January 25, 2023
Nielsen To Roll Out Controversial 'Viewer Assignment' Method, Will Model Audience Estimates In Non-People Meter Homes
- by Joe Mandese @mp_joemandese, February 19, 2015
Nielsen today is expected to announce plans to
begin rolling out a controversial method it says will fix some antiquated parts of its local TV measurement system: a process that would effectively use mathematical modeling to estimate the audience
composition and demographics of much of its local TV ratings estimates.
The move, which is part of a broader “sample expansion plan” to boost the number of households and persons represented in all of Nielsen’s ratings estimates, was announced broadly in September. At that time, Nielsen said it had previously briefed key clients leading up to that notification and that most of them were very supportive of the plan, although some felt it was questionable research methodology that needed to be tested and evaluated further.
Exact details of Nielsen’s decision to move forward with its so-called “viewer assignment” method will be disclosed when Nielsen makes the announcement, but it is expected to begin impacting official local TV ratings estimates in the next few months and initially will not be accredited by the industry’s self-regulatory media ratings watchdog, the Media Rating Council.
advertisement
advertisement
“Nielsen personnel have indicated to me that they will coordinate with MRC, our Television Committee and our auditors in moving these initiatives forward,” George Ivie, CEO and executive director of the MRC, told Media Daily News at presstime, adding that Nielsen executives had indicated they ultimately plan to comply with MRC standards and maintain accreditation for various services affected by the change.
Many of Nielsen’s non-metered local TV markets have been operating without MRC accredited ratings for a while now, and Nielsen executives have said the new method will fix many of the problems associated with its outmoded paper diary system that has been used to determine the audience composition and demographics in many local TV ratings markets for most of its existence as a ratings company.
But some researchers believe it is the most fundamental shift in the way Nielsen estimates audiences since it rolled out the people meter system for national TV ratings in 1987, and that more time is needed to evaluate the impact and validity of the method.
In a nutshell, the viewer assignment method mathematically models and ascribes audience estimates for non-people meter markets with data generated from its people meter sample.
“Nielsen will apply a demographic assignment methodology informed by [national people meter] viewing behavior to report persons level ratings,” a Nielsen spokesperson told MDN in September, adding, “We've been working on this with the industry for nearly two years.”
A Nielsen spokesperson said she was not available to comment on the new announcement at presstime, but Nielsen previously said it would release so-called “impact data” about the new method before it rolled the method out as official currency for TV advertising buys and programming decisions.
It was not clear at presstime who or how many clients may have seen that data, but at least one long-time Nielsen client characterized the move as “ill conceived and certain to do more harm than good.
“Making up viewing estimates is always a bad idea.”
While mathematical modeling is far from a new concept in Nielsen’s ratings -- the researcher routinely weights under-represented segments of its audiences mathematically to bring them into balance with the audiences it is able to capture data for -- the new viewer assignment method takes it to a new level, because it is treating the behaviors of people in one sample -- the national people meter respondents -- in a way that is representative of viewers in non-people meter households.
17 comments about "Nielsen To Roll Out Controversial 'Viewer Assignment' Method, Will Model Audience Estimates In Non-People Meter Homes".
Read more comments >
Next story loading 
The system Nielsen has been using for decades is sufficiently flawed that this might be an improvement. Maybe.
While no one is really happy to see this new plan implemented for mid-sized and smaller markets, the fact remains that the stations in such markets believe that they can't afford to pay for the much larger single-source peoplemeter type samples that are required to produce more "stable" viewer ratings in their home areas. The stations have a point, as there is little indication that national spot ad dollars, which are traditionally scarce in these markets, will support
the added expenditure for "better" rating surveys. I see Nielsen's plan as a possible---perhaps likely-----forerunner of its eventual move into "big data" ratings on a national basis. The existing big data rating operations only measure set usage----which is very misleading when it comes to targeting consumers-----and Nielsen needs much larger samples to rate the small sized, emerging viewing platforms as well as the fragmented media world, generally. So, why not link up with a big data entity and apply national peoplemeter viewer-per- home ratios to the big data set usage findings to produce more "stable" audience estimates for network TV, syndication, cable, Netflix, etc.?
Ed, you have just described the future. Stay tuned.
If what we read is true, the Nielsen National Sample Expansion Plan is ill-conceived and certain to do more harm than good. Making-up viewing estimates is always and everywhere a bad idea. Nielsen estimates need to be tabulated not formulated. To call fabricated Nielsen Ratings the industry currency, we seem to have achieved the ultimate self-deception. That MRC Accreditation and E&Y Auditing is not a pre-condition for implementation are indicative of how confused & mistaken all the players are. Further, Nielsen's claim that they are seriously improving stability (instead of validity) is apparently a sham. To reduce the SE (standard error of measurement) by half one would need to quadruple sample size. Is that happening? I don't think so.
As Spring training starts, Nielsen should be called out after two strikes: one against validity; the other against reliability. Omitting the MRC means that in the end Nielsen has already fouled out.
Perhaps it's time for government intervention through the FTC, Fcc and Congressional Hearings. Why would Nielsen denigrate a system that has worked well since 1987. And who but the Clients will pay for this madness. Onwards & upwards!
Academic & Cowboy perspectives do no one any good.
The only qualified commentator to this point is Ed Papazian ... And hopefully, yours truly.
We are dealing with matters of science, not ersatz statistics & delusional sorcery.
I'm curious as to how nielsen can "model" local news and still get buy-in from local broadcasters. Broadcasters seem to feel that their local news is different and unique and can't be compared to national trends, The only thing i can see is using homes from the NPM sample but that NPM sample in local (non-lpm) markets is non-existent in some cases and very small in others.
William, I haven't seen the Nielsen "plan" and, therefore, this is speculation on my part, however, the National peoplemeter panel measures all in-home TV usage, regardless of whether it involves local or national program content. accordingly, the local news is measured, and the plan, if it utilizes national Peoplemeter data, will, no doubt cull such data on a regional basis. So, if your a local station operating in New England, Nielsen may develop viewer-per-set ratios from about 1500 panel homes in this region and apply them to local newscasts in a small market like Bangor Maine. I assume that the actual plan is much more sophisticated that that and uses whatever local market rating data that is available---such as from Boston, where viewing as well as tuning information is available. I also assume that there are many layerings and weightings involved to make this process seem more palatable to research purists, like myself and Nick, who can't be happy with this development. What I'd like to see----and, maybe Nielsen has done it-----is a comparison of the proposed plan's viewer ratings when executed in major peoplemeter markets, with actual peoplemeter findings in the same markets. This would give a pretty good read regarding the real world validity of the proposal.
OOps!. I made a few typos, above. Joe, isn't there some way to give us posters an editing option----please. And let us do paragraphs-----pretty please. Just asking----again.
I endorse the Papazian proposal for an edit option. Perhaps it could be limited in time availability (e.g., 24 hours). I have had auto-correct options make the right words wrong inadvertently.
Thank you for your consideration.
I could have swore I talk about sampling recently with John and now this shows up. Wow.
Note to Nielsen regarding their powers of transformation: Modeling Set-Meter Data in Set-Meter Markets does not transform Set-Meter Markets into People Meter Markets, as a Senior Nielsen Officer claimed publicly today. More simply, calling a dog's tail a leg doesn't make it one, or give a dog five legs. Nielsen's self-deception, dissembling or both on this point constitutes a breach of intellectual integrity and perhaps worse. Like Brian Williams, Nielsen must misremember its origins in the scientific method and its fiduciary responsibility to deliver the truth to its clients and business partners. Nielsen: It's time for measurement principles & research quality. Better is possible. So much better. Onwards & upwards.
@Ed Papazian, we're looking into adding editing privileges, but in the meantime, if you email me at joe@mediapost.com, I will update your comments for you. Sorry for the delay. -- Joe
Glad to hear this, Joe. It will be a major improvement and a big help. As for correcting my typos, please don't bother in this case, I think the readers know what I was trying to say. Thanks. Ed
1. Yeah Benny .... spooky isn't it. Seriously, I had no idea that this issue was going to raise its head again. 2. Joe a suggestion. Rather than edit something AFTER the commenter has posted, why not do what other sites do and 'overlay' the commenters screen with a 'pop-up' of what they just typed would look like. Of course this would need to be editable. So at the bottom of the current screen you have 'Preview Comment'. The screen pops up (complete with layout and paragraphs of course) and the commenter gets a chance for a 'final read', then at the bottom of the pop-up you have your Submit Comment button. 3. We face similar problems here in Australia with small regional markets (small in population but vast in geography) that roughly equate to your local market data. They number in the dozens rather than the hundreds but all markets are people-meter markets - none are set-meter. We are investigating ways of doing exactly what Ed outlined. The objective is to use as much tuning data as possible (i.e. must pass all existing validation and edit checks) to produce the most robust set of HH ratings (lowest SE) that we can, and to reduce any potential bias unintentionally introduced by small samples. We then use the people viewing records (i.e. people-meter data) to produce demographic ratings so basically following the existing procedures. Of course there are massive issues such as whether 'pooling' across geography forces regression to the mean that doesn't reflect local nuances (yes), whether the local sample is large enough to produce the depth of demographic coverage required by the market (in general, no), whether the demographic profiles by programme when 'pooled' show the expected clustering and reduced SE (sometimes), what do you do when the above don't happen (broaden the geographic regions reported, reduce the demographics, increase the span of time over which data is calculated and average it - all very unpalatable). It is an interesting journey but one that needs to be investigated with the utmost rigour.
Realtime Nielsen Progress: This afternoon (02.19.15)
Nielsen SVP of Research, Mr. Bruce Hoynoski, has clarified the research company's position on the MRC role in methodological change measurement. To his credit, Mr. Hoynoski indicated unequivocally that the major methodological changes announced by Nielsen at Noon (i.e., National & Local Sample Expansion) will take place with the prior auditing and accreditation process offered by the MRC (Media Rating Council). This statement is consistent with those made earlier by George Ivie, the MRC's Executive Director. This commentator considers such commitments to due diligence and due process a step forward in national research quality planning. Thank you, Bruce.
That is welcome news, Nick. There are still many issues to be dealt with, as we both know, but an objective and independent watchdog is always a plus when we enter uncharted waters.
I should also note that the most common forms of "local" station programming are news by far and, secondly, syndicated fare. In the latter case, a typical major network affiliate will run some syndicated talk shows in the daytime, with a few courtroom shows mixed in. Also, there is prime access programming which consists mostly of game and gossipy newsmagazines. A typical independent station, may air talk, game and courtroom shows in the daytime but, often, switches to off-network sitcoms and, sometimes, dramas, in the early and late evening hours. As a rule, there is great similarity between the demos of most local newscasts and; also, between the main syndicated entries.So, the news "audience" gravitates to news; the sitcom audience to sitcoms, etc. You almost never find one newscast catering mostly to adults aged 60+ while a rival channel's news reaches mostly the 18-34 crowd. The median ages of most newscasts are almost always in the 55-65 year range. The 18-34s are mostly watching sitcoms or other fare.The same degree of similarity applies for shows like "Wheel of Fortune" and Jeopardy" or "Entertainment Tonight" and "Access Hollywood". So, a well thought out "attribution" approach does not have as many "demographic"risks for a typical broadcast station as it does for the far more diverse programming menus offered by cable. One additional thing to remember is that similarities in program genre appeals, notwithstanding, In very small markets, where one or two broadcast stations may compete with news while another goes with off-network sitcoms, the odds that regional generalities apply are more suspect. As a result, it may be possible for one station's news to slant considerably older, than is "normal" while a rival channel, which features a sitcom, may draw a higher than "normal" proportion of 18-34s, in both cases because the local broadcast audience, at least, becomes hyper polarized due to lack of alternative options. I wonder how the Nielsen plan will handle such situations?