topic Proc MiAnalyze "Warning: between Between-imputation variance is zero for the effect intercept" in Statistical Procedures

Proc MiAnalyze "Warning: between Between-imputation variance is zero for the effect intercept"

Peter_Hong — Tue, 28 Feb 2023 16:18:08 GMT

Hi all,

I'm running a proc mianalyze after I have completed monotone imputation.

class id time (ref = '1')intervention; 
	*class id time intervention;
	model phy_100 = time intervention time*intervention /ddfm = betwithin s; 
	random intercept/type = un subject=id g gcorr; 
	lsmeans time intervention time*intervention /pdiff;
	by _imputation_;
	ods output solutionF = aaa; 
run;
proc mianalyze parms = aaa; 
	class time intervention ;
	modeleffects intercept time intervention time*intervention ; 
	ods output ParameterEstimates = parme; 
run

the default output has my followup time and intervention group as references, fo I want to specify ref groups, but once I've done that I get the following warning messages.

WARNING: Between-imputation variance is zero for the effect intercept.
WARNING: Between-imputation variance is zero for the effect intervention.

I tried to Google the answers and seems that "no variability in the effect of intervention across the imputed datasets", but I don't understand how changing the reference group would lead to such an issue, and if there's a solution to this issue? Please advise, thank you.

Re: Proc MiAnalyze "Warning: between Between-imputation variance is zero for the effect interce

SAS_Rob — Tue, 28 Feb 2023 16:19:33 GMT

The message you are receiving can have a number of possible causes. The first is that there is some problem with the imputation model itself (I have seen where people have left the response variable off the VAR statement in Proc MI as an example). In general though, it occurs when the missing data has no influence on the sampling error of a parameter estimate. There is no fix or adjustment for that, but it does require some further investigation on your part. Usually, it is an indication of a problem with the imputation model with possible causes ranging from a poor imputation model to no real need to impute due to a very small fraction of missing information.

Without fixing the problem, when the between imputation variance is zero then the number of Degrees of Freedom is undefined so you cannot get a confidence interval or p-values. If you are unable to determine the cause then one suggestion would be that, if there are only minimum differences among these within-imputation variances, then look at the results for that variable for one of the imputations.