There is now data backing up television networks’ suspicion that COVID-era adjustments caused Nielsen to undercount the ratings.
A Media Rating Council audit of Nielsen data found that the company undercounted total use of television among adults 18-49 by an average of 5% during the month of February, the first concrete evidence of what the networks have been alleging for months — that Nielsen has failed to maintain the quality of its panel due to COVID-related changes in its measurement process.
The CEO of the Video Advertising Bureau — a trade organization representing the networks — called the findings “just the tip of the iceberg.”
The VAB has publicly accused Nielsen of failing to send field agents into panelists’ homes to ensure equipment was working properly and failing to pull homes from its panel that were no longer occupied. MRC findings show that nearly a quarter of the homes in Nielsen’s panel — 9,400 of 40,000 — had issues that in ordinary times would have justified being pulled from Nielsen’s sample. Of those, 2,400 (6%) triggered “high priority” alerts, meaning that they had issues that “would affect Nielsen’s ability to collect complete and accurate viewing data from the home.”
While the majority of programs were unaffected by the undercount — the MRC says that 93% of programs remained within .02 ratings points of original estimates — it is worth noting that the majority of television programs attract negligible viewership. It is fair to wonder whether the big ticket programs, such as live sporting events and specials, or even regular primetime network shows, would rank among the 7% suffering more severe swings.
It is also fair to wonder whether the undercount would have been more pronounced in previous months and whether it was more pronounced in other demos.
Nielsen in March began to resume in-person field operations, but said last week that it is still working to fix issues in 2,700 homes — work that will likely not be complete until the end of this month.
[MRC report via TV Technology 5.11; Adweek 5.10]
