Results from 6 independent Web experiments comparing visual analogue scales with categorical scales
Frederik Funke & Ulf-Dietrich Reips

Paper presented at the 11th annual General Online Research Tagung (GOR) conference of the German Society for Online Research (D.G.O.F.)
April 6-8, 2009 in Vienna (Austria)

From the researcher's point of view visual analogue scales (VASs) - in the studies realized as plain horizontal lines with verbal anchors on either end - offer a number of advantages over traditionally used categorical scales. In this presentation we bring together the findings from 6 independent Web experiments, conducted between 2004 and 2008. Dependent variables were lurking, dropout, item nonresponse and response times. Additionally, we assessed the seriousness of respondents' participation.
The average length of the VASs we used in our studies ranged between 250 and 400 pixels. As every pixel in length corresponds to a raw value the communication and detection of very fine graduations is possible. Data approximate the desired level of an interval scale (Reips & Funke, 2008). Researchers thus have more opportunities to analyze data than with categorical scales.
But do respondents share the positive attitude towards VASs? Do researchers ask too much of respondents when confronting them with VASs? A number of researches could show that response times with VASs were considerably higher than with categorical scales (e.g. Couper, Tourangeau, Conrad, & Singer, 2006). In our studies we replicated this finding, but found substantially smaller differences. Nevertheless, VASs have a statistically significant effect on response times even though the effect size is very low.
We examined other indicators for respondents' ability to cope with VASs: The share of break-offs, lurkers and the extent of item nonresponse. Regarding the number of data points we report mixed findings. On the one hand we observed few break-offs with VASs. On the other hand, we recognized a higher share of item nonresponse with some studies. Overall, we have more analyzable data with VASs. Our research suggests VASs lend themselves for assessing data in closed-ended self-administered Web questionnaires.