st: svyreg with missing values in predictors or outcomes

[Takahiro Higashi <taka@ucla.edu>]

I understand that if predictor variables or outcome variable of a svyreg 
model have missing values, STATA will drop the observations from the model 
estimation. (calculation of beta coefficients)
How does it affect the variance estimation? Can this cause the singleton 
PSU problem? If so, should I fill in the missing values with some arbitrary 
values and also set the weights for the observations to zero or use 
subpop() option after creating a new variable indicating which observations 
have no missing values.
Thanks.

Takahiro Higashi
UCLA Div. of General Internal Medicine

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/