Nielsen’s Panel Expansion and National Ratings Modeling: What’s Happening?

You may have seen reports that Nielsen has expanded its US National People Meter panel and is using modeling in this expansion. What does this really mean? To understand the change that Nielsen has introduced, it’s useful to lay out the Nielsen TV measurement landscape in the US.

• Nielsen measures and reports national and local TV Audiences. There are 210 local TV markets, known as Designated Market Areas (DMA).

• National Audiences are measured using people meters (NPM = National People Meter sample). A sample of 20,000 homes across USA is recruited and all TV sets in these homes are electronically monitored. Individuals’ viewing is measured as well: people in the household register their viewing using a specially designed handset.

• For local audiences, there are four different measurement techniques – people meters, set meters, the new Code Readers and diaries.

Nielsen People Meter

• People Meters are used in the top 25 markets (ranked by population). These homes are used for both national and local reporting (LPM = Local People Meters) and are an integral part of the 20,000 NPM homes.

• For markets 26-56, samples of 800-400 homes are recruited in each market. Set meters are installed in these homes – these are like people meters but only the sets are metered, not individuals. Until recently, individual viewing was measured using diary sweeps in four months (Feb, May, Jul, Nov). This measurement technique was employed solely for local market reporting, completely separate from the NPM national people meter sample.

As of December 28, 2015, diaries have been replaced by a modeling of individual viewing (known as “Viewer Assignment”) onto the set meter homes’ data, using the household meter data and the NPM as the basis for this modeling. This will happen continuously, not just in sweep months. These modeled data are the same format as the people meter viewing records, enabling these set meter homes (about 12,500) to be included seamlessly in the NPM national sample. This is the major change that is being introduced to national measurement.

• For the smaller markets (rank 57- 210) diaries will continue to be used for local reporting in most markets, This affects local reporting, not national. For national reporting, these markets have a proportional number of people meter homes to ensure representation in the NPM national sample and this number has been increased with additional people meter homes recruited.

• Code Readers:  As of December 31, 2015, Nielsen launched Code Reader in 14 of the largest formerly diary only markets effectively replacing the diary.  The Code Reader is a passive meter, similar to the Set Meters with an easier installation process.  Because the Code Reader does not have button pushing requirements, Nielsen will employ its Viewer Assignment model to assign persons’ demographics as they do in the 31 set meter markets.With the introduction of the Code Reader, the diary has been discontinued in 45 local markets.  There are still diaries in the remaining smaller markets but there is electronic measurement in 70 markets today. The net effect of this and the inclusion of the set meter homes will increase the national gross sample from 20,000 homes to 35,000.

These changes will deliver more stable national ratings with fewer reported zero ratings and the overall effect has been to increase ratings for most networks. However, as with any change in currency measurement, it has been controversial. Chief complaints have been the timing (upfront deals were based on assumptions of lower ratings so campaigns running now have a measurement disconnect) and the fact that modeling of individuals viewing has been introduced.

My personal take on the modeling question is that all research is a model of the population. The key question is: does the new methodology provide a better estimate of US Television audiences? Without access to reams of comparison data it is difficult to be definitive but the answer to that question is probably “in most cases, yes.” The new measurement will deliver a significant improvement in stability, leading to fewer reported zeroes, which makes the business of buying and selling TV easier. This should be a worthwhile trade off against the small element of bias that may (or may not) be introduced by viewer assignment. The first year of a change in measurement is always bumpy and it will be interesting this time next year to see whether the industry feels it has benefitted from this change once the dust has settled and it’s business as usual.

Pete Doe is the Chief Research Officer at clypd.