Testing the conjoint analysis technique for the measurement of passive use values of forest health

Abstract

This dissertation is composed of three papers each examining issues associated with the use of the conjoint technique to estimate passive use values of forest health. The first paper will test the equivalence of conjoint question formats to estimate the use and non-use values associated with programs to prevent forest pest infestations. Two response formats, ranking and rating, were applied to two independent samples. As has been done in the past, ratings were converted into implied rankings which were then compared with actual rankings. Further a ratings difference model was constructed with the dependent variable being the difference in rating between a "status quo" situation and a management change. This model was then compared with the actual and constructed rankings models. It was found that the constructed rankings were not statistically different than the actual rankings. This has implications for survey design as it is felt that the ratings task is less difficult for respondents. The second paper examines the use of panel estimators for conjoint analysis. Most conjoint studies require respondents to rate multiple product profiles, thereby generating a panel data set. To date, published conjoint analyses have not employed panel estimators. A comparison of random effects panel ordered probit and ordered probit finds improvement in significance levels on individual forest health attribute coefficients and gains in precision of confidence intervals on the marginal values of the attributes as well as gains in overall model goodness-of-fit with a random effects specification. The final paper uses conjoint analysis to explore the difference between resident and non-resident preferences for the attributes of forest management programs directed at western spruce budworm infestations in Oregon. Estimating national values for passive use attributes of forest health presents a challenge since it may be the case that residents' and non-residents' values will be different. It was found that non-residents do have different preferences for such programs than residents. Especially interesting is the fact that Oregon residents are more likely to favor management programs which enhance or protect commercial timber, while non-residents are not. Residents and non-residents are more likely to oppose a program which involves a risk of contamination from spraying if they are members of an environmental organization.