SoccerPop9194: Thanks for the opportunity to discuss this. There are several potential fundamental errors in your analysis:
1) According to US Census methodology, Greenville, Spartanburg, and Columbia are three separate MSA's. Aggregating two MSA's and comparing against a third MSA is nonsensical -- in census terms what you are doing is comparing a CSA (Combined Statistical Area) against an MSA (Metropolitan Statistical Area.) For more information on why an MSA is the appropriate unit of measure, please see the sources cited -- but the bottom line is what you note when you question your own results concerning apples-to-apples county comparisons. [Source: US Census Bureau, Metropolitan and Micropolitan Statistical Areas; Specific Source of MSA's:
http://www.census.gov/population/estimates/metro_general/List1.txt, December 2005]
2) You need more rigor on your team counts; you're off on all of them. For example, the Columbia MSA has 6 SCYSA teams: CSC, LCSC, NECSA, CRSA, Lower Lexington, LCDOR. All of these fall within Lexington and Richland Counties. [Source: SCYSA Web Site]
3) Breaking the population down into a 5-17 age demographic doesn't have the effect you cite. You are creating an effect by mixing county data and what may be CSA/MSA data (although it's difficult to tell.) For example, here are the 5-17 demographic for selected counties:
Greenville County: 5-17 [2000]: 17.8% 379,616 67,572
Lexington County: 5-17 [2000]: 19.3% 216,014 41,691
Richland County: 5-17 [2000]: 17.9% 320,677 57,402
Charleston COunty: 5-17 [2000]: 17.3% 309,969 53,625
[Source: US Census Bureau, 2000 Demographics]
As you can tell, Richland and Lexington actually have a higher percentage of people aged 5-18 than Greenville or Charleston. So trying to explain away gross population counts by demographics is actually a losing proposition for you. But the real trouble is that you can't just pick any county or combination of counties that you want (see point #1), you have to pick equivalent census-oriented entities. If you don't, the results are subject to incredible self-selection bias based on "stacking the data" by county/area selection. Again, see the MSA definition and history for the reason that the MSA is the appropriate unit.