Our partners deserve the highest standards of transparency in understanding how we measure TV audiences to give them confidence in the decisions they are making to research, measure, and optimize their campaigns around our insights. Comscore has for more than two decades provided detailed information about our massive reporting samples, the calibrations for data representation, and their impact. Now we’ve taken that even further.
Demonstrating how our reporting samples offer high levels of stability of our TV measurement estimates, Comscore has now introduced monthly standard error reports to our currency-grade TV measurement product, Comscore TV. These reports cover all national networks and locally reported broadcast stations, across all 210 Comscore markets, offering additional insights into the reliability and stability of our TV audience measurement.
Standard error is an established statistical concept for quantifying the expected variation in measurements caused by the size of the input data set. In the context of how this impacts TV measurement, standard error helps explain two things:
(1) How our viewing estimates (based on a large sample) compare to actual viewing of the entire population, and
(2) How stable Comscore TV metrics are with respect to statistical fluctuations.
With a level of accuracy that already sets us apart in the industry, like our 48-hour Pulse data which provides 93% predictability to enable confident campaign optimization across all 210 markets, we’re able to address both the questions above in a new level of reporting founded on the vast scale of our data set.
Big data sets allow us to lean into the "law of large numbers", which says that the more units you sample, the more stable, reliable, and repeatable the corresponding metrics will be.
Imagine flipping a coin. If you flip it twice, you'd expect one heads and one tails, but getting two heads wouldn't be shocking. Now, imagine flipping the coin ten times. It would be surprising to get all heads, but it could happen. On average, you'd likely get around four to six heads. Flip the coin 100 times, and the average heads would be close to 50% – even more so with 1,000 flips. The more times you flip, the more consistent the ratio is; this is the law of large numbers. By analogy, the more households used in a media consumption measurement, the more consistent and stable the measurement will be. Comscore’s large data set allows for millions of measurements daily, and standard error is a measure of how that helps make our data reliable.
How Big Data Sets Drive Reliability in TV Audience Measurement
Passively collected linear viewing data powers the ability to measure consumer behavior at an extremely granular market level, while also ensuring (as demonstrated by these reports) that the viewing estimates provided are consistently stable due to the sample size used. In this context, the more data there is to work with, the better the estimates about viewing behavior are.
With approximately 1 in 3 households reporting viewing nationally, and 1 in 4 or better households reporting in over 150 Comscore local markets, no other measurement provider can match the scale of our data, and in turn, low variation rates in audience measurement.
Our new reports bring transparency to this. In September 2023, for example, the standard error for the top 100 networks was less than 0.2%. For all 259 national networks it was less than 1%, for Local broadcast stations less than 1% for over 1,500 stations and less than a 0.5% for over 750 stations.
The granularity provided by these new reports proves the reliability of our data in an additional measure of data transparency, enabling more confident decision-making, better audience targeting by advertisers and publishers and ultimately more effective campaign management.
I invite you to check out the standard error reports now available in Comscore TV for the market, station or network of your choice and see how our data continues to deliver accurate and reliable metrics to power insights across the linear TV ecosystem.
Please don’t hesitate to reach out to me at this link or your dedicated account manager with comments or questions.