My organization calculates what we call the "Total Minority Population" by substracting the estimate of people that are Not Hispanic or Latino & White alone from the Total Population. I am familiar with generating MOEs for derived estimates when I aggregate data and when we created percentages and proportions, but I have not had to do this yet where I was substracting one estimate from another. My assumption is that I should use the square root of the sum of the squares rule as for aggregates, can someone please let me know if that is correct or point me to the proper formula?
An example for Glacier County, Montana from the 2019 5-yr estimates is:
Total Population: 13,732
Total Population MOE: ***** (which can be interpreted as zero for my purposes I think)
Non Hispanic or Latino and White Alone: 4,112
Not Hispanic or Latino and White Alone MOE: +/- 127
Total Minority Population (our language) = 13,732- 4,112 = 9,620
Total Minority Population MOE (my best guess) = SQRT(POWER(0,2) + POWER(127,2)) = 127
It seems like this this will often be the case for counties in the US as it appears more often than not the Total Population is a "Controlled Estimate" with a reported MOE of *****.
I am using data equivalent to that found in table DP05. Data can be found on data.census.gov at this link....
Thank you in advance for any help you all can offer.
Check me on this but I think when you see "controlled estimate" that means the result is assumed and comes from the estimates program, thus nothing to do with sampling and no MOE.
Thank you Tim, very helpful. I am beginning to understand how many datasets depend on the Population Estimates!