In a prior article in this journal, John Nyman argues that the effect on health care use and spending found in the RAND Health Insurance Experiment is an artifact of greater voluntary attrition in the cost-sharing plans relative to the free care plan. Specifically, he speculates that those in the cost-sharing plans, when faced with a hospitalization, withdrew. His argument is implausible because (1) families facing a hospitalization would be worse off financially by withdrawing; (2) a large number of observational studies find a similar effect of cost sharing on use; (3) those who left did not differ in their utilization prior to leaving; (4) if there had been no attrition and cost sharing did not reduce hospitalization rates, each adult in each family that withdrew would have had to have been hospitalized once each year for the duration of time they would otherwise have been in the experiment, an implausibly high rate; (5) there are benign explanations for the higher attrition in the cost-sharing plans. Finally, we obtained follow-up health-status data on the great majority of those who left prematurely. We found the health-status findings were insensitive to the inclusion of the attrition cases.

A Bayesian formulation of the canonical form of the standard regression model is used to compare various Stein-type estimators and the ridge estimator of regression coefficients, A particular (“constant prior”) Stein-type estimator having the same pattern of shrinkage as the ridge estimator is recommended for use.

We analyze the claims database of a large malpractice insurer covering more than 8,000 physicians and 9,300 claims. Applying empirical Bayes methods in a regression setting, we construct a predictor of each physician's underlying propensity to incur malpractice claims. Our explanatory factors are physician demographics (age, sex, specialty, training) and physician practice pattern characteristics (practice setting, procedures performed, practice intensity, special risk factors, and characteristics of hospital(s) on staff of). We divide physicians into medical and surgical/ancillary specialty categories and fit separate models to each. In the surgical/ancillary specialty group, physician characteristics can effectively distinguish between more and less claims‐prone physicians. Physician characteristics have somewhat less predictive power in the medical specialty group. As measured by predictive information, physician characteristics are superior to 10 years of claims history. Insofar as medical malpractice claims can be thought of as extreme indicators of poor‐quality care, this finding suggests that easily gathered physician characteristics can be helpful in designing targeted quality of care improvement policies.

"Counting on the Census? Race, Group Identity, and the Evasion of Politics." Journal of the American Statistical Association, 96(453), pp. 341–342

This comment, offered by one well-positioned respondent to the study on which this article is based, reflects the widespread belief that alternative dispute resolution methods, particularly mandatory binding arbitration agreements, have become the rule in health care delivery.' This apparent trend has spurred vigorous debate about the merits of using such agreements. Our study is an effort to ascertain how widespread mandatory arbitration agreements between health plans and providers and their enrollees and patients really are, to assess how decisions regarding their use are made, and to evaluate the prospects of their future use. We found, contrary to popular belief, that arbitration agreements are not used widely in the medical setting, and, where they are used, it is typically because organizational policy explicitly directs their use.

A simulation model that estimates individual health care spending as a function of the structure of indemnity-type insurance plans is presented. The behavioral models that form the basis for this work were developed as part of RAND's Health Insurance Experiment, HIE, a randomized clinical trial. The randomized design and statistical methods provided estimates of the effects of insurance on use, uncontaminated by sickness or selection effects. The demand for medical care was modelled using episodes of treatment. Within the simulation, episodes occur independently and randomly through time according to a Poisson process with rates depending on individual characteristics and insurance. Empirical results from the HIE indicate that insurance primarily affects individual decisions to seek treatment episode frequency, but has only minimal effects on episode costs. The response to changes in price insurance is modelled as a Bernoulli censoring process on episode frequency. The model is used to address issues on the effective design of insurance plans.

Merit rating is not widely used in setting medical malpractice insurance premiums. A statistical analysis of two different data sets shows that actual malpractice claims experience is inconsistent with the notion that claims occur randomly among physicians within each specialty class. Consequently, a statistical model allowing physician specific claims propensities is fit to a recent data set. Calculations using this model indicate that the additional effect of four years of a physician's claims experience on his or her expected claims rate is comparable to the effect of knowing the physician's medical specialty. Consequently, merit rating deserves more serious attention in medical malpractice insurance. Merit rating is commonly used in many types of insurance (e.g., automobile insurance). By contrast, medical malpractice insurers do not. Premiums typically vary by medical specialty class and by large geographical region frequently by state. Recently there have been some physicianowned insurers that have used underwriting judgments based on a physician's individual experience in deciding whether to offer that physician insurance or to assess a surcharge, but this is the exception rather then the rule among medical malpractice insurers. The conventional view is summarized by the senior actuarial officer of a leading medical malpractice liability insurer who says there is no possibility of experience rating in the near future, because there is such a large chance element in malpractice. Most valid claims are produced by careful, responsible doctors who just make a mistake. [1O, p. 82]. The opposing view is expressed by the president of a doctor-owned mutual liability insurer who says we're now cancelling some doctors who have track records. We also charge some four and five times the normal premiums.' [4, p. 236]. If malpractice results from a bad apple problem, not using claims experience in setting premiums means that good doctors have to pay for the errors of John E. Rolph is a Senior Statistician with the Rand Corporation and the Institute of Civil Justice. He has a Ph.D. in statistics from the University of California, Berkley. Dr. Rolph has served on the faculty at Columbia University, the University of London, UCLA, and the Rand Graduate Institute. The author thanks Patricia Munch Danzon and William Schwartz, for interesting him in medical malpractice insurance and for helpful information and suggestions on an early draft of this paper. Comments by Edward Ignall, Emmett Keeler, and Charles Phelps are also gratefully acknowledged as is computing assistance from Bryant Mori. This work was supported by a grant from the then Department of Health, Education, and Welfare.

Using data on all applicants to U.S. medical schools in 1975, we analyzed how an applicant’s characteristics affect the probability of his admission to medical school. Specifically, separate logit regressions for minority and majority applicants are performed to estimate this probability as a function of the applicant’s academic attributes (admission test scores, grade point averages, etc.) and of his non-academic attributes (state of residence, age, etc.). The coefficients of the state of residence dummy variables in the logit equation are estimated by discriminant analysis and then modified by empirical Bayes methods to give more accurate estimates of the state of residence effects. These modified estimates show that state of residence has a much larger effect for majority applicants than for minority applicants. An exploratory regression analysis indicates that legal residents of states with high ratios of medical school places to population are more likely to be admitted to a medical school.