Somali Population in MN. Why such a difference between U.S. Census and Current Pop. Survey?


I have seen the 2011 data for the Current Population Survey identifying Somali populaion in MN to be at 124,768. The U.S. census for 2010 is about 32,000 in MN and the American Community Sample for 2011 is at 33,887 in MN.

While it is commonly assumed that the Census is substantiallly undercounting because they do not use ethinicity other than simply race (African American, Asian Pacific Islander, other, etc.) it is shocking to see the 400% increase. How can we explain this dramatic variance? I understand that the CPS uses smaller sample size but how can this explain such a dramatic increase in population? Can I trust the CPS result?

Is there someone I can talk to so to better understand this?

Michael VanKeulen



The only way to identify the Somali population using IPUMS-CPS is with Birth Place (BPL) and Parental Birth Place (FBPL and MBPL), and none of these variables are used in the weighting procedure for CPS. Further more, the sample size of CPS is too small for this kind of analysis and I would not trust numbers generated about such a small, specific population. However, just using FBPL in the March 2011 sample to perform a quick check, I only found a weighted total of 67,183 persons in MN reporting Somalia as their father’s place of birth (un-weighted, this is only 93 respondents). The CPS numbers may not be perfect, but as I am not sure how the 124,000 figure was calculated, I cannot speak to its accuracy. I would recommend contacting the source of that figure.

I hope this helps.