Combining successive waves of magazine readership data, or creating doublebases have been common practices for the purpose of providing more stable data than would be produced by a smaller sample of data. An MRI doublebase consists of two studies of approximately 20,000 adults each, and therefore about 40,000 adults. Each study is based on combining two survey waves of approximately 10,000 adults each.
However, the combining of samples has traditionally been based on equal weighting—which has to be considered arbitrary.
The differences observed in two successive report periods of readership data may be partially due to sampling variation, but may be partly due to actual trends in readership for at least some magazines.
Our hypothesis is therefore that combining data that gives more weight to the more recent report period may produce averages that are more predictive of the next set of readership estimates. If so, this would represent a very useful finding for media planners and buyers, since they generally look at readership data for the purpose of estimating a future schedule. If it turned out that 50/50 weighting could not be improved upon, this would serve to support the current practice of doublebase.
Symposium: 1999: Florence, Session 6 - Refining Our Models
Authors: Baim, Julian, Frankel, Martin, Gray, Stuart
Organisations: BBDO, Mediamark Research Inc
Topics: Analysis Issues