Web Survey Bibliography

Title A Comparison of Ranking versus Rating in Online Surveys
Author Thomas, R. K.
Source RC33 6th International Conference on Social Science Methodology: Recent Developments and Applications in Social Research Methodology, 2004
Year 2004
Access date 10.09.2004
Abstract Traditionally, rating and ranking measurement operations each have had their proponents. Some criticisms of rating tasks have been that they are susceptible to end-piling (giving high ratings to all targets), often have a high degree of multi-collinearity, and often fail to get people to sufficiently differentiate between factors affecting their decision-making. Some criticisms of ranking tasks have been that they: 1) force dependence in responses which reduces the power of statistics that can be applied to this kind of data, 2) the subjective meanings of a given rank can vary substantially among individuals, making comparisons between individuals, and 3) do not provide distance information between the responses (i.e., is the difference between number 1 rank and number 2 rank equal to the distance between 2 and 3?). Thomas and Shaeffer (2003) and Thomas, Johnson, and Behnke (2004) have reported that, when compared to ranking tasks, rating tasks were less burdensome on respondents and more valid in terms of relating to key criteria in areas as diverse as grocery store attribute evaluations to political issue support. This study sought to extend our understanding of the differences in rating and ranking tasks in yet another area – body image. In addition, we conducted a recontact study to assess the reliability of rankings versus ratings. Method. In our first wave, over 10,000 respondents from the U.K. participated in an online survey. They were asked about their overall perceptions of their appearance and their satisfaction with the appearance of 4 aspects of their body (face, upper body, midsection, lower body). They were then randomly assigned to either rate or rank each of the 4 aspects in terms of physical attractiveness to others. Approximately 30 days after the first wave, we recontacted a subset of respondents to invite them to participate in another survey containing the same items measuring body image, of which 2832 respondents completed the second wave. Results. First, as has been obtained in the other studies comparing rating and ranking methods, we found that the order of means of the elements did not vary substantially whether they were rated or ranked. Second, we again replicated the greater validity of the rating scales (with an average validity twice that of ranking). Third, we found that rating tasks took far less time to complete. Fourth, we found that the average test-retest reliability for the rating scales was .65 while the average test-retest reliability for the ranked values was .43.
Year of publication2004
Bibliographic typeConferences, workshops, tutorials, presentations
Print

Web Survey Bibliography - Thomas, R. K. (64)

Page:
  • 1
  • 2
Page:
  • 1
  • 2