I tender an open research project to anyone who might have access to historical TV data in Canada. This is not Google-able. You would probably need access to a database, if not the books themselves.
I'm interested to hear about the numbers for the premiere episodes of high profile scripted shows on Canadian networks going back to 1997.
Especially premiere numbers for shows on Global, CTV, and CBC.
The Associates, The Eleventh Hour, Tom Stone, Robson Arms, Degrassi, Jozi H, Falcon Beach, Whistler, This is Wonderland, DaVinci's Inquest, Intelligence, and anything else I may be forgetting.
29 comments:
Never underestimate the power of Google. Though I just used my TV, Eh? archives. So it's not everything you asked for, and a few you didn't, but, from Playback (here and here):
Da Vinci’s Inquest: 900,000
Whistler: 550,000
Intelligence: 443,000
Alice, I Think: 251,000
Jozi-H: 195,000
Rumours: 168,000
I wanted to be a librarian in a previous life.
Oooh, Diane! You rock.
Would 'Corner Gas', 'TPB', 'Jane Show, 'Billable Hours', Rent A Goalie, etc. apply?
Regenisis? Terminal City?
And doesn't sound like you're looking for the 'Stargate Atlantis' or 'Beastmaster' or 'The Collector' or 'Adromeda' or 'Mutant X's, right?
Diane, thanks, as always.
Will: Nope -- the lower tier cable numbers are too hard to explain. For the purposes of the comparison I'm thinking about, I want to compare apples and apples.
Also from Playback, Robson Arms started out as follows:
On its debut night, June 17, the first two back-to-back Robson episodes brought in 672,000 and 704,000 viewers, respectively. When the next two episodes aired, again back-to-back on June 24, the viewer numbers came in at 401,000 and 437,000.
Because I'm a nerd and this is kinda fun:
Corner Gas: 1.15 million
Instant Star (from same article as Corner Gas. They call it a preview episode - what the hell does that mean?): 1 million-ish
This is Wonderland: 630,000
Falcon Beach: 576,000
Eleventh Hour: 404,000
how about, made in canada, pr (remember that, kevin sullivan) snakes and laddders, rideau hall, american in canada (they did two seasons)
I'm intrigued Denis.
m
Ratings, huh? Anyone wanna poke the elephant in this particular room? I'm talking about how much we should trust Nielson numbers in the first place. Come on, you've all wondered.
According to Nielson's Canadian website (http://www.nielsenmedia.ca/):
"For our national ratings estimates, we use a sample of more than 3,000
households, containing over 7,000 people who have agreed to participate. Since there are over 12 million
households with TVs in Canada, it might seem that a sample of 3,000 is just not big enough to represent the
nation."
No kidding. BUT, as the website goes on to explain, statistical theory says this sample size is just dandy, as long as the sample selection is an entirely random process:
"Our sample homes are selected strictly through chance. Naturally we'd like to accommodate people who offer
to be in our sample, but doing so would violate basic laws of sampling practice. The sample would
immediately become biased because those who asked to join may be systematically different from the
population at large.
Instead, we carefully draw our sample in a way that offers every Canadian television household an equal
chance of being selected. Once the homes are selected and agree to participate, Nielsen Media Research
protects their privacy by keeping their identities confidential."
Notice something there? "Once the homes are selected and AGREE TO PARTICIPATE..." Slipped that one in there, didn't they? If the samples have to agree to participate, the selection process isn't exactly random anymore is it? Now consider an additional fact: TV professionals and their families are not allowed to become Nielson homes.
So what do you think? Does having a sample population that consists ONLY of people who want to be there but that EXCLUDES folks who work in the TV industry sound like it might be "systematically different from the
population at large"?
I'm wondering how all that might skew the precise statistical work that churn out the numbers that dictate how millions of dollars get spent and entire careers rise and fall.
House of cards or impregnable fortress? I'd love to hear from some statisticians on this one.
Jeremy
I would probably include Instant Star ... they premiered it after the Canadian Idol finale though since have punted it around the schedule pretty mercilessly.
Jeremy, you think people should be forced to be Nielsen participants? Of course they have to agree to participate - that's true of any kind of survey except maybe the census, where the government gets testy if you refuse. Oh, and the networks have yelped in the past when viewership has declined, claiming perhaps participants are reluctant to participate.
And it's standard to exclude people who have a professional stake in the results of a survey.
A bigger problem with Canadian ratings measurements to me is that they haven't caught up to changing viewing habits, like measuring time shifted viewing.
Ratings are estimates, and are obviously not completely accurate, but at least they're comparing apples to apples. Plus they're the only measurement we have of viewer numbers.
Hmm, speaking of TV professionals, Mark and Will, how come you're throwing out names but no numbers? Gee, too bad someone with connections to, say, Made in Canada doesn't stop by here.
made in canada, pr, snakes and laddders, rideau hall, american in canada, and instant star would all be good, too, for the exercise I'm thinking of?
I feel like the teacher now..
Okay? Anyone? Anyone but Diane? Can I see a hand?
Diane = Lisa Simpson...
aw!
Hey! I ... I ... would object, but sadly, I don't think I can. Gotta go practice my sax now. (No, not really.)
I gave Instant Star numbers above (pout).
But since you weren't satisfied with the "over a million" vagueness for the Instant Star preview episode, there's more here on what I guess was its non-preview, non-Idol timeslot premiere:
"CTV also premiered its new homegrown show Instant Star with back-to-back episodes on Jan. 23, at 7 p.m. and 7:30 p.m. The first episode attracted 507,000 viewers, rising to 534,000 for the second."
OK, I'm done proving Google's omniscient power and my uncanny resemblance to a yellow cartoon girl.
well i googled with playbackmag.com (just go site:playbackmag.com "made in canada" premiere)and found:
made in canada 935,000
snakes and ladders 408,000
couldn't find pr or american in canada.
My point, Diane, is that Nielsen justifies using a small sample size to represent all the TV households in Canada by saying the system works as long as the samples are randomly chosen. I'm saying, clearly, it's not such a random selection process, and therefore the sample size may be too low to produce a Good estimate. I'm surprised that so many people are willing to bank so much on what might be an extremely flawed estimate without questioning how it works.
And yeah, I think it's terrible that "they're the only measurement we have of viewer numbers". A quick google search shows I'm not the only one who's been bothered by all of this (from http://www.weshouldallcount.com/):
"Do you think less than 9,000 handpicked households should dictate what more than 250 million people watch on television? Neither do we. Now you can do something about it.
Welcome to WeShouldALLCount.com: America's single resource for the latest news and filings related to the erinMedia federal antitrust suit against Nielsen Media Research (NMR)."
I'm calling it: Emperor, no clothes.
I've got my assistant looking for some of the numbers...after she picks up my latte, of course.
"then Will wakes up"
DK -
There are far worse things in this world than being compared to a multi-billion dollar generating, genius, outspoken activist that is utterly adorable, loves jazz and has her father's heart.
Go get 'em, tiger...
Da, da-da duh-da, da-da, dee-dadadada...
Aww, thanks Bill. I think.
Jeremy, you want to know what statisticians think of the Nielsen system? Statisticians developed the Nielsen system. That's who Nielsen employs.
Where I come from, the basics were taught in high school math. Google can help with what it means to take a random sample, too. There are formulas to identify how many participants are needed to represent the population you want to study and give statistically accurate results. There are formulas to identify how many people you need to randomly select in order to get enough people for your sample, because there will always be people who choose not to participate.
There are flaws to the ratings system, and a too-small sample size may be one of them (though statisticians may disagree), but objecting to the fact that people can choose not to participate is ludicrous.
What's the solution - giving a company the power to monitor our TV habits without our permission?
Now that Nielsen and BBM have merged, if people have a better idea of how to collect ratings maybe they could try launching another competing corporation instead of wasting time signing a petition (gee, how many names do they need to represent a statistically significant number of people who have a problem with the system?). And if they object to the statistical method in general, they might want to expend some energy into objecting to other areas where it's applied, like, say, medical research.
Well, Diane, other people who were awake in their basic high school math class had this to say:
"Another criticism of the measuring system itself is that it fails the most important criteria of a sample: it is not random in the statistical sense of the word. Only a small fraction of the population is selected and only those that actually accept are used as the sample size."
That's from Wikipedia, which carries a few more criticisms of Nielsen for anyone who cares.
And no, I'm not saying people should be forced to participate. I'm saying the world would be full of unicorns and rainbows if 1) ABSOLUTELY, YES PLEASE Nielsen had some real competition so we could have some other numbers for comparison; and 2) Nielsen used a much bigger sample size.
Nielsen itself understands why a larger sample size is better. From a Nielsen Media Research
Executive Briefing called
"PROGRESS IN MEASUREMENT
First Edition: National Sample Expansion" (Feb. 25, 2004):
"When complete, this expansion will yield an NPM
panel with a gross sample size of 12,580 households, or an effective sample
size of 9,685 households. The increase in sample size, for example, will
enable Nielsen to report on more precise audience definitions, including
upper income levels. By roughly doubling the effective sample size of the
National panel, Nielsen will increase the reliability and stability of its audience
estimates by reducing the sampling error by 28%."
Is Nielsen similarly expanding in Canada -- anyone know?
And now, yes, I'm going to drop the subject and go to medical school.
did anyone mention davinci's city hall? does billable hours count? dragons den? naked josh?
someone else will have to find the numbers...i don't know where to look...
diane:
sorry, i posted and then had to run to a basketball game (my son's team won) and mentioned shows without doing the legwork.
since then mic is up. someguy had it at 935 (the second one was way down I remember that)
the oddd thing about that show is that it was ratings proof. the subject matter and especially the lead (rick) guaranteed that it wouldn't be cancelled.
okay i offially suck at this. no luck on p.r. or american.
let the roasting of me continue.
mark
See Jeremy, there's the intelligent criticism of the Nielsen system - the sample size, not the fact that people need to accept in order to participate.
Aw, Mark, that wasn't roasting, it was good natured ribbing! I got called Lisa Simpson ... you got off pretty easy, I think ;-)
Diane, Jeremy.
It is a legitimate concern if the need for agreement is skewing the selection.
Forcing people to participate is not a possible solution, of course. But increasing the selection size doesn't address the issue if the selection method is biased.
The way to accommodate biased selection is to weight the results to counter-balance it. Now, you can't by definition get any info on the people who don't participate in your survey, but you can always compare your results with those of a proven source, such as Stats Can.
For example, if Stats Can says you should have 20% people over age 70, and you've got 40% then you know they are twice as likely to agree to participate in your survey, and you give that demographic half as much weight when you come up with total numbers.
If Neilson employs actual statisticians, I would hope they've done something like this.
They do, and that would be why the networks yelped when Nielsen recruited more young men to address the fact that young men were more likely to refuse to participate. When ratings for young men dropped, the networks claimed it was because young men were not actually participating, even if they agreed to be Nielsen families. In the US, same happened with Hispanics. Nielsen tries to address imbalances between their sample and the population at large, but if it doesn't go in the networks' favour, they complain. And it's very possible they have a legitimate beef - if people in the sample aren't actually participating, that skews the results. But the somehow don't complain when things get skewed in their favour.
There are many, many things wrong with the Nielsen ratings, but Jeremy needs to drop his initial objection, the objection that started all this pedantry, if he's going to have a credible argument that doesn't fly in the face of accepted statistical methods:
Notice something there? "Once the homes are selected and AGREE TO PARTICIPATE..." Slipped that one in there, didn't they? If the samples have to agree to participate, the selection process isn't exactly random anymore is it? Now consider an additional fact: TV professionals and their families are not allowed to become Nielson homes.
The Nielsen selection process is very specific. I know a Nielsen household (even though I am not supposed to know they are a Nielsen household and they didn't tell, I just happened to see the paperwork on the coffee table and beat it out of them). This particular person was approached because they were a specific sex, marital status, income level, particular kind of pet-owner, home owner status all within a set postal code (in my area, about a six block radius). That said, many broadcasters have complained for years about the sample. APTN has probably the most legitimate complaint ... most of their viewers are in the north and Nielsen is underrepresented in those regions and in Aboriginal households(partly due to servicing logistics) despite the fact that they watch tv more per capita than people in other regions of Canada.
Diane: "And it's very possible they have a legitimate beef - if people in the sample aren't actually participating, that skews the results."
Jermey: "If the samples have to agree to participate, the selection process isn't exactly random anymore is it?"
I don't understand. Doesn't that mean the same thing?
No. It doesn't. Which is why this whole digression is depressing. Guys, if you want to go and study statistical theory and then come back and debate it, that's fine.
But the problem with the whole "internet objection to ratings" movement is that it's the case of a little or no knowledge being a dangerous thing. It reminds me of Creationists talking about the flaws in the theory of evolution. Depressing.
Post a Comment