Web-based data collection yielded an additional response bias--but had no direct effect on outcome scales.

OBJECTIVE

To assess and to evaluate possible effects arising from Web-based data collection on the results of a study.

STUDY DESIGN AND SETTING

We analyzed participants of the German Weight Control Registry (GWCR) of whom 328 chose to use Web-based questionnaires and 139 preferred to participate via a traditional postal survey. Furthermore, we included data of 212 individuals sampled independently from the general population who fulfilled the study's inclusion criteria-giving us the chance to differentiate between response bias (concerning Web-based data collection) and general selection bias (concerning participation in the GWCR).

RESULTS

In addition to selection bias (GWCR participants are overall better educated, more likely to live in a partnership, more often female, and older than the general population), we also found a substantial response bias: Participants using the Internet were younger, better educated, and more often male compared with participants preferring the paper-and-pencil version. However, after adjusting for these differences, we found no additional direct effect of Web-based data collection on any of the outcome variables.

CONCLUSION

Web-based epidemiologic studies still do not attract the same participants as postal surveys, even in highly industrialized countries. However, after adjusting for this bias, the same results can be expected.