Use the data in CHARITY.RAW to answer these questions.
(i) The variable respond is a binary variable equal to one if an individual responded with a donation to the most recent request. The database consists only of people who have responded at least once in the past. What fraction of people responded most recently?
(ii) Estimate a probit model for respond, using resplast, weekslast, propresp, mails- year, and avggift as explanatory variables. Which of the explanatory variables is statistically significant?
(iii) Find the average partial effect for mailsyear and compare it with the coefficient from a linear probability model.
(iv) Using the same explanatory variables, estimate a Tobit model for gift, the amount of the most recent gift (in Dutch guilders). Now which explanatory variable is statistically significant?
(v) Compare the Tobit APE for mailsyear with that from a linear regression. Are they similar?
(vi) Are the estimates from parts (ii) and (iv) entirely compatible with at Tobit model? Explain.