by Kyle Watson
Hello blog readers,
This is a story about "which side of the tracks" you are on and how demographic analysis comes into play. More importantly how to identify certain anomalies when one side is completely opposite of the other - and the sum doesn't equal its parts.
To illustrate this point let's say we are a national franchising organization. Hundreds of new applications come in a month each with three or four proposed sites. Our commercial real estate group then gives us analysis, maps, and reports of newly proposed franchises. We spec out the sites, they come back with the goods. They hand us a "PASS/FAIL" report that tells us if we've met our initial criteria.
The initial criteria is always:
- A 7-mile radius
- Population of 250k - 350k
- Average Household Income of $120k (+/- 20%)
So this is what we get back in a nice report. If the three variables are not met - FAIL. And we move on to different proposed sites. If the criteria is met - PASS. We further breakdown the area and continue with the suitability process.
Below is an example PASS/FAIL map (of which I annotated to show the anomalies). At first glance all criteria is a match, right?
WRONG!
We take a more in-depth and determine that even though the average household income of the 7-mile ring is $121k, there aren't any block groups in the surrounding area that are actually in the $121k range. Here we are dealing with a real life situation where the north side of the proposed location (the working class Pontiac, MI area) and the south side (one of the most affluent areas in the U.S., Bloomfield Hills, MI) border each other back-to-back. Polar opposites.
The extreme differences in demographic makeup cause an anomaly, thus we can't consider the 7-mile ring a uniformly, normalized area. In this case the medians and averages for the ring don't tell us the whole story.
This concept is often related to ecological fallacy. Take, for example the linear graphic below. This is a representative situation of the map above. The average and median are not representative of any data considered.
Quite often when averages are skewed, medians are the next in line for analysis. In this case neither help. You could split the regions into a north and south trade area (which I did using the Draw Area tool in Business Analyst desktop), but this only reaffirms that one side is too low and the other is too high. The common practice remaining is to expand your PASS/FAIL criteria. You may need to add a second tier of demographic qualifiers (competitors, age, race, market saturation) before making a final decision.
I hope the above reaffirms the notion that unless you do analysis in an evenly distributed Utopian society, make sure you understand the data. And add in some secondary checks to ensure you make the right decision.
Cheers,
Kyle - born in a Pontiac - Watson
