Web Survey Bibliography
With the recent rise of Internet based public opinion studies, panel studies have been utilized with much more frequency. Although the benefits of panel methods are well known (e.g., Sharot 1991), using a panel risks bias in two ways. First, since panels rely on re-interviewing panelists, systematic panel attrition can produce a panel that is un-representative of the target population. Second, interviewing and re-interviewing panelists may change the opinions/behaviors of the panelists – creating unrepresentative panelists. To investigate the prevalence and impact of these possible biases, I investigate a panel that is particularly suspect to these sources of bias – the panel of Knowledge Networks. Knowledge Networks’ panelists are not only given an interactive TV appliance and Internet access, but they are also surveyed weekly. In this paper I both examine the extent (and effect) of panel attrition in Knowledge Networks’s panel over a 7 month period, as well as report the results of an experiment designed to isolate the possible opinion/behavior changes introduced by panel participation. I find little evidence of either type of bias in the Knowledge Networks panel.