As more markets roll out PPM data we are learning more about the new methodology or maybe it just adds to the confusion. One thing is very clear is the low sample base creates problems. Arbitron states and correctly so that there are more samples of radio listening with PPM than with the diary. However, the pool of people sampling radio has been reduced to around one third of the diary level. So a market that had 3000 diaries - 3000 different people sampling radio - now has 1000 meters. From experience we all know what even five or ten diaries meant to the end result. Arbitron proposes that because of the additional sample days the numbers are more consistent and less likely to 'bounce' or 'spike' as in the diary method. Logic says if you have the same person listening for 28 days by human nature they are probably going to have a routine which includes a pattern of radio listening. Especially when you compare it to four people sample seven days each.  However, it needs to be totally clear that the projections for rating listening is vastly different - using the previous example there is a significant difference between a one thousand and a three thousand person sample. If there isn't why hasn't Arbitron agreed to do such a study? Please don't give me the statistical bs on sampling - I studied mathematics in college even though that was a lifetime ago! Granted having one thousand panelists listening over 28 days is more accurate than three thousand diary keepers listening seven days but it is a different sampling set - it measures the usage of the one thousand panelists. The margin of error is still significant for the market because you have a sample of only one thousand versus three thousand to reflect the market. Where is it most noticeable - with stations that have a smaller core and cume base.

Then there is the whole other issue of distributing the existing one thousand panelists to represent the population accurately. In one market (that will remain anonymous) a single meter from a male 25-34 represented nearly 5300 men. In the same market a meter from a male 50-54 represented 2900 men. The same problem occurred with the diaries but not to this extreme because the sample base was much larger. 

So what is the end result? PPM is a better methodology than the diary; there is no question that. It is much better to have actual behavior rather than recall. The issue is the sample size which must be improved!