Question about design factor for replicate weights in ACS


I’m using the SAS code posted on the IPUMS website for incorporating replicate weights:…

Does the design factor include assigning a value of 2 for very large or very small estimates, as indicated on the ACS documentation?…

If not, what is the SAS code that would address this?



After looking through the documentation it looks like the design factor method for generating standard errors (detailed here) is slightly different from the replicate weights method (detailed here). This being the case, I am not familiar with available SAS code for implementing the design factor method. However, there is fairly detailed documentation for calculating standard errors with design factors in the Census Bureau document you’ve linked above.