What's in a Number?
They operative phrase here is “its estimates.”
Arbitron does not provide audience estimates for national programs. NPR and other radio networks create their own numbers from the Arbitron data they buy.
It appears the process the networks use to crunch their own numbers overstates the size of their weekly audiences. There is strong evidence that the National Public Radio network Cume is overstated by as much as 15 percent.
The problem may not be limited to public radio as most commercial radio networks use the same process for calculating for their national audience estimates. There are not enough publicly available data to know.
Network audience estimates can come from several sources of Arbitron audience data. This article focuses on two of those sources: the National Regional Database (NRD) and the network version of Nationwide.
When Cume audience estimates for the exact same public radio network line-up of stations are created in each database, the Nationwide produces an estimate that is 15 percent larger than the estimate from the NRD.
Here’s how that happens.
The networks buy Arbitron’s Nationwide database, license software to analyze it, and add up the audiences for the network stations.
Here's where it gets tricky.
Some people listen to more than one station in the network. In public radio, some people listen to the same network program (Morning Edition or Car Talk, for example) on more than one station. In the network version of Nationwide, these listeners get counted twice when they should be counted just once.
This creates the need to erase the double counting. The audience estimate software providers work very hard to create algorithms that account for listener duplication. The problem is that the algorithms don’t erase enough of that duplication, at least for public radio stations.
Arbitron’s NRD software is designed to eliminate the double counting problem from the outset. That’s one of its most important features. The NRD produces a lower Cume audience estimate than Nationwide for the exact same line-up of public radio stations.
Still, the public radio networks use the higher Nationwide numbers.
The networks justify this because the software used to analyze Nationwide is easy to use and is an important part of the “currency” in network advertising sales. The networks use the bigger number because they fear that switching to the NRD estimates would cause them to lose millions in sales revenues to competitors using overstated numbers. And I was recently told by one public radio network executive that, “everyone knows about this is okay with it.”
Not everyone, apparently.
Shortly after I started researching this topic, Arbitron announced it was working with its clients and vendors to develop a process for conforming all audience estimates across its databases and products.
Good for them for responding. Sometime in the next year, NRD and Nationwide will deliver the same Cume estimate for the same line-up of stations. You can expect a few million “listeners” to disappear from the public radio networks’ audiences when that happens.
But Arbitron’s fix addresses just part of the issue.
Today, there is a high level of transparency and accountability in radio station audience estimates. Anyone can buy the book and look up a station’s numbers. Arbitron, correctly so, is looking to increase that accountability with its new Portable People Meter (PPM).
That same level of accountability does not exist for network audience estimates created through the Nationwide. It is a world built on the concept of “trust, but don’t verify.”
That has to change. It just makes sense that what’s good for individual station audience estimates would apply to network audience estimates.
I’ll start the campaign for an independent, objective source of national network audience estimates.
Allowing radio networks to control the conversation about the size of their audiences is unacceptable in this age of shrinking audiences, greater competition for ad revenue, and higher expectations for transparency and accountability.