how to get a single P-value across exposure categories when using multiple imputation (MICE)?

[Danah Abdul]

To get a P-value across all categories of a particular exposure I use the likelihood ratio test comparing the full model to the null model, and if the P-value is significant I can state that there is a statistical significance across all exposure categories. When *not* using multiple imputation (complete case analysis) I would usually type that:

clogit death i.ethnicity, group(matchedid) or
est store Danah
clogit death if ethnicity !=., group(matchedid) or
lrtest Danah

However, I have been trying the following after employing MICE for missing data but I get error messages.

mi estimate, or: clogit death ethnicity, group(matchedid)
est store Danah
mi estimate, or: clogit death if ethnicity !=. , group(matchedid)
lrtest Danah

when typing this command:

mi estimate, or: clogit death if ethnicity !=. , group(matchedid)

I get:

e(V) must be saved in all estimation results

Can anyone help. How can I get a single P-value across categories while using multiple imputation?

[Rich Goldstein]

I don't know of any way to get a likelihood ratio test (remember that MI estimate is NOT an MLE); however, tech support sent the following to me when I asked about a Wald test:

 There are currently two ways of performing omnibus tests for a set of categories on -mi- data.

One way is manually listing the categories like the following example:  
webuse mhouses1993s30
mi estimate: regress price tax sqft age i.nfeatures
mi test 1.nfeatures 2.nfeatures 3.nfeatures 4.nfeatures 5.nfeatures 6.nfeatures 7.nfeatures 8.nfeatures    

The other way is using the -fvexpand- command (see -help fvexpand-); for example:  
fvexpand i.nfeatures
mi test `r(varlist)'

[Danah Abdul]

Rich Goldstein Both suggestions worked perfectly well. Many thanks and appreciation, that saved me!

Kindest regards
Danah