Friday, January 18, 2008

Radio Ratings Part 4: Arbitron

I'm trying to steer clear of the TV ratings systems that are ancillary to my whole radio theme: Telepulse, Videodex Trendex, Telerad, TPI ratings, etc please bear with me. With all the buzz going on between Hooper, Crossley and Nielsen you actually could miss that there were other smaller ratings companies selling equally legitimate products. This included; Sindlinger, Trendex, Inc., the American Research Bureau..and many others. Some of them used very similar methods, others basically branded themselves on the fine differences. The three survey methods were:
1. The Roster Recall Method
2. The Re
call method
3. The Coincidental Method

Take Sydney Roslow and the Pulse for example. They used face-to-face interviewing instead of phoners. The interviewees were selected through the same kind of random sampling as Hooper and Crossley. But instead of asking about the last hour or the last day, The Pulse asked them to name the radio stations the interviewee had listened to in the last day, and week! Interviewees were shown a roster of local radio station callsto help them remember. It was called the "roster-recall" method. It sucked. Under this methodology interviewers went from home to home interviewing. It was fast and cheap, and produces a much larger sample, up to 67,000 households, sixty times the size of some competitors. but the samples were less random and therefore less representative.

Sydney was president of The Pulse for 34 years. He went on to pioneer demographic research breaking our Hispanics early in the game. But in the 1970s after a decade of growth they began to lose clients to their competitors. The diary technique was more trusted, and the radio pool was too small for the number of services. They folded in 1978. More here.

Sindlinger and the American Research Bureau were all advocates of the diary method. Sample households are given a diary and asked to record all the radio programs they listen to in a week. On the upside it's cheap and produces a large sample. American Research managed 2,200 radio diaries. As an added bonus, the system also reached low-income households that lacked a phone. The phone surveys of hooper and Crossley eluded that demographic.

Sindlinger was fascinating in it's own right. Eighteen years after even CAB had dropped the recall method Albert Sindlinger brought it back from the dead in 1962. Pulse was using aided recall all throuout, but Sindlinger was using unaided recall alone. The president of ABC networks, Robert Pauley drove Sindlinger to break out the old, dead and debunked method. The reason was the previous to this Sindlinger was running an electronic monitoring system called RADOX but the system was ruled to infringe on one or more of Nielsens patents so proceeding with that platform woudl require paying a royalty to Nielsen.

ABC used the new pathetically innaccurate Sindlinger phone surveys to pick a fight with Nielsen over out-of-home listening. So Robert Pauley convinced Sindlinger to return to radio after being run off once already. The Sindlinger Radio Activity Service (RAS) used about 16,000 samples in his survey so it was at least more sound than the old Hooperatings. he also pioneered the use of computers. He was using an IBM 1620 to help crank out reports faster than Nielsen. ABC just liked it because the ratings were higer with Sindlinger than Nielsen. In the end NBC had enough of the malarkey and hired a NYU statistician to debunk Sindlinger. Af courswe even Albert himself knew it was all crap. In the book "Lies damned lies and Statistics he wrote:The American Research Bureau's system by comparison was pretty substantial. The diary system worked and is still relied upon today. It was founded by Jim Seiler in 1949, and became Arbitron in the 1960s. Arbitron still uses diaries and still knows they have flaws but it's far superior to all three recall methods and lacks the out-of-home listening problem that plagued Nielsen. Today Arbitron has pioneered the Personal People Meter (PPM) a portable device that records radio listening everywhere the subject may travel. It's only in a few markets so far and it has some demographic issues in its sampling but compared to all previous methods it is the most reliable by far. More PPM here.