Hi all, are there set criteria others are using to flag estimates produced using ACS files in PUMS as potentially unreliable? I am also working with National Survey of Children’s Health, which I believe is also administered by the Census Bureau, and their online data querying system flags as potentially unreliable any estimated percentage with a 95% confidence interval that is either (1) more than 20 percentage points wide, or (2) more than 1.2 times the size of the estimated percentage. IPUMS doesn’t seem to flag anything as unreliable unless I am missing some functionality. I also don’t see any set criteria around this in Census’s documentation around the PUMS files. Is there any sort of uniform criteria out there that I’ve missed? Or are people coming up with and using their own criteria?
I’m assuming you are using the IPUMS USA online data analysis system to get your estimates. You can have the system calculate standard errors and confidence intervals for you, but there is no automatic flagging of unreliable estimates. It is up to data users to determine what they consider unreliable based on the information provided by the system.