TNS vs Nielsen Media Research

I recently read of a deal between TNS Media and DirecTV which allows TNS to create an opt-in pool of 100,000 DirecTV users to monitor their viewing habits. Recording directly from the DirectTV box, TNS will be able to compete with Nielsen and deliver second-by-second viewing ratings. I’m always in favor of more research, particularly cutting edge digital measurement that is finally catching up to the way people actually consume media (will those antique Nielsen diaries ever disappear?). Right off the bat, I had a few questions and concerns about this deal.

Their sample size is probably not representative of the overall US demographics. For one thing, utilizing only DirectTV participants excludes great segments of the population that are not satellite consumers, so their sample is most likely skewed to upper income homes. Many people ask me why Nielsen doesn't simply tie into their cable box and record from there. This seems like a simple, quick solution, but in actuality is more complex. For one thing, roughly 10% of homes do not have cable or satellite, getting their signal over the air via antenna only. This is a particularly large segment of homes to be ignoring.

The sample they have is an opt-in sample, verses a selected sample that balances to universe estimates. By default, generally in opt-in samples, you get a skewed group of people who have high regard for television in general and want to share their voice. It goes back to a simple theory of research – people are more willing to participate in surveys about things they like. (i.e. People who like to cook will participate in cooking surveys. People who never cook won’t waste their time on the survey.) This naturally skews the survey in favor of these high-television viewers. This is a problem we had in Tampa, until Nielsen changed the methodology, allowing more non-tv viewers to participate. All of a sudden, we saw viewing levels drop across the board, and it was mainly due to the fact that there are more people in the sample that watch comparatively less television.

TNS is making a big deal about their sample size being 100,000 homes, but in statistics, big is not necessarily always better. You generally come to a point where the sample size stabilizes, and getting a bigger sample won’t statistically affect the results. I haven’t the statistical analytical skills or the necessary data to run the equations to figure out what this optimum sample size is for the whole US, but I’d be interested in seeing whether 100,000 homes really matters that much more than Nielsen’s 14,000 homes, in terms of the statistical results.

TNS has a long way to go before they could compete with Nielsen on a national level. They are positioning this new service as something that “fills in the gaps” of Nielsen research, not necessarily directly competing. A sample size of DirectTV homes might give a market some insight on consumer viewing behavior, but the market would never accept this sample size as a source for overall, end-all ratings, due to the sample bias towards DirectTV homes.

I was very interested to read that TNS’ new service was able to see that although popular thought is that all sporting events are watched live, there was some DVR recording and playback of the Superbowl and its commercials. This type of research gives validity to the DVR ratings released by Nielsen and will hopefully give us some ammo to fight against advertisers who are convinced that everybody in the world is fast forwarding through every single commercial on television.

Second by second ratings?! Wow, my head spins thinking about it! The good thing, at least for me right now, is that none of this will filter down to the local level for a long time. National agencies and broadcasters will argue over minute-by-minute versus second-by-second ratings, but my local agencies are still struggling with basic concepts of Local People Meters and DVR ratings. It has always been my contention that the national research firms (particularly Nielsen) do not do a good enough job educating the smaller local agencies about methodology, and it is the local television stations that bear the brunt of educating the market. I shudder to think of the advertisers' facings staring blankly at me when I try to explain second-by second ratings in the future.