The dominant readership data collection method for over 2 decades has been â€˜Recent Readingâ€™ derived from a single interview survey. A Binomial convolution model (or some approximation thereto) is then used to calculate the usual schedule statistics (Net, Gross, average frequency/FD). This has required in addition the collection reading frequency. We do not challenge the utility of this readership model. We do say that in this complex multi-media age it is woefully inadequate.
There are 4 principal inadequacies:
1. Readership is modelled as occurring all at once.
2. No account is taken of the repeat reading of a single issue. Gross reach is underestimated for all magazines.
3. The measure reports issue exposure and not advertisement exposure
4. The independence assumption which underpins the Binomial Convolution model is never tested. In some cases it is clearly wrong (parent and supplements, daily reach of dailies).
Do these issues really matter?
Symposium: 2005: Prague, Session 5 - Technical Problems and Solutions
Authors: Masson, Peter J, Sumner, Paul
Organisations: Bucknull & Masson
Topics: Data Collection, Fusion and Modelling, Readership Model