The evaluation of new health care technology: The labor economics of statistics

This paper interprets the market for observations as a labor market and discusses the impact of market forces on experimental designs, focusing on clinical trials in the evaluation of new health care technology. First, the impact of the supply and demand for observations on the sampling distributions of common treatment effect estimators, as well as the equilibrium sample size that the supply and demand lead to, are analyzed. Equilibrium samples imply that standard arguments about the consistency of common estimators, which assume tha populations sampled do not change with the equilibrium size of the sample, are applicable less often than commonly argued. Second, the impact that extensive public regulation of the market for observations has on sampling distributions is discussed, particularly the effects of mandatory, as opposed to voluntary, randomization. It is argued that such regulations may be interpreted as ovecoming an econometric moral hazard problem in the production of treatment effect estimates by producers who do not have the truth-preferring tastes implicity assumed throughout normative statistical analysis of experimental design.