nested logit估计中的数据问题

nested logit中z变量是哪个？数据例子是第一个个体，对应4个选择，有两个分支，那么使用最大似然估计时，x_j|l是第l个分支中的第j行。那么对应的z是多少？比如格林第七版769页。 attributes of the choice sets  z_ib是什么？请以他的例子说明。

If you want to learn Nested Logit model, maybe you should try to use Stata manual.
There are very detailed explanations about the theortiecal background of Nested Logit model as well as an emprical dataset to help you to realize the estimation of nested logit model.
See help nlogit.
nlogit depvar [indepvars] [if] [in] [weight] [|| lev1_equation [|| lev2_equation ...]] || altvar:  [byaltvarlist] case(varname), [options]

stata中的命令能够用来做了，但还是不太明白这里面的含义，所以想问一下z变量的含义。