It's been said that it's a television ratings disaster when the World Series comes down to two teams that are from smaller markets, and/or closer together. That is, the thinking goes that if St. Louis plays Kansas City, it's a Midwest series and people outside MO and KS don't care, and consequently a ratings disaster. Yet somehow this same concern isn't expressed for the Superbowl, which always seems curious.
I'd never seen actual data on this. So, I got data for all Superbowls through 2013, and all World Series from the same period (starting in 1968).
The inputs I looked at (i.e. things that could influence viewership) were as follows: 1) total combined home market size of each team in the championship (i.e. greater metro area)[a][b], 2) driving distance between the two teams, and 3) time zone distance between the two teams. The logic for using home market size is obvious. The idea behind using distance is that the farther apart the two teams, the greater ratings might be because the broader the appeal of the game (i.e. maybe people in L.A. don't care about Seattle, but if Seattle is playing some East Coast team, well that's different, and ratings would be better; vs. if the championship were between San Diego and Los Angeles, maybe people in the Northeast wouldn't care, and ratings would be worse.) I also looked at time zone distance for this same reason.
My outputs were: 1) absolute viewership, and 2) viewership as a percentaqge of national population. Because the average audience can change over time, I also tried looking at each year in comparison to the 11-year moving average that bracketed the year (average calculated from 5 preceding and 5 succeeding years). I then looked at scatter plots of the data.
The answer is that there is no relationship. That is, the television ratings of both the World Series and the NFL are NOT influenced by home market size or distance of the competing markets from each other. So if the A's and the Giants play each other, fine! (Not counting any earthquakes that might be induced thereby.) So it turns out that demography is NOT destiny, at least not in football and baseball ratings. The highest R^2 for a linear trend anywhere in these comparisons was a worthless 0.08; there's no point in showing you some uninformative crappy scatterplots.
Just for grins I also looked at ratings against the Excitement Index calculated for the playoffs since 2001 (were the games snoozefests or edge-of-your-seat games); if there were exciting playoffs, there might be better ratings. Again, no relationship.
It's clear from the graphs below (total viewership over time, then % of US population viewing over time) that even if baseball isn't more sensitive to geography, it has other issues.
[a]In cities which have two teams I ran two versions, one which assumed that everyone in the home market would follow their home team when it's in the championship, and another that assumed only half the sports population would support each team.
[b]It's clear that there are some strange geographic distributions of fans around the country, but for simplicity's sake I just used metro area population to get a number for home market.