Dealing with Count Data

Instapundit mentions some data about military recruitment that Donald Sensing has also discussed. The topic is Marine recruiting numbers coming from Northern California and Nashville. This is what Rev. Sensing had to say:

Just in case you missed it, that means that the single city of Nashville, population 1,187,521 in 2000, sends more men and women to the Marine Corps than half the entire state of California, population 34,501,130 (for the whole state; how many live in the northern half I don't know). And the Chronicle is apparently worried that the place is becoming too pro-military. Yeah, right.

Professor Reynolds misspeaks a little when he describes these numbers as recruiting rates. The rankings under discussion are based on counts of recruits. To fairly compare the "fertility" for recruiting, we should divide the counts by some appropriate area of opportunity. Sensing mentions total population, but an even better area of opportunity would be population in the appropriate age range.


The world of management is shot through with this kind of error in data analysis. If you look around, you will probably find many examples.