I recently uncovered a fine peice of radio arcana regarding thsi more statistical side of radio. In order to estimate the number of listeners a radio staiton has in any given period of time, one must define "listener." Is that one song long? or the average legnth of one ad, about 30 seconds? Nope, actually the threshhold is currently set to five minutes.
The criteria for inclusion as a "listener" is five continuous minutes in one quarter-hour, per week recorded in an Arbitron diary. That sure isn't much usage, I will admit. Fortunatly for radio stations and their advertisers Arb diaries indicate a lot more individual radio usage than this bare minimum. But does PPM? that I do not know.
Their distant, but only real competitor Bridge ratings takes a harsh view on this "Bridge Ratings has no confidence in average-quarter-hour measurement nor its cousin time-spent-listening. The convoluted calculation required to generate such numbers raises the margin of error for such estimates beyond reliability. At Bridge Ratings more confident estimates rest with CUME and FAVORITENESS. Favoriteness simply seeks listenering behavior related to the station audiences listen to most often. Years of studies tell us that a listener devotes between 66 and 80 percent of their weekly listening to their favorite station. This statistic represents loyalty and when combined with cume reflects a station's ability to convert its cume to favoritness. " Favoriteness? Is that like Truthiness? http://www.bridgeratings.com/
Basically they avoid the traditional ratings schema entirely. they generate useful but unrelated quantizations of a stations listener dominance in a market. RAJAR in the UK has an interesting article on this as well: http://www.rajar.co.uk/documents/admap-paulkennedy.pdf