Tracking Sample Characteristics

So, I've spent the morning tracking our market's metered sample characteristics that are NOT balanced to Universe Estimates to try to estimate how much of an affect these have on our ratings. Nielsen sample balances the ratings to a few demographics, including age, race, presence of cable/ADS, presence of children and county.

But I feel these limited demographics are simply not enough to truly represent the market. Although I can receive information about the income, education, occupation, precense of DVR, HD and home phone lines in our sample, these characteristics are not balanced to universe estimates. A 35-year old Caucasion person who is college educated, with a DVR, and a household income $75,000 plus is going to watch television much differently than a 35-year old Caucasion person who has only a high school diploma and a HHI of under $10,000.

Demographics alone do cannot predict behavior. Consider this case study: Two men, both white, both older than age 50, both with household incomes above $100,000, both are homeowners with children. Both are married, college-educated and have professional occupations. According to demographic, qualitative research, we would assume both men would watch television, surf the internet, consume other media, and purchase similar products. Essentially, demographically, they are the same.

Who are the two men? George Bush and Howard Stern.