide new estimates of affects of insurance type on amount of treatment and will assess the biases that can result from studies that employ less sophisticated empirical models. Thus, new estimates of differences in the quality/quantity of treatment received by an underserved group (Medicaid recipients) will be provided. The model requires data sets that provide information on the quantity of treatment provided to a patient, patient characteristics, and multiple observations per sampled physician. The present study is intended as a pilot demonstration of the method and therefore will be somewhat restricted in implementation. The present study will limit to seven patient conditions and use the amount of time the doctor spends with the patient as a measure of amount of treatment provided. Future studies are anticipated that can use the methods developed here to study a wider range of conditions and different measures of the amount of treatment.