The number of Internet users is growing fast. In 1995, 2 percent of the Swedish population 16-79 years old were using Internet at least a few times per week. In 2001, the percentage had increased to 51. If those who use Internet more seldom are counted, the percentage 2001 was 70 (Timander and Österman, 2002). High Internet penetration is also reported in many other countries. In, e.g., the United States, about 54 percent of the population were using Internet in September, 2001 (Cooper and Victory, 2002). The European Commission (2001) reports household Internet penetration ranging between 11 and 68 percent in the 15 countries of the European Union. The highest penetrations (over 45 percent) were reported in the Nordic countries, the Netherlands, United Kingdom, Ireland, and Austria. As the number of Internet users has grown, the use of World Wide Web for survey data collection has more and more become an interesting alternative for the survey industry. Among the attractive features are low field work costs and a variety of possibilities to design the questionnaires. Today, “Web surveys” encompasses a great number of methods, with different purposes and populations, see, e.g., Couper (2000).
However, as Web surveys have become increasingly popular, the survey industry has identified a number of severe methodological problems, such as coverage-, nonresponse-, and estimation problems. In particular, this is true for Web surveys of the general population.