Content area
Full Text
Abstract
This study reports the findings of a comparison between different nonprobability sampling methods applied in a WWW-user-survey. Respondents were recruited in four different ways, namely a one-page press release in a weekly business news magazine, newsgroup postings, hyperlinks from other web sites and an e-mailing.
The results indicate that (1) newsgroup postings and e-mailings generate very high speed response, (2) all non-probability sampling methods possess different caveats which force researchers to apply them cautiously, (3) the recruitment methods generate responses of equal quality and completeness, (4) the various recruitment methods are not suited for generating stable results concerning demographic and webographic dimensions of the WWW-user population and (5) the complementary sampling methods allow useful inferences about web users' attitudes towards the WWW to be made.
Introduction
With a population of over 40 million users worldwide, a number doubling every year, the Internet has become a medium with distinct business opportunities. To academic and commercial researchers, the Internet* is appealing for two reasons. First, the Internet is interesting as a research subject. Accordingly, the Internet and its user base have become the object of study in several well known Internet-user surveys (e.g. the GW and Nielsen studies). Secondly, the Internet can be considered a valuable (electronic) instrument for data collection. Up till now, academics as well as practitioners have paid little attention to the Internet as a new research method. First attempts to fill this knowledge gap have been made in the studies by Mehta & Sivadas (1995), Tse et al (1995) and Comley (1996). These studies compare the (dis)advantages of traditional and new research methods, such as mail surveys, e-mail surveys and web-based surveys. Despite these instructive studies it is fair to say that there remains a lack of specific understanding about the use of the World Wide Web (WWw as a survey research tool. The well known WWW-user-population studies (e.g. Nielsen Media Research 1995; Pitkow & Kehoel 1997) are the first research examples where WWW surveys are continually applied as a research method Within these studies two different sampling procedures are applied for studying the WWW population, namely, random sampling (e.g. random digit dialling is applied for recruiting respondents) and non-probability sampling (e.g. recruiting respondents via press announcements, hyperlinks, newsgroup...