1
$\begingroup$

This might be trivial, but I'm used to HLM7 software output and now I'm switching to Stata (xtmixed).

To give an example imagine I have students (level-1) nested within schools (level-2). Running an empty model, in HLM, I can easily see the variance component associated to each level, to see how much variation is at level-1 and how much is at level-2. Starting from that, I also calculate the intra-class correlation coefficient.

Now I see the variance components, in the xtmixed output, these are reported as the standard deviation estimates of the intercept sd(_cons) and of the residuals sd(Residual). How do I calculate the associated p-value to see if there is significance?

$\endgroup$
3
  • $\begingroup$ Well, this is probably because you shouldn't look at these p-values, Stata knows that and this is why doesn't show anything there. Look at this discussion statalist.1588530.n2.nabble.com/… $\endgroup$
    – Steve
    Aug 26, 2014 at 12:15
  • $\begingroup$ Thanks. Can you better explain here also so we have an answer here? Moreover, how do I calculate the p-value even if not totally accepted as valid? $\endgroup$ Sep 1, 2014 at 14:24
  • $\begingroup$ Your thoughts on that issue have been discussed in CV before. For your specific question I couldn't give an answer. Look the answers in this post stats.stackexchange.com/questions/4858/… $\endgroup$
    – Steve
    Sep 1, 2014 at 16:29

1 Answer 1

1
$\begingroup$

Concerning the display of the results, specify the option variance if you prefer variances over standard deviations.

Concerning the significance, you can run an OLS of the dependent variable on all independent variables with exception of the level 2 identifier (i.e. schools), using the command regress. Store the estimates you obtain through estimates store [name1].

Then estimate your multilevel model using xtmixed and again store the estimates by estimates store [name2].

The difference between these models is the random intercept you allowed in the multilevel estimation but not in the OLS estimation; hence testing whether the unconstrained model performs better is equivalent to testing significance of the random intercept. lrtest [name1] [name2], force will do this for you. You will need to specify the force option; otherwise Stata deems the test invalid.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.