I am working on calculating approximated MOE's for summed estimates in ACS data. I understand that when at least one estimate 0, the MOE calculation should use only the largest MOE value for the estimates that equal zero (slide 26, https://www.census.gov/content/dam/Census/programs-surveys/acs/guidance/training-presentations/2016_MOE_Slides_01.pdf). Does anyone have experience with programming this rule into SAS (or Stata)?
Thoughts/guidance would be most appreciated.
In reply to Stas Kolenikov: