Hi, I have a question
I am trying to compare changes in the Puerto Rican population in metro areas from 2000 to 2015. Are the 5% ACS 2000 and 1% ACS 2015 datasets comparable?
In other words, if the proportion of PR in let’s say Miami in 2000 was 1.5 and the proportion of PR in Miami in the 1% ACS 2015 is 2.5, can I interpret this a one percent increase? I am concerned about the accuracy of this kind on interpretation because the sample sizes different (5% vs 1%)
I am using the 2013 metroarea variable to account for changes in the boundaries between the two datasets.
Any information would be very helpful.
Cheers,
Dr. Ali R. Chaudhary
As long as you are using the appropriate sample weights (in this case PERWT) when calculating your estimations, then the different details of sampling shouldn’t make much of a difference in your estimates. The one worry is with the changing definitions of the boundaries of metropolitan areas over time. Over time, and particularly over a 15 year time period, the boundaries that define metropolitan areas can change considerably. So, one caution in performing the sort of analysis you describe above is recognizing that both demographic characteristics of areas and the geographic area defined as a given metropolitan area can change. It is difficult to disentangle these two changes.
Following up on this answer: Since your analysis only goes back to 2000, you should be able to use the MET2013 variable. This variable has the benefit of using the 2013 definitions for metropolitan areas for all years. Therefore, using this variable should allow you to make reasonable estimates of changes in demographic features of metropolitan areas over time.