Nielsen said the data is being reexamined by its statistical research team, because "a number of our samples had lower than usual in-tab levels for that day as a result of the server issue previously communicated," and the team needs to ensure that the local metered market ratings meet Nielsen's minimum standards.
Last week, Nielsen executives said a "firmware bug" led to a failure of the computer system used to collected data from its people meter ratings system. That glitch led to the delay of national TV ratings for the week, and it apparently affected ratings in the local metered markets as well.
Nielsen said the local metered ratings have already passed some key tests indicating that the data collected is representative of the markets and had valid usage levels. But after reviewing why many households fell out of its tabulations, Nielsen said its statisticians believe that "higher tuning households may have faulted disproportionately," and whether that might have contributed to a "bias" in the ratings.
"We are taking these extra steps as a precautionary measure," Nielsen said. "This is a unique situation. Typically low in-tab levels are caused by weather related events and therefore the standard sample quality testing that statistical research employs is geographic based. Accordingly, we decided that more testing based on additional criteria was appropriate."
Nielsen said that all of the testing would be completed by the end of today, and that if any metered markets fail, they will be excluded from future ratings reports issued by Nielsen.