Key points are not available for this paper at this time.
Participants' understanding of the treatment in social science experiments is raised as an issue in the specification of statistical models of policy evaluation. A statistical model is offered which shows that a failure to consider participants' understanding may intro duce biases of unknown direction into policy parametric estimates. Data from the New Jersey-Pennsylvania Negative Income Tax Experiment show that the negative income tax treatment was not well understood by the experimental participants, that under standing was correlated with experience and the treatment parameters themselves, and that the inclusion of "knowledge" variables in the labor supply equations used in the formal evaluation of the experiment changes some estimates of work disincentives.
Nicholson et al. (Sun,) studied this question.