Sample size and structure are inadequate for audience measurement

Sample size and structure are inadequate for audience measurement Sample size and structure are inadequate for audience measurement I feel that the survey samples and panels have been designed without the knowledge of data users. Very strangely, they are often structured as reduce models of the overall population. There should be an overrepresentation of the third of the population which aim the two-thirds of advertising campaigns, this is rarely the case. Much of the information collected is not used for much. In addition, the sizes of these samples appear to be really important tests that would like to make. How it comes that the designers of these systems do not being asked about the uses to be made of these data? Do not they have asked for help from statisticians? How it comes, there were so anxious to have very large samples, if uselessly representing every part of the population, including those not interested data users? Addition to the inadequacy of the sample structure, research manager and data users have reported to me that people who are recruited to participate in these surveys or panels are sometimes poorly trained and profiled: the minimum criteria, especially when compared to what offer the new media on the Internet. However advertisers and agencies communicate easily their criteria. I met with advertisers and media agencies who proceeded to risky interpolations on the basis of these truncated data. In addition, despite the fact that there are basic criteria, it is difficult and expensive to access data extractions audience crossing these criteria. What could guide the designers of these systems in these choices? Are they knew that making it more difficult to access these directly measures they reduced the turnover of media that fund?