Clients Weigh In On Nielsen Plan To Give Less 'Weight' To Internet Households

Nielsen will huddle with clients this morning to discuss a controversial plan to begin mathematically adjusting its TV ratings estimates this fall based on several new factors, including the number of households in its sample that have computers with broadband Internet access. The adjustments, known in the parlance of the TV ratings business as "weighting controls," are designed to make Nielsen's audience estimates more representative of the actual TV universe by giving greater mathematical weight to viewers who are underrepresented in its sample, or giving less weight to those who are overrepresented. Based on estimates Nielsen began providing clients over the past week, its national TV ratings sample currently over-represents households that have computers with Internet access, and under-represents those that have no computers, and some clients are concerned that the changes will be made before their impact on the TV industry can be fully understood.

While Nielsen has been reviewing and altering its weighting controls annually for years, it normally gives its clients plenty of time - at least a full year - to evaluate the changes before it makes them. The discussion surrounding this year's changes are based on only a few weeks of data, and some clients fear Nielsen is giving them the bum's rush, because the changes impact other parts of its business plans, especially its decision to begin reporting online viewing of TV programming in its ratings this fall.

advertisement

advertisement

For their part, Nielsen executives say they have made no decision to implement the changes, and will only make the adjustments if their clients approve them. They also say the actual impact on TV ratings for specific television networks will likely be negligible, and that the biggest reason for implementing the controls now is to give Nielsen the ability to recalibrate its ratings when there is a big shift in the number of households that have new computers with Internet access. Pat McDonough, the Nielsen executive overseeing the process, says that usually occurs during the Christmas holidays, and Nielsen spends a few weeks adjusting its ratings based on a variety of new consumer electronics equipment, including computers and broadband access.

McDonough said the timing surrounding the new weighting controls is unlike any Nielsen has implemented in the past, because it only just compiled universe estimates for households with computers connected to the Internet, and it will begin reporting TV viewing via the Internet this fall.

According to the data Nielsen has begun circulating with its clients, 69.7% of U.S. households now have a computer with broadband Internet access. Based on its current sample, Nielsen's panel over-represents those households by 1.7%, and under-represents households with no computers by 0.8%.

"The changes in ratings are really, really small," McDonough said of the impact those weighting controls would have on the ratings TV networks use to sell billions of advertising to agencies and advertisers.

She estimated that the rating adjustments created by those weights would ultimately compute to "thousandths" of a ratings point, or numbers so small that they wouldn't event impact the statistical rounding that goes into the numbers Nielsen reports.

That is small comfort for some Nielsen clients who are concerned that the ultimate, long-term impact of the weighting controls could be more significant, especially for networks that skew especially high or low with viewers living in households with or without Internet access. They also note that those estimates are based on only a week or two of data, vs. 52-weeks of data Nielsen normally gives its clients before making such changes.

"Even fractional declines in ratings can become costly over time," one Nielsen client notes. "Networks whose audiences are older are more likely to experience negative changes. Conversely, the younger the age profile of network the more likely the change is to be neutral or positive. In the final analysis, a measurement company does not weight unless its wants to see audience data change in some way because its sample is out of alignment."

Another concern among clients is that Nielsen's new estimates for Internet-connected households may not be accurate, and that it is not being benchmarked off of some reputable, objective, third-party source such as the U.S. Census.

"It is unclear how Nielsen collects some the information that goes into developing its sample universe estimates," said the Nielsen client, adding, "Full disclosure is essential."

7 comments about "Clients Weigh In On Nielsen Plan To Give Less 'Weight' To Internet Households".
Check to receive email when comments are posted.
  1. Douglas Ferguson from College of Charleston, July 27, 2010 at 9:02 a.m.

    Since when is greater accuracy and representativeness so controversial? Oh, that's right. Unnamed clients worry, which is all that most clients know how to do. I would worry more that ratings are less accurate and less representative. And I would wonder how many of these unnamed clients are really trying to protect the likely-undeserved advantage they presently hold.

  2. Andrew Crowley from ggiyt, July 27, 2010 at 9:13 a.m.

    to say "it is unclear how Nielsen collects data" is an understatement. The ratings bible has to be re-written.

  3. John Grono from GAP Research, July 27, 2010 at 10:04 a.m.

    This is a responsible and prudent action. Here in Australia were are undertaking a "determinants of viewing" CHAID analysis in our regional markets to understand what really drives viewing levels. If variables like PVR penetration, broadband penetration or number of home PCs are shown to have discrete correlations (uncovered by the other variables such as number of TVs, household structure etc) then they will be added. Of course, not all of these variables are accurately benchmarked via a Census (and we have five-year Censuses here) then an Establishment Survey - generally 10 times the sample size - is conducted to establish the overall PVR penetration rate etc against which to calibrate the panel. This is commonplace practice around the world, and I would be stunned if US Nielsen aren't meeting or exceeding these criteria. I would go so far to say that I am 95% confident that they are.

    If such procedures are clear to the marketplace here in Australia I wonder (assuming that such procedures are also conducted in the US) why they are unclear to the US marketplace. Maybe it says more about the US marketplace than Nielsen US.

  4. Walter Hammock from Showcase Enterprises, July 27, 2010 at 10:31 a.m.

    Ratings are a joke and have been for the last thirty years. Nielsen has little desire for accuracy. Everything is simply lip service.

  5. Jonathan Mirow from BroadbandVideo, Inc., July 27, 2010 at 1:12 p.m.

    Lies, damn lies and statistics. Guess we have to add "Nielsen Ratings" to this list...

  6. Rob Frydlewicz from DentsuAegis, July 27, 2010 at 4:49 p.m.

    If hundreds of millions are guided by the fairytales of the Bible I have no problem with billions of dollars of ad time being guided by Nielsen. And I think Nielsen ratings are more believable.

  7. Jim Blundell from Survey Audits Pty Ltd, July 28, 2010 at 5:57 p.m.

    Weighting is a complex business. It would be nice to correct sample imbalance across the widest range of population characteristics that are relevant to TV viewing. There is a practical limit to how many variables and which ones can be included in any weighting system. The most important and their interaction with one another must first be determined by detailed analysis. Having broadband internet in the household may nowadays qualify as important enough for inclusion but the question still remains - how to best include this in the mix of weighting variables.

Next story loading loading..