In Phaneuf et al. (2013), the authors use a discrete choice experiment to measure the value of remediating the Lower Buffalo River in Buffalo, New York, which was classified as an Area of Concern (AOC) due to past contamination. A sample of recent homebuyers completed choice tasks in which their actual purchase was compared to an alternative home that differed in the following attributes (attribute levels):

home size (+25%, +15%, no change, −15%);
AOC condition (full cleanup, partial cleanup, no change, more pollution);
distance to AOC (2 miles closer, 1 mile closer, no change, 2 miles further);
home price (+30%, +15%, no change, −10%).

The full data set ptb_dce is available for download on the website, along with variable definitions. The data are based on a sample of 281 people, each of whom completed eight choice tasks that compared their actual home to one designed alternative.

(a) Using these data, estimate a main effects model using log transformations for the home size and distance to AOC variables, and individual dummy variables for the environmental conditions. Make sure you also include an intercept term to distinguish the real from the designed alternative. What is the marginal willingness to pay in home purchase price for one additional square foot of home size, for a representative 1,500 square foot structure? What is the marginal willingness to pay for an additional 0.10 miles distance from the AOC, for a representative home 1.5 miles away from the river? Predict the average willingness to pay in home purchase price for remediating the river from its current state to full cleanup.

(b) Repeat the exercise in (a), but now include interactions between the distance to AOC and environmental condition attributes, as well as the main effects.
Compare your estimates and willingness to pay predictions from this to a specification that excludes the main effects for the environmental conditions (while maintaining the interactions with the distance to the AOC).

(c) Return to the main effects specification in (a). Examine the extent to which preferences for the distance and environmental condition attributes depend on household characteristics, such as income, presence of children, and age of respondent.