Why inconsistent Hispanic pop 2010 in ACS.

Hello Jeff, Thank you for looking into this.

I draw a sample from the one-year ACS 2006 to 2015. Key variables include perwt hispan.

Using STATA the code is:

gen byte latino=1

replace latino=2 if hispan==0

label define latino 1 “Latino” 2 “NonLatino”

label values latino latino

table latino year [fw=perwt]

I’ve attached an Excel file of the results.

Note the strange year-over and % year-over increase in Latino pop and similar decrease in NonLatino pop.

Thank you for any clarification you can provide.

Best regards,

Werner Schink

ACS US Latinos by Ye.xlsx (10.3 KB)

Thanks for sending over more detailed information. I was able to replicate the fluctuations you have observed over time in Hispanic/non-Hispanic classifications. I don’t have a clear answer to why this occurs in the data. Note that in 2008, the ACS changed the wording to the question for this variable, switching to “Hispanic origin” instead of “Hispanic”. While this likely identifies more people as “Hispanic” than before, it is curious that these changes don’t fully materialize until 2010. Also, I did verify that the data found on IPUMS USA does match (or come within the margin of error) to official statistics reported by the US Census Bureau. So, while these fluctuations could be caused by the changing of question wording, I’m not able to rule out the possibility something else may be at play here as well. You might find this Census Bureau study on “Factors that Influence Change in Hispanic Identification” informative.

I hope this helps.