Nielsen has defended its ratings measurements over the last year, saying that the coronavirus pandemic amplified changes in media behaviour in the United States but many of the trends were already established. It follows industry criticism of systematic undercounting since Nielsen lost a fifth of its measurement panel. The research company concedes that its audience estimates are subject to an increase in margin of error but insists that this is not material.

Viewing of live and time-shifted television among adults in the United States fell from 4 hours 14 minutes in the third quarter of 2018 to 3 hours 56 minutes in the same quarter of 2019 to 3 hours 41 minutes in the same period of 2020.

Television advertising advocacy group the VAB claims that television usage was undercounted during the pandemic as the number of households supplying usable information from the Nielsen panel fell by 20%.

The average weekly reach of total television usage fell from 92% in 2019 to 89% in 2020 and fell further to 87% in the first weeks of 2021, as reported by Nielsen.

VAB said that the reported decline in television usage in 2020 was contrary to a mountain of other data. It suggested that an increase in the number of people reported to have watched no television in a week reflected the number of empty or non-responding homes in the Nielsen panel. It said these included homes with more than four people and those with more than three television sets, while Black and Hispanic homes lost a quarter of their representation within the measurement panel.

Sean Cunningham, the chief executive of VAB, expressed incredulity at the accuracy of the Nielsen numbers.

“Nielsen has streaming declining among 18-34 in 2020 versus 2019. How could this possibly be true?” he said in a statement.

Nielsen paused all in-home activities relating to its media measurement from March 2020 and changed the way in which it managed its audience panel. While this resulted in a 20% reduction in the panel size, Nielsen maintains that it remained robust and representative. In any case, the effective sample size remains 90% larger than it was when it was expanded in 2014. It continues to be weighted to reflect national demographics.

Although total audience estimates will be subject to increased standard error, Nielsen says this is not material. Nevertheless, it will add a disclosure to its reporting for the period that figures are subject to an increase in standard error as sample sizes declined.

Nielsen responded that television viewership, as well as other media usage, spiked when people were obliged to stay at home, with increased viewing of news. After the initial shock of the pandemic, shifting media habits continued.

The pandemic cut programme production and side-lined sports events. The combination of more repeats, delayed premieres, programmes produced in homes and basements, led some to seek new programming. Nielsen reports that subscription video programming, news and nostalgic comedy programmes benefited, but the overarching shifts in total viewing were bigger than specific programme genres.

Between December 2019 and December 2020, total minutes streamed directly to the television screen, across streaming-capable homes, increased from 117.7 billion to 132 billion. People aged over 35 now make up the majority of the streaming audience in the United States, Nielsen said.

“The pandemic has both bottlenecked and accelerated aspects of the media industry, acting as a polarizing factor for consumer segments,” noted Nielsen. “Media consumption has been fragmenting for years, particularly as consumers become exposed to new options and platforms. Streaming options are an easy scapegoat for the recent declines in TV viewing but given that total TV viewing includes connected TV viewing, the reality is that viewing is fragmenting beyond the TV glass.”

In other words, Nielsen suggests that the television industry should not blame the measurement methodology but should take a look at itself to ask why people are not watching as much as they once did.

The case continues.