Modeled VPVH, 0.44; peoplemeter sample VPVH, 0.38; difference, 16 percent.
The VPVH estimate for the two-and-more member households containing the demo member (step 4) is the largest source of potential error.
One way to do this is to start with the national demo VPVH value among two-plus-member demo households for the time period (peoplemeter data) and adjust that value up using the set-meter panel comp of demo households tuned to the program, divided by the set-meter panel comp of demo households tuned to the time period.
The key insight is variation in VPVH for a viewer demo will be seen in variation in the demo composition of the group of households tuned to the program.
A high Male 18-34 VPVH for a program like Survivors will be signaled by more households with a Male 18-34 in residence tuned to Survivors.
Large set-meter panels using independent peoplemeter VPVH values for viewer modeling promise to produce better ratings for the money than the current systems.
(4.) The range in VPVH for programs is from 1 to about 2.5.
(6.) For an ingenious method of collected VPVH from a setmeter panel, see Baniel and Ephrou, 1999.
(10.) Monday Night Football appears to attract a disproportionate number of households with more than one 18 to 49 year-old man in residence, This can be accommodated by modeling VPVH for households with one and with two-or- more target members, separately.
(12.) Even if a constant VPVH value is used, the number of local demo households tuned to a program will vary in each market, which allows the set-meter tuning measurement to capture variation in demo audience, just as a peoplemeter panel would.
These examples suggest that our modeled VPVHs are within the sampling error range of a 1,666-household-peoplemeter panel and should be statistically indistinguishable from measured data.