Using on-line interviewing in access panels for media research has been a reality for the Dutch national readership survey since 2002. The quantity aspect is evident. The quality of such an approach needs to be assured by measures taken to overcome its drawbacks. The innovative and yet traditional survey design taken for NPM is a good example of how quantity and quality can meet media research standards. On-line interviewing for media research is not without danger. Since the internet is itself a medium, the users of the internet are almost certainly biased. Samples drawn from on-line sources can hardly be expected to provide a representative sample of the population.
Still, for years NPM uses on-line access panels for – a part of – its fieldwork. The reason why is obvious: the costs of an on-line interview are far below those of a face-to-face interview, involving travelling and personal visits by an interviewer. On-line panels certainly have advantages, but there are also threats to the quality of the resulting data. A mixed mode approach, in which data can be used complementary, combines the best of both worlds: the cost-effectiveness of on-line panels and the quality of random probability sampling. In 2006, NOM redesigned the Dutch national readership survey NPM to build in guarantees for a more representative sampling and weighting, using a two-step approach (Petric & Appel, 2007). In the first step, the foundation of the survey is built on a random probability sample of homes, which is then used in the second phase as an enumeration survey for the sampling and weighting of the on-line sample. This combined approach is unique in its efforts to enhance overall quality of an economically viable study. Where currently research clients are faced with increasing field costs, this is a practical solution for the longer term.