SBS boss: Oztam ratings are ‘very unscientific’ and underestimate size of NITV audience
The boss of SBS has described the official ratings system for television OzTam as “very unscientific” and “not accurate”, and argued it significantly under represents the audiences the multicultural public broadcaster achieves on channels such as indigenous focused channels such as NITV.
Michael Ebeid, managing director of SBS, told Mumbrella after last Friday’s 2015 Upfront event that he did not believe Oztam was accurately reflecting NITV’s audience, which has previously posted a 0.0 per cent audience share in the official TV ratings since its launch two years ago.
“The thing you have to understand with the way our ratings system works is very unscientific,” Ebeid told Mumbrrella. “For example, there are no measurement boxes anywhere in the Northern Territory (NT) so any viewer in the NT is not counted anywhere in our measurement system.”
Ebeid went on to argue that Oztam disadvantaged the broadcaster but that there was little that could be done to address the situation.
So does Ebeid dispute that Dateline has consistently held its audience of 250,000 , one of the highest rating programs on SBS now facing an uncertain future after 30 years? How does he and new director Jim Carroll justify cuts to that program? Either way you cut it with ratings formula, it is a quality high performer for SBS .
Firstly, it sounds like Mr Ebeid is not sure of OzTams remit, as highlighted by Pfeiffer.
“(The lack of ratings boxes in the NT) would obviously be a disadvantage for NITV in terms of measuring their numbers”
While true, it’s still not going to add a great deal to NITV’s viewership, given that the NT is about 1% of Australia’s population. That’s probably less than the margin of error within all the ratings in any case.
mmm night woman – need to get your facts right about Dateline – never got an audience of 250,000 in its life – otherwise would be SBS’s highest rating show weekly and it’s definitely not. Lucky if it scrapes 150.000 weekly and is way behind Insight in its audience performance which does regularly get 250k.
Sadly, that is a pretty a basic misunderstanding of the fundamentals of what are some pretty important data.
* OzTAM measures the five mainland state capital cities, so I’d be concerned if OzTAM had NT homes in its panel.
* RegTAM measures the regional markets and sub-markets, and commercial decisions meant that Darwin/NT was deemed not of sufficient size/commercial worth to be measured. That decision can always be reviewed.
* I was also intrigued by the assumption that NITV’s low share was because of the NT not being measured by either OzTAM or RegTAM. So I looked at NITV’s reach over the past four weeks in the five major metros. Guess what. NITV reached 3.25m people during that time – one in five people (20%) tuned to NITV at least once in those four weeks. This compares favourably to SBS2’s 6.1m (38%), and SBS1’s 9.4m (59%). In RegTAM NITV reaches 1.5m (21%), so very similar. All up, around 4.75m people watch NITV in a month.
* The issue of low ratings and share for NITV is not that people aren’t viewing NITV at all, it’s that they watch it very infrequently and/or for very limited durations.
* However, it must also be said that OzTAM does find it hard to recruit LOTEs (Language Other Than English). It also must be said that every research organisation faces this problem, and of all the media currencies that I work on OzTAM’s panel is closest to the LOTE population as per ABS data.
No-one is a bigger proponent of a National TV panel than me, so I say stump up some money, talk to the other broadcasters, start work on some contractual arrangement and then jump on board the National Panel train and let’s build it!
@ John Grono – Do you see LOTEs as being important? Why do we need to measure them at all?
Given that around one-in-five speak a language other than English at home – an unequivocal yes.
Hey No BS . I am quoting figures in Friday’s SMH. So you better tell the journo to get her facts right.
“The comments signal a rift in the SBS news division, with Hogan singling out the prime-time news bulletin for underperforming while Dateline maintained consistent audience figures of 250,000 people”.
Read more: http://www.smh.com.au/entertai.....z3JKKDHgSq
Hey No BS and night woman. I think you will find that you are both quoting data on different bases. No BS I think you are referring to Metropolitan-only data, while night woman is referring to Metropolitan + Regional.
In the 39 Dateline episodes to date this year, in Metropolitan markets it has averaged 150k and in Regional markets it has averaged 70k for a total of 220k for the YTD average.
It has fluctuated from a low of 155k (W/C 04/05/14) to a high of 317k (W/C 07/09/14).
I think this confusion adds weight to my comment about the need for a National Panel.
So does anyone really believe that surveying around 6,600 homes gives you an accurate representation of the video viewing habits of the roughly 7m homes in Aus..Maybe decades ago when we had a few FTA stations to watch – maybe then measuring a few thousand homes was a good proxy for the nations viewing patterns..but now?? at a time when viewing of video is so fragmented across not just FTA but of course more importantly viewing video on the internet on a non TV device, let alone using specific services on the large screen (Apple TV, Netflix etc etc) ….Also does anyone seriously believe that people today watch the same amount (or more?) of FTA TV as they did before we had all these options…seriously no one has swapped material FTA viewing time for time watching video on their laptop/tablet/phone or using youtube etc etc on their TV screen???? Might be interesting to see what the motivators are for measurement…
1. Yes. But it can always be improved upon- especially for the smaller channels.
2. No. And that is what the data is saying – FTA has dropped since we started electronic measurement over 20 years ago. Total TV viewing is also dropping marginally – 2014 could be the first year we see viewing per person dip below 3 hours per day, but I think it will squeak in by a minute or two per day. And that drop is partially because we’re defining “TV” as the fixed set in the room and not (yet) including portable devices.
3, Feel free to contact me for in-depth factual and empirical support of the above, rather than opinions.
4. Remember that we are “in the industry” and WE’RE the odd ones out. Don’t project your own media consumption habits onto the entire population.
John as stats major I would say that when you look at small option data pools then small sample sizes are generally sufficient (a la the viability of using a very small sample size of homes vs the large number than actually exist when the option set for people at night was TV, radio, reading)…In a world where you have a very large set of viewing options and a general population trend to niches then the small sample set of not sufficient to provide reasonable analysis of the behaviour of the actual base….
Also I must have a very non-normal set of friends..All used to watch a lot of FTA TV and all are now watching at least 50% less…Clearly given we are told that overall viewing of FTA is generally flat then my group (large group that it is made up predominantly out of non media types) must be balanced by a group that are watching MORE FTA TV now than before…love to meet these people and find out what they see on FTA that is so rivetting…
Daniel, also as a stats major and 37 years of applying that, you are spot on with what you describe as ‘small option data pools’. I prefer to call them dichotomous or binary distributions. The perfect example would be … ‘are you watching the TV right at this moment’ … you either are or you aren’t. As you’d be well aware we can apply p * q / SQRT n to establish the likely error level of the estimate (and all ratings are just that – estimates).
But let’s look at what is behind the TV ratings. OzTAM has a 3,500 home panel and RegTAM has a 2,015 home panel (though some home in overlap areas count to both). The average DAILY reporting sample size for MTV this year has been n=8,456 and for RTV n=4,811. That is a total sample of n=13,267.
Now the panels operate every minute of the day, every day of the year, so across the 365 days in a year, we’re talking a cumulative sample size of over 4.8 million ‘person days’ a year of TV viewing. This is important because if we keep seeing similar results for programmes that are shown weekly or daily our degree of confidence in the estimates is increased. This is with the caveat of the 95% confidence interval coming into play .. or as I like to think of it, there are 20 weekday news bulletins in a 4-week period, so one of them by random chance should produce a “what the?” rating. I must say it doesn’t happen often.
But back to the dichotomous distribution – did you or didn’t you watch TV.
In essence the system collects every minute of every day for those 13,000+ people. It calculates the rating based on the percentage of people that were watching and then accumulates that data into either a daypart (e.g. primetime) or All Day. Clearly this would be the estimate with the lowest standard error I am sure you would agree.
So I processed the 46 weeks to date of this year and calculated the average time spent viewing TV. For All People in MTV it is spot on 3 hrs per day. For under 18s it is 1 hr 57 mins, 18-39s is 2 hrs 12 mins, 40-54s is 3 hrs 29 mins, 55-64s is 4 hrs 16 mins, while 65+ is 4 hrs 51 mins. It is roughly 10% higher in the Regional markets. That includes ALL television – FTA, Foxtel, Fetch etc, along with time-shifted viewing, but it doesn’t include catch-up and streaming. And you know what, when I first calculated this in 1991 it was 3 hrs 14 mins. So across the past roughly quarter of a century it has gone down 14 minutes.
But you are 100% correct that the standard error for the smaller niche channels is much higher (as their ‘p’ is lower), and that their ratings are subject to higher error levels and more fluctuation. There are methods that can help alleviate this – such as the use of Return Path Data to establish Household Tuning levels that can be used in tandem with the panel to translate that HH Tuning into People-based viewing levels.
And Daniel, my viewing habits are NOTHING like those in the above data.
But try and track down a study by Jim Spaeth and Mike Bloxham of the Media Behaviour Institute using USA Touchpoints to research the (claimed) media consumption habits of media professionals compared to the general population. Yep, we’re nothing like the average Joe, and the people we mix with aren’t either. If I find a link I’ll post it.
Cheers.