In this paper we use conjoint analysis to predict participation intent
ions for an optional energy service offered by Niagara Mohawk Power Co
rporation (NMPC). In contrast to conjoint studies that use ordinary le
ast squares regression to estimate a utility or preference function, w
e use an ordered probit model. For validation, we compare predicted pa
rticipation from the probit model to actual behavior. We also compare
predictions from Sawtooth Software's Adaptive Conjoint Analysis (ACA)
software to actual behaviour. The predictions obtained from the ordere
d probit model are closer to actual behaviour than the predictions obt
ained from the linear model. We discuss possible reasons and implicati
ons for future research. (C) 1998 Elsevier Science B.V. All rights res
erved.