Question (a) of problem 10.2 
===========================
y_i is bernoulli. We have done likelihood for Bernoullis before. 
Check your class notes. 
So p(y_i) = product of ....? 
But in your likelihood function, substitute for what the p's are..... 
convert from logit(p) to the expression for p in terms of the regression coefficients. 

Question (b) of problem 10.2 
============================

Instead of doing what the author requests, assign a Normal prior 
distribution with mean 0 and large variance (almost 
non-informative). 
Doing what he wants requires us to assign priors to the p's first and it 
is a rather complicated and indirect way.