SirRoxalot said:
Simple. The agencies, for whom radio buys ratings, wanted faster, electronic delivery with greater granularity and a newer methodology. Radio, under the "client is always right" truism, supported Arbitron's efforts to advance the meter concept it had been developing since around 1991.
And agencies badly wanted a measure of impressions, not memory. The got that, too.
One of the objectives of the PPM was to have frequent books, eliminating the lengthy 12-week diary survey period... a system that failed compared to instant hit and click data from the web and ovenights on TV. Thus the PPM could make radio very competitive on timliness. To do fast turn around surveys, the daily and weekly samples had to be larger than the very small and totally uneven ones in the dairy.
So we have daily and weekly samples that are larger than the monthly diary samples, and the ability to look at specific days and dayparts with the same accuracy as an entire book... something impossible to do with the diary.
The cost of the PPM raised subscriber fees by 60% on the average. Some dropped out, some complained about the high cost. Arbitron has made it plain that they will gladly add more sample if all subscribers in a market pay for the addition. So far, radio has not backed a sample increase of this kind.
Arbitron, who never did panel surveys before, had allied with Nielsen for the PPM. The earliest Philadelphia and Houston tests were done with Nielsen. The meetings I attended about the Philadelphia 2002 tests included as many Nielsen executives as Arbitron reps. But the alliance broke up.
And Arbitron had to go it alone. They selected a phone base sample frame for all markets except Houston, which was set up with Nielsen's system using an address based sample frame. Houston quickly won and has kept accreditation.
Arbitron, at considerable cost, is using address based recruiting as well as a cell-phone-only sample to improve the reliability of the sample itself. As this rolls out, they are getting more and more accreditations, including the #1 revenue market, LA.
As they resolve some DDI (response rate) issues inherent to panels, it is pretty likely all 48 markets will be accredited.
As to the earbud thing, I don't think that any subscriber uses the cumbersome passthrough device But since stream listening does not accrue to the broadcast station it originates from (except for NPR) the earbud thing is not relevat at present. What is needed is to have the PPM in a smartphone app, but the current battery and microphone technologies are not able to make that happen for some time to come.
That's the easy one: statistics is the only science where "error" is not a dirty word. Statistics measures probability by taking a sample of the universe under study, and projects into the entire population group.
An error of one or two percent can make a political poll show the "wrong winner" while in most marketing functions, such as radio, errors of that size or greater are eminently tolerable and in no way compromise the data.
"Who's on first?" is not the purpose of ratings. The purpose is to give a reliable estimate of "how many are on first at the same time." Being first or second or third is not very salable, as there is no way to evaluate comparative cost based on rank. So the battle is not about rank, but about rating. Having the right delivery of the client target demo at the right CPP is essential.
And since the metric used is rating, keep in mind that a .4 rating can cover, in most 6-Mid cases, stations with anything from a 3.6 to a 4.5 share, even though there is a difference of around 25% between the lowest and the highest of the possible 0.4 rating range base shares. That simply shows that agencies are not looking at little 0.3 or 0.5 share differences... and it also shows that all the degree of precision some are looking for is totally obfuscated by the buying process.
