The IPA has undertaken some comparative work between the old and the new BARB data, comparing coverage and the effect of the Negative Binomial Distribution calculation.
A pool of almost 300 actual schedules (or parts of schedules) running during July was analysed for the IPA by DDS on both the old panel using the old methodology and the new panel with the new methodology.
The data was inevitably restricted to the time period,and thus some campaigns were very small. Others, however, were up to 500 TVRs. These schedules were analysed for both simple (1+) coverage and for 4+ coverage. In the case of the new panel runs, the analysis was done both with and without the NBD option.
Each of the schedules was analysed against its own target audience, which in most cases was housewives.
The average of the schedules was: Old Panel Coverage – 62.2% New Panel Coverage – 62.0% without NBD New Panel Coverage – 63.9% with NBD
Thus the new panel measures a slightlylower coverage than the old if the NBD is not used, but if the NBD is used (as properly it should be) the coverage is measured slightly higher – in fact 2.7% higher.
Not surprisingly, this figure varies with the size of the schedule: Up to 100 TVR +7.5% 100 TVR-200 TVR +3.6% 200 TVR-300 TVR +2.4% 300 TVR-400 TVR +2.7% 400 TVR-500 TVR +1.5% 500 TVR or more +0.3% All Schedules +2.7%
Over all of the schedules, the new panel measures TVRs about 10% higher than the old panel. On an approximate calculation this should produce an increase in coverage of about 2.5% to 3%. Thus the extra coverage appears to be entirely the result of the increased TVRs.
Both systems exhibit a very similar relationship between coverage and TVRs.
This conclusion holds for all of the schedule size categories shown, but there are inevitably some small variations by area.
The data for 4+ coverage produces much the same conclusion: Old Panel Coverage – 23.9% New Panel Coverage – 23.5% without NBD New Panel Coverage – 25.8% with NBD
In fact, in this case the expected 4+ coverage given the change in TVRs is about 26.5%, suggesting that the new panel might be estimating 4+ coverage a little lower than the old panel did for any number of TVRs.
