OzTAM to lift number of ratings households, track online viewing.
OzTAM is set to lift the number of households in its survey panel and begin tracking online viewing of TV in 10% of homes.
- Published by David Knox
- on
- Filed under News
Ratings service provider OzTAM is set to lift the number of households in its survey panel and begin tracking online viewing of TV programmes in 10% of homes.
OzTAM CEO Doug Peiffer said, “We are also pleased to announce that over the course of 2012 the OzTAM panel size will increase from 3,035 homes to 3,500 homes, ensuring OzTAM’s continued world’s-best positioning compared to any TAM panel in any market.”
It will raise the number of panel homes to 950 in Sydney; 900 in Melbourne; 650 in Brisbane; 500 in Adelaide; 500 in Perth.
10 per cent of panel homes will have both their TV sets and their PCs metered for viewing of broadcast TV content.
“We are currently working with Nielsen to develop the options for reporting viewing across multiple screens and to provide initial top line insights to the marketplace later this year.”
OzTAM has also extended its contract with data suppliers Nielsen for television audience measurement (TAM) services in the five Australian metropolitan markets to 2017.
“This new agreement leverages OzTAM’s substantial investments including the introduction of its Time Shift Viewing service, upgrading all panel homes to the state-of-the-art UNITAM metering system and decisions this year to increase the size of the OzTAM panels,” said Peiffer.
David Ellem, CEO of Nielsen’s Television Audience Measurement business in Australia, added: “Television viewing habits are evolving as new technologies create additional opportunities to view and we look forward to working with OzTAM to help Australian media owners, agencies and advertisers understand and leverage these developments in the coming years.”
In the recent Audience Inventory, 80% of TV Tonight readers indicated they had never been asked to participate in a Ratings Survey.
OzTAM is jointly owned by Seven, Nine and TEN.
55 Responses
@kats.. While it may suit a network or networks to select specific people, that suitability would tend not to last long and given the amount of programming changes we sometimes see and the unhappiness those draw, networks trying to game the sample group could wind up badly bitten. I would never advise anyone to try and cheat such systems; it just doesn’t pay off and, as suggested, it can wind up doing more damage than good Always best to conduct yourself ethically and accept the results as they may fall.
@kats The channels have joint ownership over Oztam, but it’s independently run I believe.
To all the people saying that 3500 houses isn’t many, you should consider that most houses usually have more than one person living in them so its more likely to be around 5000-10000 people being surveyed.
Also, to people saying to blindly sample everyone, the surveys also give stats based on demographics (age, gender) so it’s not a just a simple view counter.
The amount of people surveyed is also more than enough to give an accurate indication of viewing stats; Think about it, 3500 people from a broad range of demos is gonna give the same stats as sampling all of Australia isn’t it? It’s a pretty big sample size.
And adding to @TasTVcameraman comment, does anyone know why they don’t survey Tasmania?
@SusanP,thank you.From what I’ve been reading here it sounds like the sample survey is enough to give us accurate enough figures.Thank you to those who explain how statistics work.However once again,I’ve got another question along the same lines as my last post.If OzTAM is jointly owned by the three commercial channels,aren’t they going to choose households that best suit their needs?
@kats..not a naive question…a very important one. Oztam don’t say how they select…simply that it’s random. Households are only on the panel for 4 years and people can leave at any time. I am guessing they use socio-economic factors but ethnicity too? No idea.
Again Tasmania does not rate a mention !
Sorry..in previous comment this: “55.2 and 55.8% ”
should have read “55.2 and 58.8%”
@Trix… Just remember that in statistics and mathematics, deviation is considered normal.
My question is how are the 3500 households chosen?How does the general population know that these households are representative of all demographics?My questions perhaps sound naive,but just wondering.
Isn’t it correct that some homes have more than one box allowing for the fact there is more than one television or broadcast device in the house? I believe that is the case so knowing numbers of homes isn’t quite the full picture. If someone has 3 sets (and thus 3 boxes) and another home only has 1, you potentially skew in the former direction. BUT that is complicated by the array of choices open to the first household and one person (or more) may have chosen the same as the second home. Now sure how they account for these specific differences, or if they even try to.
The sample size is satisfactory for a 95% confidence level with a confidence interval variation of around 1.8 (going on 22 million and example size of just under 3000). So, if say 57% of sample gives ABC as an answer, you can be sure that if you asked the same question of the entire population that between 55.2 and 55.8% would have said ‘ABC’ also.
The sample size would not be satisfactory for a different confidence interval level.
I’ve reached the heights of my maths here so anything further would need more expertise!
Is it wrong to become slightly aroused by stat nerds?
@DeeBee, Phillo, ptrg: tracking every house through set top boxes will never be the answer. giving every tom, dick and harry the power to affect the data will never be more accurate than a sample survey.these 3500 houses are designed and monitored especially carefully to represent the Australian population. doing this on a large scale would never give an indication of how many people watching one screen, the demographic of the viewer, or their viewing habits without breaching their privacy. not to mention that the data would always skew toward a more tech-savy group and they could never rule out networks cheating the system.
@Maev. yes. toilet/kitchen breaks are all taken into account.
Tex is 100% correct in what he says. Probably.
@MJL – you’re confusing “homes” with “individuals”. Your numbers only work out if every OzTAM household consists of just one person.
It’s also more complicated than that because you’re also assuming that the structure of households is the same for each city, which is certainly not the case; eg around 40% of households in Per and Ade are 1- or 2-person, while in Syd and Mel it’s about 32%. There are a lot more 5+ person households in Syd and Mel than in Per or Ade.
Lastly, for those that made it this far, a joke:
How many statisticians does it take to change a lightbulb?
1 to 3, alpha = 0.05
I agree that we should be able to track individual household numbers but if you look at clinical trials of current medications – the numbers represented are much less compared the prevalence in society and so the same things applies here. It isn’t fool proof but its is a temporary patch till we are able to track actual numbers
@MJL – I agree with you. I have no issue with the sample size as such, i just dont know why they arent proportionately split to better represent the population spread.
@Bazza and @simmo3 – I understand why you think the 3000 odd homes are unrepresentative. However there have been studies done that suggest these numbers are about right.
From my limited understanding, if they doubled the amount of sampled households to around 6000 then the accuracy of the data would only improve by less than half a percent if not less.
The most accurate way of determining the figures are rolling out a box to every house, but the cost of that would be horrendous.
Long term… the NBN will provide interesting opportunities. When we look at the world of IP based TV, accurate figures of viewers will be readily available for content delivered via IP.
@ Bazza, DeeBee, etc: To give an example I’m currently working on: to be 95% sure we can detect a contaminant present at 2 parts per thousand, we only need to take ~25 1kg samples from a total mass of 75000kg.
Curiously enough, that works out to be about the same sample intensity (0.03% of the total) as the OzTam ratings households (3035 out of 9 million)…
@ Dr. Rudi…there are 3 types of lies…Lies, Damned Lies and Statistics
What we have is a rough calculated guess of who watched what
And this information they are relying on….how many of these 3035 folk…go to the bathroom or the kitchen during ads…or simply mute them because they are annoying….I wonder if their little box tells them that.
With todays technology they could easily have Set top boxes/PVR’s upload what was watched and recorded during the week. This would provide a much closer picture of what is rating well. With the current system the 3500 surveyed could not watch a program that in reality millions watch. Such a small snapshot is pointless. IMO
@Bazza – big respect, but I suspect you know nothing about statistics, surveys, or sampling.
@simmo3: That’s because you don’t get the power of statistics. I’ve done a very rough analysis here before somewhere; IIRC, if their 3035 homes accurately represent the demographics of the whole population, then the ratings figures should be accurate to better than 2%.
The fact that the networks get all excited about sub-1% changes suggests that they don’t really understand statistics either…
@simmo3 – it’s called a sample survey. Three thousand is a quite a large sample, and would have a margin of error of less than 2%. Why does this affect what is on and off tv? Money. Advertisers want eyeballs for their product; and Networks need to generate income.
Statistically speaking, a sample of 3000 (or 3500) from a total of 22,750,000, is absolutely useless data.
I can’t figure out the sense of these numbers as it seems some ratings homes will represent more people depending on which city.
Going off the 2012 Universal Estimates at Oztam these will be what each ratings home counts for in the Cities
Syd = 4963
Mel = 5140
Bris = 4666
Adel = 2860
Perth = 3780
So why isn’t each home equal?
UE’s found here oztam.com.au/documents/Other/UEs%20Metro%20TV%20Q1%202012%20-%20141211.pdf
still cant work out how 3035 homes having ratings boxes multiplies out to millions and why this rules what shows they put on and take off….to me its crazy