A Group Decision Making Approach to Model Household TV Channel Choice.

详细信息   

• 作者：Su ; Lei.
• 学历：Doctor
• 年：2011
• 关键词：viewing choice modeling ; television rating ; group
• 导师：Kwong, Jessica Yuk-yee,eadvisorLau, Kin-nam,eadvisorJia, Jian-min,eadvisor
• 毕业院校：The Chinese University of Hong Kong
• ISBN：9781267099259
• CBH：3491974
• Country：China
• 语种：English
• FileSize：9587319
• Pages：236

文摘

An accurate television viewing choice model is an important tool for television industry executives, as well as advertisers. An efficient model can help television channels maximize ratings by improving both scheduling and the characteristics of their shows. On the other hand advertisers can predict ratings and demographic composition of audiences with better accuracy. Though there is considerable evidence to suggest that individual viewing choices are strongly affected by ones family members, quantitative models in marketing literature typically focus on the individual as the unit of analysis without incorporating the influence of family members. This thesis proposes a three-stage model to capture the process of household television viewing behavior. We divide the household viewing process into three sequential and interrelated decision stages pre-decision, joint decision, and final-decision) according to the group decision making framework suggested in prior research. By defining utilities of different programme types on different channels, and weighting parameters of each family member, each family members three decisions pre-decision, joint decision and post-decision) are modeled as a function of these parameters with three sub-models. The model was estimated with maximum likelihood estimation, duly validated with simulation studies. Meanwhile, the model was extended to be time-dependent to allow past viewing history to influence current viewing choice, and applied on the people meter data for primetime telecasts on weekdays for the whole of 2006. The results indicate that our model has better prediction accuracy compared with models being currently used Rust and Alpert 1984； Yang et al. 2010). Furthermore, we are able to demonstrate that models that ignore the influence of family members yield biased estimates. Our model also has better prediction accuracy compared with the traditional model proposed by Rust and Alpert 1984), and has more flexibility to fit households with different compositions. Finally, we find that there exist different household decision structures, initial latent preferences, and influences of past viewing history across different families and their members, and the heterogeneity can be explained by demographic variables.