This paper describes an experiment designed to assess the ability of contingent donations to predict participation in and benefits of a voluntary environmental program. In our earlier work a theoretical model was developed to show that voluntary contributions are "theoretical lower bounds" on the Hicksian surplus measures. Our interest is in exploring how to provide robust estimates of this lower bound based on contingent donation data. Studies have consistently found contingent donations to overstate actual donations. This overstatement could be due to respondents in the contingent donation treatments not free-riding or it could be due to the hypothetical bias associated with the contingent donations, some combination of these two factors, or some other factor.

In the earlier study, we found that participants who expressed a higher level of certainty about their response to a dichotomous-choice contingent donation question in the sense of being very certain they would donate the amount they said they would, were very similar to study participants in the actual donation treatments who chose to donate. Use of this follow-up certainty question to differentiate respondents who say "yes" to a contingent donation question and would actually donate from those that say "yes" but would not really donate if given the opportunity, allows for use of contingent donation data to predict actual donations. In this research, we are replicating the approach with a different type of public program, across a wider range of offer amounts, and with a different type of payment schedule.

This experiment involves a split sample design in which one treatment group is asked to actually participate in a voluntary program to purchase electric power generated by a wind turbine and in turn displace an equivalent amount of coal generation. The other treatment group is asked in a parallel manner about hypothetically participating in such a program (the "contingent participation" treatment). A mail survey is the chosen mode of administration.

Eight focus groups were used to develop and pretest the description of the wind power program. Survey questions were also tested in the focus groups. We conducted a pilot study (N=900) to test both the survey materials and survey procedures. The final study is scheduled to be in the field in October ~997.

The data will allow us to assess the predictive validity of the contingent participation treatment. We will also estimate the social benefits of the wind power program. The measures elicited in the survey will be useful for identifying the characteristics of both participants and non-participants. We will also be able to investigate the relationships between environmental attitudes, past participation in energy programs, and socio-demographic characteristics and probability of participating in the wind power program.