mel:
320 Replies! WTH Robert — more people post her about Dollhouse than are watching it!

maybe it means these shows tscc and DH have more viewers than indicated by neilsen?

Unlike others I won’t play in universalist statements. Instead I inend to speak in subjective tones: I like Dollhouse very much and have from the get go. I don’t like 24 (to much repetitiion). I don’t like Fringe (Abrams is the most overrated “auteur” which is quite an accomplishment for a nation that produces, for the most part,juvenile media mediocrities). I think Bones is an OK mediocrity. I don’t particularly like Terminator and hope Summer Glau ends up on Dollhouse if the latter survives.

As to the prophetic utterances I will just note that in the US longevity is determined in the final analysis by popularity rather than on the basis of quality (presumably in such a environment Ulysses would never get published). Caveat: only time will tell whether Dollhouse will be cancelled.

So they really wouldn’t be more accurate at all. We’d have no idea of the age/gender/number of those watching. The data would be useless to advertisers.

All I’m saying is that a relatively small sample can give very accurate models of the wider population. And that 25,000 is not a small sample.

The idea of my television telling someone what channel I’m tuned to is a pretty horrific one. And, of course, all it does is tell the company what channel I’m tuned to – not whether I’m watching, who’s with me, how much attention I’m paying or whatever.

The Nielsen system, as I said, has problems. Every system does. Sci Fi and the people who advertise on it know that they’re under-represented. I’d rather have a system where people know it’s roughly right, that keeps adjusting its model than everyone’s television watching them as they watch television.

‘For advertising purposes only’ … so, say, if a politician wanted to place some political ads, he’d have access to the data? Great.

TV viewing habits can not compare to most polling. There are just far too many variables. The same percentage of Niche Demos does not exist in every normal demo that Nielsen has to go after. If they’re looking for a college educated female between age of 18 and 34 with an income under $60,000 and single that’s fine. However when you start strategically going after groups, you lose the fact that niche demos in viewing aren’t going to get near close to being measured accurately.

SCI FI is a great example. There are certain demos that the percentage is higher for there being a SCI FI fan. Lets call it group X. Group X is more likely to watch SCI FI. Nielsen strategically measures group X in the same “national percentage,” that they’re in. The problem is Nielsen believes this provides statistically valid data. The problem is that SCI is a Nich Demo, and thus harder to accurately measure. So a SCI FI show is more likely to be inaccurately measured. TNS data has shown this in the past (TNS measures about 20 times the people that Nielsen does). The problem is that its only Niche Demos that are inaccurate and thus there aren’t enough people unhappy with Nielsen’s results.

Cable companies and Sat companies should just make it part of their agreement, if you order their service your viewing habits will be measured and shared for advertising purposes only.

Then we’d have more accurate numbers, but less “data,” about the viewers. Combine that with Nielsen and you might have something.

Paranormal Cop Show sounds right. A “science fiction” show with no science… bizarre.

I’m depressed. And I don’t mean because I’ve seen T:SCC recently.

Yeah. It’s basic statistics – as long as you make sure the sample is representative, and you ask the right question, a sample of 1000 Americans will get the same answer as polling every single American to within a couple of percentage points.

Or, to put it another, way – how many people do you think you need?

There’s a great explanation here:

http://www.socialresearchmethods.net/kb/sampstat.htm

But you see it with online polls, assuming the polls don’t get crashed – by the time ten people have responded, you can usually see roughly what the result is going to be.