The Voice rates 2.6m for Nine, Ten gets less than 10% share

Can someone tell me how many TVs are used to calculate the ratings figures?
For, say, NSW or Victoria?

I’ve heard it’s bugger all.

Sydney’s sample size is 833 Homes (2,272 TTL PPL)
Melbourne is 703 ( 1,882 TTL PPL)

actually pretty robust in comparison to other countries

Is it just me, or does that seem an awefully low amount? How is the sample selected?

It does seem like a very small amount – I guess people see it also as an ‘invasion’ on what they are watching (although I am sure that most people would forget that their tv viewing is being recorded).

how can we get a sample size; if we are only getting a small portion of the population, without an even spread of demographics?

You want a Stats 101 lesson? First no sample size is ever big enough – we can always do with more. Extra sample costs … and costs a lot. More importantly it has to be a representative sample – and that costs even more.

Many people wrongly believe that you need total enumeration to get an accurate measurement. This is not true of relatively homogenous populations.

Two common examples are blood tests – around 5ml of you 5 litres of blood are needed. I surely don’t want ‘total enumeration’ for a blood test. A second example is cooking a cake or boiling peas in a pot. You stick the straw in or check a single pea – if they are cooked you know the cake or peas are ready.

We have a national sample size of over 5,050 homes covering over 10,000 TVs and around 13,000 people. That measures every minute of every day of every year and produces TV ratings the following morning without missing a beat.

Sure on some days it struggles with some niche channels or niche programmes but overall it does the job it is meant to do – to provide reliable and robust TV audience estimates.

Nielsen internet panel is AT BEST 7,500 people (and not representative of the overall population makeup) … so the TV panel is pretty strong.

raising questions about sample sizes when they’re over 5-10,000 shows more about the persons ignorance around statistics and research than the validity of the sample and method.

These people are professionals. I know the idea of the TV industry dudding the market and misreporting viewers is a nice conspiracy (especially for those seeking money from the medium) but it’s just not true.

The average daily REPORTING sample last week was n=8,048 for Metro TV and n=4,954 for Regional TV for a total of n=13,002 (my 13k estimate was pretty darned close!).

There are actually more people that this on the panels, but due to a variety of reasons around 8% of the panel do not ‘qualify’ for reporting, plus there is around 1%-2% of homes “in the pipeline” ready to come into the sample.

So Logic, while I agree with you on the sample size, I would have to disagree about the representativeness of the population.

The panels are VERY CAREFULLy selected and constantly monitored to reflect what the panel is designed to report on. The population is stratified on key determinants of viewing – such as the number of TVs in the household, the lifestage (age and size) of the household, and the presence of Subscription TV. The monitoring shows that every cell in the stratification is within 2 percentage points of the ABS population estimates and indeed the gross majority are withing 1 percentage point.

However, it is important to note some things about the system. For starters, not all markets are included – just those that are economically feasible to measure. So ‘solus markets’ such as Mildura, Darwin etc. are not included. All up they would probably account for around half a million of our population.

The panel is also designed to represent the in-home broadcast television viewing population. Obviously, homes and people without a TV are excluded from the measured population (around 1% of homes). Television viewing that occurs in public places or ‘on the move’ is also not captured or reported on. Certain devices such as tablets and smartphones are still being worked on. It is also salient to remember that we are measuring ‘broadcast television’ and not ‘all video’. The panel is currently being upgraded to allow the reporting of ‘catch-up’ TV viewing over the internet.

All up these ‘exceptions’ would represent a few percentage points of the gross viewing levels not captured and reported on – and frankly they are not (currently) worth the cost and effort to do so. However, none of the above caveats should be in any way construed as the panel not being representative.

I would suggest that what the panel was originally designed to report on is increasingly different from what marketers actually need to know about the performance of TV shows.

Using at best broad demographics to measure the performance of TV shows is a woefully inadequate approach in this day and age.

‘Reach and Frequency’ or Tarps are not an end.

I agree Chris. We’re talking about an audience measurement system and not a campaign measurement system – which these days is virtually by definition cross-media. No single medium can provide total campaign measurement. They can however provide the fundamentals of communication planning (contacts, reach, frequency etc.).

While marketers need to know about the performance of TV shows, each marketer has a different view on what they define as ‘performance’ – in essence they want bespoke demographics. I agree 100% that this is different to what the panel is/was designed to do. I await to see the flood of marketers financial backing to provide such bespoke demogarphics, and until then age/gender and their various other incarnations remain the surrogate cohorts.

One thing we do agree on Chris, is that TARPs and R&F are merely ‘inputs’ into campaign performance modelling systems which work out which parts of a campaign drove sales.

Hang on….

If you have a sample of say 8,000 and report that 2.4 million Australians are watching The Voice.

What are the other 7,000-odd people in the sample watching? Cause population percentage would suggest most of them aren’t watching TV at all.

@ Me – correct. The ‘most popular’ channel on TV is “Off”. Always has been, always will be. There is no medium that commands the majority of the population on a regular basis.

To follow The Voice example using last Sunday night, here’s some of the salient ratings data for Metro TV only:
* Programme average (18:33-20:30 ) 2.815m (using latest programme logs and viewing)
* This is 17.9% of the population – just under every fifth person was watching it
* Average TV audience for that same time frame was 7.481m
* This is 47.6% of the population – virtually every second person was watching a TV channel

If we look at the people who watched ANY of The Voice or ANY TV during that same two hours we see:
* 5.403m people watched any of The Voice, which is 34.4% of the population – every third person
* 10.098m people watched an TV, which is 64.35m, almost two out of every three people.

“The ‘most popular’ channel on TV is “Off”. ”

Thanks for the answer, John. Sure I’m gonna get a lot of mileage out of that quote.

@ Me – feel free to. I have for the past few decades!

Just also remember that the majority of the population are also not on the Net, reading newspapers or magazines, listening to the radio, in a cinema watching a movie, or driving past billboards, at any given time of the day.

The implication that it is unusual that the majority of the population is not interacting with a medium at any one time is pretty wide of the mark – people are just too busy living their lives.

Yes, “off” rules all media..

And lets mot forget those who are out shopping with an influx of POS media…

This is why John Grono is without doubt the most respected researcher in this industry…
appreciate all your comments

What about the biggest entertainment industry in the world, video games? I note they’re rarely discussed here and there doesn’t seem to be any way of officially tracking their effectiveness (possibly because it’s too difficult, they’re so diverse.)

Is there a place in advertising for someone with experience in games? (Self-serving question, stuck in #MSM and have a mind to do something different.)

@ Me. I’ve yet to see any empirical evidence that video games is the biggest entertainment industry in the world. Large – sure. Biggest – doubt it.

There are various sources which can give pointers but none that provide a picture in totality. This is largely because as an industry they haven’t got there game together (pardon the pun) to produce industry-endorsed metrics – and more importantly to put the money on the table to produce an industry ‘currency’.

You are able to garner some information from Nielsen Online as to online gaming usage. Last month it was around 1.4 billion minutes (around 4 hours per month per person) – this is around one-third of the average daily minutes of TV viewing in Australia. Their meters MAY be able to produce information about off-line gaming usage on a computer as well as the meter is at an OS level.

For usage on a device connected to a TV, OzTAM’s meters know when the TV is on and what broadcast TV is being watched, and therefore OzTAM report on the broadcast usage of TV. But there is also this “bucket” of unreported “non-broadcast” usage of TV, and it is unreported because the meter doesn’t know if it is gaming consoles, connected PCs, other connected devices etc. But this may give an indication of the maximum that gaming on the TV set could be.

Finally, regarding hand-held gaming devices, you would need bespoke research as to the incidence of ownership, frequency of play and duration of play.

If you were able to source all three and put them together you’d start to have a picture.

As Jerry Maguire said … show me the money!

@ Paul. Thank you for your generously kind comments. Mind you it is a sample of one and shouldn’t be extrapolated to the population!

and it is also a unique IP too

John – to clarify I was talking about the the representative issues with the Nielsen internet panel not the Oztam panel. I agree with what you’re saying.

Logic, you raise a salient issue.

I agree that it is easier to get people to agree to have their TV metered in their home than their PC.

Further, there is a lot of usage ‘at work’ and getting permission to instal metering devices in workplaces is problematic – especially for large businesses, government, finance etc.

That is why we moved to a hybrid system that captures the total traffic and places reduced reliance on the panel. That is, the total quantum is pretty accurate (traffic tags are not always accurate) and does a lot of the ‘heavy lifting’.

The role of the panel then is more to establish ‘propensities’ – such as duplication, multi-site access, multi-device access, multi-person access, demographic profiles etc.

Having said that it is still ESSENTIAL that we have as representative a panel as possible. Using ‘convenience’ panels tends to deliver large sample sizes at the cost of representativeness. We use a stratified random sample as it gives greater control but with the higher refusal rates online it is a much more costly and time-consuming exercise. The Nielsen panel is reported on monthly to the IAB and it is improving every month.