Abstract:

The author has conducted analyses of viewing patterns among respondents within households in a test panel of peoplemeter households in the United States, investigating how predictable and habitual viewing behavior is within a household. Pattern analysis suggests that viewing is well-structured within households by time and set and this structure is an effective discriminator between respondents within a peoplemeter household. This preliminary pattern analysis indicates that 'intelligent meters' could be developed that learned the habits of peoplemeter respondents after as little as a month's learning and these intelligent meters could use this information to control the delivery of prompts. A heuristic algorithm, by which peoplemeter prompting is controlled by meter uncertainty is discussed by the author. Such algorithms are called heuristic algorithms, as a heuristic system is a learning system. Such heuristic algorithms might be expected to work better in smaller households then large households where the number of potential respondents among whom the heuristic must discriminate is greater. Research suggests this may not be the case, as large households are often characterized by more sets and more variation in demographically driven viewing patterns.