t-test with standard error cluster

ElaineLiu — Sun, 07 Nov 2021 16:13:39 GMT

Hi everyone,

I want to do two sample ttest with standard error cluster. I googled that proc mixed would work, but I understand that it also has fixed effect. But I don't want fixed effect, I only want error cluster.

My data looks like this

fundid date var1 group

group is dummy variable.

I want to cluster at fund lever, that is fundid

my code is

proc mixed;

class fundid group;

model var1=group;

random fundid;

ismeans group/diff;

run;

Can you let me know whether it's correct? do I have fixed effect included in the test?

Thank!

topic t-test with standard error cluster in Statistical Procedures

t-test with standard error cluster