SirRoxalot said:
David, you have considerable nerve calling someone else "contentious". Have you ever NOT carried an arguement to the point of ridiculous?
There is a point here: extrapolations, done by hand or via X-trends, yield only one possible result for each case. There is no guesstimate, no "my results are better than your results" in any case. Using the word "guess" is not correct, and conveys to those unfamiliar with the "math of Arbitron" a totally mistaken impression.
For those unfamiliar, extrapolations are done or created in software for phases I and II of each trend in continuously measured markets to get a very close firgure for the indivuidual month just measured.
For example, if we just got phase I of Fall, and got a 5 share 12+ for the 3-month rolling average, we can know almost exactly what we had in October by taking the actual months of August and September from the Maximiser Programmer's Package and subtracting the sum from the trend number multiplied by 3. Voila! We have September.
To see how close extraps are, we have NY and WCBS.
Extraps yielded 12+ 2.5 4.8 3.7 for the phases, while the actuals were 2.6, 4.5 and 4.1. In 25-54, the extraps showed 2.9 5.1 3.7 for the three phases, while the actuals, out just 10 minutes ago, showed 2.9 4.9 4.0 Very close, and very useful for seeing developong trends.
Arbitron does not support the use of extrapolation, nor to they guauarantee the accuracy, in a diary study, of single months as they represent only a part of a total sample, and can be influenced by things like under or oversampling in individual demo cells, things which are weighted for the full sample before release.
Thre you have, also, the story on CBS FM for the three months of Summer... lots of curiosity and binge listening in August, and off to a lower, but still respectable showing in September... 7th in 25-.54, several notches above the Jack showing in its best book...