Suchergebnisse

Abstract
In the world of television-making, the media industry relies on measurements from a people-meter panel in order to find out how many people watch a certain program or commercial on television. A people-meter panel is a special-purpose panel in which the television-watching behavior of household members is measured using a special device attached to television sets. The measuring device can also register the television-watching behavior of guests to the panel homes. The reliability of estimates based on the people-meter panel primarily depends on the sample size. Due to the complexity of the sample design, the effective sample size cannot be determined straightforwardly. This is even more the case when one incorporates the viewing behavior of guests. This paper describes a methodology that determines effective sample sizes by computing design effects and which extends the methodology in such a way that it incorporates 'guest viewing'.

"In this paper the authors study the selectivity in the recruitment of respondents for one of CentERdata's Internet panels (the CentERpanel). This recruitment is based on a probability sample. It involves three stages: participation to a first telephone interview, willing to be re-contacted and final agreement to participate in the Internet panel. By matching data of the recruitment process with registries of Statistics Netherlands they are able to distinguish selectivity with regard to age and income in all stages and with pc-ownership in the latter two stages only. Interestingly, the authors hardly find any selectivity with key variables on living conditions. Finally they will make some explicit recommendations for the recruitment process." (author's abstract)