Convergent validity of alternative dependent variable specifications for individual travel cost models

Abstract

Applications of individual observation travel cost models have employed two alternative dependent variable specifications, (trips) and (person-trips), defined as (trips*groupsize). For 58 National Park Service data sets, willingness to pay (WTP) was estimated using both the trips and person-trips construction. Significant differences were found in pairwise comparisons of the alternative WTP estimates in 29 of 58 cases. For a subset of 31 data sets where statistically significant travel cost parameters could be estimated under both dependent variable specifications, 23 of 31 models showed statistically significant differences in WTP between the two models. In all 23 cases of a significant difference in WTP, the specification using person-trips as the dependent variable was greater than the cases using trips as the dependent variable. Additional analysis showed that lack of dispersion in the trips variable, as measured by the percent of visitors reporting taking only one trip to the site, was positively correlated with the differences in the WTP estimates.