Larry Rosin: Kabrich PPM Conclusions Flawed
July 10, 2009 at 10:13 AM (PT)
After eyeballing the data crunching of the HOUSTON PPM by consultant RANDY KABRICH (NET NEWS, 6/29 and 7/1), EDISON MEDIA RESEARCH Pres. LARRY ROSIN wasn't completely sold on KABRICH's conclusions. So "I had one of the statisticians in our office run some statistical tests," he wrote on The Infinite Dial blog.
"First, we ran a correlation. The correlation between the Black 6+ sample size and the AQH reported for KMJQ is a .27. The correlation with the AQH for KBXX is .30. Correlations run from -1.0 to 1.0. A correlation as small as these is essentially no correlation. My eyeball estimate was correct.
"We also ran an Analysis of Variance test (ANOVA) and found that the sample size is not associated with the AQH for either station.
"Now, we believe that ARBITRON's clients should demand adequate sample sizes and should 'get what it pays for' from PPM, Diary, or anything else. But by the same token we need to be fair when we criticize. Mr. KABRICH's assertion that the results are "startling" is simply incorrect. To call them even 'suggestive' would be a stretch."