I am working on an analysis of some revenue figures that are given to me by a Crazy Geographic Breakdown (CGB) that is particular to this business, it doesn’t correspond to any standard geography. I need to convert this to revenue by DMA.
I do know for each of their CGBs, how their zip codes map to DMA zipcodes, but the my revenue figures are totaled for the CGB, they aren’t broken down to that DMA level. So my thought was… if I assume every zip code has roughly the same population, then it would be reasonable to assume that each zip code has about the same revenue percentage, and apportion the CGB revenue based on the number of zip codes.
That is:
CGB is composed of 10,000 zip codes. 6,500 are in DMA 1, 3,500 in DMA 2. It made $400,000. So I would allocate $260,000 (65%) to DMA 1, and $140,000 (35%) to DMA 2.
But – is this assumption true? It doesn’t need to be exactly true, but is it pretty close, close enough? I’ve been surfing the USPS and Census sites all morning, and can’t figure out if this is even roughly correct or not.
This could be a very dangerous assumption. Two adjoining zip codes may have radically different demographic profiles. To take an extreme (but not that uncommon) example, Harlem, with an average household income in the $40,000-something range, abuts the Upper East Side, with an average household income $100,000 higher.
However, you might not have a ton of choice in the matter if you can’t actually get to the zip-code-level detail in the CGB. You might do the analysis as you describe, and then put plenty of caveats on it. And try to ascertain something about the demography of the area you’re analyzing. It could be that there’s more uniformity than in my example above, in which case your approach would be safer.
I agree with Sal, this is likely to be a very inaccurate assumption.
Zip Codes are set up by the Post Office for their business, delivering mail. So they are based on the volume of mail in an area. This can sometimes match to population, but only very roughly.
Areas with a lot of renter, poor people, transient students, etc. often get much smaller volumes of mail than established middle-class home-owners, for example.
Some business or government agencies can get huge volumes of mail. In larger cities, it’s quite common for City Hall to have its very own zip code.
Also some large commercial buildings. The Empire State Building has it’s own (10001), but I don’t think it has any ‘resident population’.
And the tiny towns of Norwood/Young America, Minnesota (recently merged, creating a metropolis of over 3,100 people) have 27 zipcodes assigned, about 1 for every 114 people. (A huge coupon/rebate processing business is located in town.)
Since each zip code is served by a single post office, I kinda doubt they’d be willing to torch some perfectly serviceable 100 year-old building just because a lot of the population has moved to the 'burbs. The imbalance would probably have to get pretty large before they redraw the lines.