xtmixed produces different results from mixed

[Philip Leissner]

I am planning to run a mixed effects model and I am trying the various xtreg, xtmixed and mixed options.

xtmixed and xtreg are producing similar results, as they should do. However, the mixed command gives very different results for the random-effects parameters. Why is this? Previous posts on the topic states that xtmixed and mixed are interchangeable. And if they were different, I would expect mixed to atleast produce similar results to xtreg?

I run the commands:
xtset ExternalReference OP
xtreg phq_tot_T AAQ Age Sex Education MaritalStatus Diabetes DS14 Psychmed, re
mixed phq_tot_T AAQ Age Sex Education MaritalStatus Diabetes DS14 Psychmed || ExternalReference:
xtmixed phq_tot_T AAQ Age Sex Education MaritalStatus Diabetes DS14 Psychmed || ExternalReference:

(OP is the Observation Point variable)

xtreg gives:
sigma_u | 2.1066733
sigma_e | 2.4060045
rho | .43395939 (fraction of variance due to u_i)

The random-effects parameter estimates for xtmixed are:
sd(_cons) | 2.120249
sd(Residual) | 2.391614

The random-effects parameter estimates for mixed are:
var(_cons) | 4.495457
var(Residual) | 5.719819

Thank you

[Joseph Coveney]

The first two are reporting standard deviations and the last is reporting variance. Try the stddeviations of mixed.

[Joseph Coveney]

Try the stddeviations of mixed. → Try the stddeviations option of mixed.

[Philip Leissner]

Thank you very much Joseph, that resolved the issue