Looking to give demanding local TV advertisers more stable viewership, consumer data and reliability for their local media campaigns, Nielsen is expanding its panel of portable people meters and return-path data into a number of top U.S. markets.
These new “currencies” start Oct. 3.
Nielsen says all these efforts to increase sample sizes will improve reliability, reduce errors, and result in reductions in “zero-rated quarter hours.”
Portable people meters (PPM) -- which measure out-of-home viewing, for example, in gyms, doctors' offices, hotels, workplaces and other people's homes -- will be integrated into 25 Local People Meter (LPM) markets and into 19 set-meter markets.
advertisement
advertisement
In addition, Nielsen says, National People Meter homes will be integrated into those set-meter markets.
All this will approximately double the panel size in the top 44 markets (25 local people meters and 19 set-meter markets). For example, for persons ages 25-54, the sample size across those 44 markets will increase to 60,000 from 30,000.
In addition, Return Path Data [RPD+] -- which comes largely from set-top boxes from DirecTV, Dish Network and Charter Cable -- as well as National People Meter homes will be integrated with panel data for 12 set-meter markets and 15 code reader markets.
This will sharply lift the panel size of those markets --- to a range of 12,500 to 300,000 homes (depending on the size of the market) from its home panel size of 400 to 800.
Local people meters are in 25 of the largest 26 markets, measuring 50% of U.S. TV households. Set-top meters are in 31 mid-to-large markets -- 21% of all U.S. TV homes. Code readers are in 15 markets, representing 5.5% of U.S. TV homes.
RPD+ measures are in 137 mostly smaller markets, representing 23% of U.S. TV homes, converted from diaries in July 2018. Code readers make up the “+” in RPD+, installed in markets to capture over-the-air viewing that is missed by set-top boxes.
Overall, Nielsen is adding two million RPD+ homes to its samples and more than 1,200 National People Meter homes with the change.
From all of this, Nielsen says better granular data will be the result, with those 27 set meter and code reader markets seeing -- for example -- an average 83% reduction in “zero-rated” quarter hours for 25-54 viewer measurement.
Local Broadcast Television is a national resource, with tremendous value that can only be realized by automation in the ad ecosystem. I'm impresed with what Videa is doing in that space, particularly with their new open-source APIs developed with Sinclair Broadcast Group and Imagine Communications.
Wayne, Nielsen may be increasing sample sizes in the 44 large markets by virtue of bundling together various types of meter panels but look at the numbers. The average sample size for adults aged 25-54 will now be 1364 persons per market---per the figures you cited. with an average TV rating for broadcast TV and cable being, say 1%, that means that the average quarter hour program episode sample size for this demo in these markets would work out to 14---and that's if all of the meters were working 100% of the time---which is very unlikely. And what about the low rated shows--of which there are many? Here, we are talking about sample sizes--per quarter hour episode--- of a couple of people? iIt can be argued that the way much local time is sold---in rotating packages--- that you can use averages across a week or a month for five times per week shows like local news or syndicated entries such as "Wheel Of Fortune"---which does raise sample sizes somewhat, but let's face it. Compared to what is considered minimally accecptable nationally, local TV rating measurement is not even close when it comes to sample sizes for viewer data.
As far as set-top-box tune in being added to this mix of methodologies, that's fine ---providing that there is some way to render such data as representative of the total market---but how do you get viewers aged 25-54 or any other viewer info by using this method of enlarging sample sizes?All it gives you is set usage.
Correct, as usual.
Well-expressed, as always.
Thank you, Ed.
In William Shakespeare's Much Ado About Nothing, deception is the means by which the characters bring about change.
I am not accusing Nielsen or MediaPost of deception, but when will media and marketing professionals learn that there is more to media research than a "stability" purchased at the expense of accuracy, validity or truth.
I shall not reiterate the careful and appropriate assessment of Ed Papazian to whom I shall always defer for his expertise and experience. I shall also not take issue with Wayne Friedman who cannot always recognize the sins of omission committed by the subjects of his reports.
However, can anyone really believe that Portable Passive Meters (PPM's) or Return Path Data (RPD+) from the set-top boxes of satellite homes and Charter Cable really enhance the quality of ratings data? What has become representative surveys from area probability samples?
It seems in research as in politics, down is up, black is white and fiction is truth.
Welcome to the DRS: Dystopian Research Society
Ed, your n=14 derivation is one of the possibilities.
But as we both know, (i) ratings are based on average minute audience, and (ii) the incidence of people watching every minute of a programme is quite low.
So, based on your average n=1,364 sample size for P2554, a 1% rating would equate to n=14 people. But they are 'average minute audience' people, and not sample size.
For example, assuming this was a one-hour programme 1% rating it could be an 'active' sample of n=14 people who watched all 60 minutes. It could also be an 'active' sample of n=28 people who watched only 30 minutes. Or an 'active' sample of n=56 people who watched only 15 minutes. And, so on.
And playing devil's advocate, there are two ways of determining an audience. One is to enumerate those who watched it. The obverse side of that coin is enumerating those who did NOT watch it.
So sticking with your n=14 of n=1,364 who watched itout , means that n=1,350 people who did NOT watch it. Given that sample size of n=1,350 of n=1,364, it would be a pretty confident estimate!
And regarding your question as to whether all the meters were working that is a good point. It actually goes beyond whether the meter was working, but also whether the raw data is 'believable'. For example, the TV might be on but no-one is registered as watching (maybe the dog was). Conversely, the TV might be on and tuned to a single channel for hour upon hour with people registered, which is highly suspicious unusual behaviour, and due to the volume of 'tuning/viewing' may slightly shift the ratings needle, meaning those viewing records should be edited out for that day.
While I don't have the US incidence of equipment issues and editing out suspicious data, in Australia, on a typical day around 95% of the sample would be used. It should also be recognised that on a typical day there would be a 2%-3% 'buffer in the sample size to cater for such issues as well as to provide 'wriggle-room' to allow panel representation to be maintained.
John, unless it has changed, local TV ratings in the U.S.---as opposed to national ratings---are not average commercial minute but quarter hour total audience. As regards the down time of the meters in the old days when all we had was a national household panel of a few thousand homes, the typical drop out rate per day was, I believe, over 10-12% and possibly higher. This was because the TV sets, themselves, were not working as well as the meters. These days TV sets are far more reliable so their breakdowns are much less of a problem. As for the various meters now in use I don't know what the corresponding down time percentage happens to be but I suspect that it's higher than 5% per day. Would anyone from Nielsen care to comment?
Fair point Ed regarding local TV ratings are/were based on quarter-hour diaries.
I was jumping the gun - I seem to recall an announcement that meters was to become the default/preferred methodology, and my top-of-mind was thinking about the AU system where all local markets are metered.
John, I believe that even in loocal metered markets Nielsen tallies quarter hour not commercial minute audiences.