coefficients and confidence intervals for discrete values of random effects variable

[Hillel Alpert]

Could anyone please advise how to generate the coefficients and confidence intervals for the discrete values of a random effects variable?

I am generating predicted coefficients and variances (blups) for these by following a svy: melogit model with the command:

.predict coefficient, reffects reses(variance)

Is that correct? How are the confidence intervals derived if they are not directly available in the output, please?

Hillel

[Clyde Schechter]

The model in -melogit- is that the random effects are normally distributed. So normal theory applies and the 95% CI would be the estimated value +/- 1.96 * the standard error.

By the way, I wouldn't use the name variance for the standard error as the two terms obviously mean very different things.

[Hillel Alpert]

Thank you.