2010 Census 5% Sample

I see the 5% sample for 2000 and every 10 years before that but not for 2010. Was this because 2010 5% sample was canceled or something?

Thanks In Advance!

The 2010 sample that is currently available on the IPUMS-USA website represents the American Community Survey PUMS data. Unfortunately, we are not sure when the Census Bureau plans on releasing the 2010 census 5% PUMS files, so we do not have a release schedule for the corresponding IPUMS-USA file.

I hope this helps.

I have an additional question to the above one: Will there be a smaller sub-sample of the 2010 census with more variables? I am mainly interested in educational outcomes, work and income. I heard such variables from the 2010 census are available on the county level, so I assume they were collected. Just to avoid confusion: I am not referring to the ACS but really to the microcensus because it is conducted on the 1st of April every year which would help me calculate birth dates. Thank you in advance!

The ACS actually took the place of the long-form Decennial survey, with 2010 being the first time the Census Bureau did not field a long-form. I checked on American FactFinder and I was not able to find any evidence that Education or income information was collected as part of the 2010 Decennial Census (except in U.S. territories where the ACS is not collected). I’m sorry I could not provide the data you are looking for.

Ok! Thanks a lot for checking!

Is the 2010 10% sample the aggregation of the ACS performed between 2001 and 2010 or is it the 2010 decennial census? Why is that sample unweighted?

The 2010 10% sample is a 1-in-10 random sample of the US population. It is a distinct sample from the ACS samples in other years, and therefore is not a combination of the ACS samples. The term “unweighted sample” refers to the samping framework used to generate this sample. That is, every individual in the US population was equally likely to be selected into this sample. Therefore the sampling weight values will be constant for all observations in this sample. This is contrasted with “weighted samples” where depending on observable characteristics different people have a different likelihood of being selected into a given sample.

Amazing. Thanks! I’m working with DC data starting from 1950. For 2010 do you recommend me to use the 2010 10% sample? or the ACS 2006-2010 5-Year sample?

The 2010 10% sample is from the 2010 decennial census. (Unfortunately, that’s not stated explicitly in the sample description on IPUMS USA, so it’s understandable why it’s not totally clear!)

The reason you’d use the 10% sample is to have the largest possible sample for your study, improving your estimates. The reason not to use the 10% sample is that the 2010 decennial census included a much smaller number of questions than the ACS, so the 10% sample provides a very limited set of variables (basically: age, sex, race/ethnicity, household size & relationships, housing tenure, and group quarters status).

If you need more variables, you’ll need an ACS sample. For ACS data representing the year 2010, you can use the 2010 ACS 1% sample. If you prefer a larger sample, the 2012 5-year sample (2008-2012) would better represent 2010 characteristics than the 2006-2010 sample.

Great. Thank you!! That’s exactly what I needed