April 28 - May 2, 2008

[Roman]

Yeah.

It stinks.

Ok?

[YRBB]

The episode were Storm kills himself and Brooke runs and finds him got 4,5 million viewers!!!! There IS a God!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Toups, I think you're right. I'm not 100% sure for LML's reign, but under Bell, Alden, Smith, the show NEVER got that low. How sad.

[JackPeyton]

i find it intersting that every soap has higher ratings on monday then on friday. i always assume poeple tune in friday.

[Sylph]

Nielsen's use the wrong statistical theory.

[MarkH]

Okay, now THIS I want to know about.

What does Neilsen's use? I assumed probability sampling.

What are you proposing as an alternative?

[Sylph]

Yes, they use probability sampling (also known as random sampling) and stratified sampling, among other things. And that is their main and major defect – the sample just isn't big enough. There are many serious issues involved as there are many types of criticism – of which genuine, the one that wishes to better the technique, is the rarer type. Mostly people actually don't care about Nielsen's methods, they just hate its monopoly.

That whole thing with which they justify using a small sample size to represent all the TV households in the US by saying the system works as long as the samples are randomly chosen is just... Anyone with a little knowledge of high school maths should know how a sample should be chosen.

I'd really like to have a chat with a Nielsen statistician.

As for the alternatives – I haven't really thought about it thoroughly... Did you?

[MarkH]

Okay, thank you! Yes, I'd like to speak with one of their statisticians too.

You know, I think a relatively small Neilsen sample works--like a Gallup or Roeper sample--as long as it is truly representative. Obviously central limit theorem says that the bigger the sample gets, the better, but fairly small samples can give good results.

From what I know, my problem with the stratified sample is more that they fail to adequately sample from settings in which much viewership occurs. I know there was an attempt to fix that within the last year (e.g., dorms, bars)...but I'm not sure that it made a difference.

[Sylph]

Well, I really hope those statisticians Nielsen employs know what they're doing. There's just too much criticism from their colleagues to be putting any trust into it anymore... In the next few days, there's a chance for me to meet a statistics professor who was involved in those types of studies, so I'll let you know if the meeting materialises.

Yes, I wonder what happened to those dorm measurements. Is that still in the experimental stage?

[Jess]

There is nothing wrong with the way Nielsen does ratings. It randomly selects participation using geographic selection and also randomly-generated telephone numbers. It uses meters in larger markets and dairies in smaller markets. The sample size is very large -- I read somewhere that it samples 25,000 metered households, although that seems very large. I'll edit, that large sample probably means they do local ratings too.

There is nothing wrong at all with Nielsen ratings. It is developed and monitored by very skilled statisticians. It uses neighborhood outreach to avoid many of the problems associated with random dialing. It also uses random dialing to future ensure a representative sample. Today the overwhelming majority of homes have at least one television so it really isn't hard to develop a sampling frame.

Finally, it also collects data from DVRs. Soaps are tanking. You don't need ratings to tell you that. Believe me the day that students sit in the college lounge and watch soaps are over. You don't even hear people talking about "their stories". Talk to a college student, very few still watch soaps. Oh yes, I know someone now will say "I am college student and I watch soaps" Good for you, you are the exception and not the rule. Let's save the pointless argument.

I did a survey, plus or minute 3 percent gang. One of the questions was what shows do yo watch. People are not watching soaps.