Those %$#@*&! numbers

A thread on Matt Ottinger's game show board got into Improv-a-Ganza's crummy ratings. Matt himself posted a caveat:

I have maintained for years that at the bottom of the cable barrel, the numbers are so small that the differences between them are virtually insignificant. Remember when we used to hear about the "margin of error"? The margin of error for this sliver of audience measurement is probably greater than the largest rating GSN usually gets, meaning that the whole thing is just a great big guess.

Um, not really. GSN's biggest audiences are five or six hundred thousand people. That's way above margin-of-error territory for Nielsen's huge (by statistical standards) sample. And GSN isn't at the bottom of the cable barrel. It's a mid-sized network that often sneaks into the top forty basic cable outlets.



Matt's a PBS guy in Michigan, so he's probably not a big fan of Nielsen and their pesky people-meters. PBS is not known for its sky-high numbers. To be fair to Matt, he's hardly alone in his dislike of Nielsen. Most everybody involved in American TV has cursed those irritating but almighty numbers.



But when Nielsen keeps saying day after day that nobody's watching Improv-a-Ganza, well, it's almost certainly true. The sample might wobble back and forth but after a while the probability is with The Nielsen Company. Central limit theorem and all that.



UPDATE: Wouldn't you know, Improv-a-Ganza scored its best numbers in a long time on August 7. But only in the 8:00PM hour. In the 11:00PM hour the numbers returned to their usual suckiness.



For the ratings week of August 1-7 overall, GSN got a prime time average of 340K viewers, its best in many weeks. And well above the margin of error.



UPDATED UPDATE: Nielsen is now getting some competition from Rentrak, a company that uses set top box (STB) data from Dish Network. Not to be outdone, Nielsen itself has made a deal to get STB data from a hundred thousand or so DirecTV households for use in local markets. Nielsen has no plans yet to use STB data in its national ratings.



There are serious issues about how representative and reliable STB data is. But I'd love to see real competition for Nielsen. The company has monopolized TV ratings for decades. High barriers to entry, as the economists say.



Plus, more numbers from more sources would make any actuary happy.