Complex survey settings

[Sara Suzuki]

Hi,

I am trying to set-up a complex survey data set in Stata and Mplus so that they have the same settings.
I am having trouble getting them to match up.
I am not getting the same standard errors for simple descriptive statistics, suggesting that the survey structure is not being accounted for in the same way across Stata and Mplus.

In Stata I am entering:
svyset IDSCHOOL [pweight=TOTWGT], fpc(TOTALSCHOOLSPERSTRATUM) strata(IDSTRAT) singleunit(scaled) || IDCLASS

In Mplus I am entering:

VARIABLE:
WEIGHT = TOTWGT;
CLUSTER = IDSCHOOL IDCLASS;
STRATIFICATION = IDSTRAT;
FINITE = TOTALSCHOOLSPERSTRATUM (POPULATION);

ANALYSIS:
TYPE = TWOLEVEL COMPLEX;

Does anyone have experience with this and know how I can set the survey structure to be equivalent?

Note: I set singleunit(scaled) because in Mplus the "THE VARIANCE CONTRIBUTION FROM A STRATUM WITH A SINGLE CLUSTER (PSU) IS BASED ON THE DIFFERENCE BETWEEN THE SINGLE CLUSTER VALUE AND THE OVERALL CLUSTER MEAN."

[Steve Samuels]

You need singleunit(centered) to match the Mplus description. Note that the lack of an fpc for IDCLASS in the svyset statement implies that sampling of classes was with replacement.