topic Re: Proc Mixed nagative variance in Statistical Procedures

Proc Mixed nagative variance

User29 — Mon, 05 Oct 2026 00:39:42 GMT

Dear community,

I am using METHOD=TYPE3 in PROC MIXED to retain negative variance-component estimates following Littel, Stroup, Dixon. However, I obtained a negative F value. I am new here, could you please help me? I know the nobound option is an alternative, but I would like to work with type3 method.

Re: Proc Mixed nagative variance

jiltao — Mon, 05 Oct 2026 14:36:26 GMT

Do you get a negative variance estimate or a negative F statistic? A negative F value is not possible -- please send us the output for that.

If it is a negative variance estimate, that can happen. These negative estimates arise for a variety of reasons, such as the following:

The variability in your data might be large enough to produce a negative estimate, even though the true value of the variance component is positive.
Your data might contain outliers.
A different model for interpreting your data might be more appropriate. Under some statistical models for variance component analysis, negative estimates are an indication that observations in your data are negatively correlated.

Thanks,

Jill

Re: Proc Mixed nagative variance

User29 — Mon, 05 Oct 2026 15:45:54 GMT

Code:

PROC MIXED data=mydata method=type3;

class rep A B C;

model y = A B A*B C C*A C*B C*A*B / ddfm=kr2;

random rep rep*A rep*B rep*A*B;

RUN;

PROC MIXED yielded this error term for my rep: MS(rep*A) + MS(rep*B) - MS(rep*A*B).

Note the red minus.

Re: Proc Mixed nagative variance

jiltao — Mon, 05 Oct 2026 17:18:18 GMT

That is expected -- the error term for the REP effect turns out to be the sum of two MS's, subtract the MS of rep*A*B. Math works out that way.

Re: Proc Mixed nagative variance

User29 — Mon, 05 Oct 2026 18:43:21 GMT

This happens with only one of my variables that ranges from 5 to 33 g.