Before we go any farther about TV ratings, I think it's important to discuss how they're created.
Take the Nielsen ratings for example. In the U.S., there's about 80 million households (probably more) - however, from what I can gather Nielsen only gathers data on 25,000 households (5,000 national, 20,000 local). The local users are allocated among the largest 46 markets - so if they're spread uniformly (likely not the case), each market's local ratings are determined by 435 people. If they're weighted according to each market's household population, then obviously cities under the average will get more. (Someone else can crunch those numbers.)
But let's say all 25,000 households are used to measure national ratings. It would be like going to a city of 500,000 people and trying to draw accurate conclusions on what the city is like by talking to 156 people - what's the chance that your view from those 156 people doesn't really measure what's going on there? There's *a lot* of room for measurement error here, and critics have alleged this for a long, long time. (Go read why Star Trek was cancelled - it's a great discussion of how Nielsen got it completely wrong.) Since it's really just 5,000 households determining the country's ratings, then it's similar to using 25 people to "accurately" measure the characteristics of a city of 1,000,000 people. Does that strike anyone else as completely bizarre and off-base?
(Note: if there are 80 million households, Nielsen should be using another 3,944 households to measure national TV ratings more accurately; if there are 100 million households, they need another 5,000 - perhaps Dr. No will expound on this point to explain how to determine a proper sample size.)
Now supposedly Nielsen uses households that make up a representative sample of the U.S. population - but what if hockey fans aren't in the typical representative sample? We hear all the time about how hockey fans are hard-core, rabid, die-hard fans ... what if their personal characteristics are largely on one side of the population base and thus underweighed (and undermeasured) by Nielsen? It would be interesting to see a comprehensive study of the demographics of hockey fans, because I suspect they don't fall in line with Joe Average.
Back to the local markets - Nielsen has long been criticized for their measurements here, as local TV stations pointed out that local numbers have been lower than the national numbers for much of the time since Nielsen started this. If this is indeed the case, then how much accuracy can we put into any local ratings if they're undermeasuring true TV interest? And if the demographics of hockey fans in local markets vary more from Joe Average, wouldn't that mean that hockey ratings are probably understated even more?
So ... the use of TV ratings is questionable even as a starting point; to discuss hockey "support" in any market regardless of how large a market we're talking about (national, regional, or metropolitan) is even more questionable. It's much more likely that Nielsen's ratings are flawed to begin with, and depending on the mismatch between the demographics of hockey fans and the general population, it could be even more flawed against getting an accurate measurement of how many people are watching hockey on TV in any region. This would logically translate to possible over/undermeasurement of other TV programs and genres.