Missing Standard Error for 5% 2000 estimates using Cluster/Strata

USA
1

I am attempting to calculate standard errors for the 5% 2000 PUMS data file. Specifically, I am attempting to caclulate the average household size for certain dwelling characteristics. I have collapsed the original rectangular dataset such that each observation indicates a household. For each household, I have indicated to stata to keep the first observed value for the number of person records, the strata variable, the cluster variable, and hhwt.

As indicated by the user guide, I svyset my data with the following command: svyset cluster [pweight=hhwt], strata(strata)

However, when I attempt to caclulate the mean value of NUMPREC, I receive a single point estimates with no standard errors and no 95% confidence intervals. In addition the following note is included: “Missing standard error because of stratum with single sampling unit.”

I’m not quite sure why I am getting this error. Is it something to do with the Indiana subsample perhaps being too small? Or perhaps I have erred in a command. Any help would be appreciated. Thank you.

2

I believe the issue you are seeing can be resolved by using the subpop() option for the Stata svy command as described in the “Syntax for Subpopulation Analysis” of the Analysis and Variance Estimation with the IPUMS page. You can also use the over() option for the mean command to get estimates with Std. Err. values like so:
svy: mean numprec, over(statefip)

I hope this helps.