关于PROC MIXED 中class的疑惑

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

在论坛中看到如下的对重复测量数据分析的过程（原帖地址https://bbs.pinggu.org/thread-998488-1-1.html），class中加入和不加入time有什么区别？是否应该加入time？谢谢！

Proc mixed data=a_b covtest method =REML;
  CLASS group pnt;
  Model y=group time group*time time*time time*time*group /S;
  Repeated /type= TOEP subject=pnt R;
  Lsmeans group / tdiff pdiff;
run;

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：proc mixed Mixed mix ROC Repeated method 测量

放不放在class语句里就看这个变量是不是分类变量。如果属于分类变量就要放进class语句。

楼上说得没错，如果模型里的time是continuous variable，那么class语句中不应加time；相反，如果模型里的time是discrete variable，那么class语句中要加time。在time只取0/1的情况下，无论是哪种类型的变量，估计出的结果是相同的。
In particular, if time is treated as a discrete variable, you should remove time*time and time*time*group from the model specification presented above, since these two variables will be collinear with time and time*group, respectively.

mannlee 发表于 2014-6-10 08:58
楼上说得没错，如果模型里的time是continuous variable，那么class语句中不应加time；相反，如果模型里的ti ...

更正一下，用quantity variable versus categorical variable 来区分比较好，请忽略上面 continuous variable vs. discrete variable 的划分。

I think continuous is just contrast to discrete.
In this case, clearly time was treated as continuous variable and thus should not be included in CLASS statement.
JingJu