(This continues question 10.) The assignment variable Xi is defined as follows: Xi = 1 if i is in treatment, else Xi = 0. The probit model says that given the assignments, subjects are independent, the probability of success for subject i being #(Xiα + wiβ), where # is the standard normal distribution function and wi is a vector of personal characteristics for subject i.

(a) Would randomization justify the probit model?

(b) The logit model replaces # by /(x) = ex /(1 + ex ). Would randomization justify the logit model?

(c) Can you analyze the data without probits, logits,...?
Explain briefly. Hint: see discussion questions 9–10 in chapter 6.