席慧慧 发表于 2020-4-13 22:07 
你描述的不是很详细。显示你分组了?猜测自变量有缺失值还是因变量只有一个值?
我控制了企业和年份的固定效应进行logit回归 xtset id year
sum 因变量是:
amount_whe~r | 19,206 .1315214 .3379785 0 1
但回归结果中观测值只有六千
xi:xtlogt y x i.year,fe结果如下
note: _I籍贯_21 omitted because of collinearity
note: multiple positive outcomes within groups encountered.
note: 1,630 groups (12,447 obs) dropped because of all positive or
all negative outcomes.
note: _I籍贯_11 omitted because of no within-group variance.
Conditional fixed-effects logistic regression Number of obs = 6,678
Group variable: firmid Number of groups = 616
Obs per group:
min = 2
avg = 10.8
max = 13
LR chi2(49) = 651.60
Log likelihood = -2597.3896 Prob > chi2 = 0.0000
请问下怎么样解决固定效应logit回归中大量观测值损失呢