Multilevel data with correlation within subjects

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

I have a dataset with 1206 deputies from two different chambers (1998 and 2002, respectively). In addition, there are 18 parties, and some deputies are in both chambers (the ones who were reelected). I am interested in the relation between party discipline (in a scale from 0 to 100) and the probability of geting appointed as party leader (0 or 1), but I am also interested in separate estimates for each party.

A classmate told me that the following mixed effect model would be appropriate, but I don't know what to do with the correlation within subjects:

model <- glmer(leader ~ discipline + (1 + discipline|chamber:party),                data, family=binomial)

What is the structure of these data and which model would be appropriate in this case (using R)?

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏0 回帖

关键词：correlation Multilevel relation subjects Subject discipline interested different appointed following

Gelman and Hill (2007) say that there is nothing wrong with having groups with few individuals (even when there is only one). But I am not sure whether I can add an intercept for the deputies or not.