Notice: the WebSM website has not been updated since the beginning of 2018.

Web Survey Bibliography

Title Visual Heuristics and Answer Formats in Rating Scales
Year 2009
Access date 12.08.2009

In interpreting questions, respondents extract meaning from how the information in a questionnaire is shaped, spaced, and shaded. We carried out five experiments using five questions measured on a five point scale to investigate how the effects of visual heuristics affect the answers to survey questions and test whether effective question writing (using numerical and verbal labels) may overrule visual effects. The experiments were fielded in the CentERpanel. This is an Internet-based panel, although there is no need to have a personal computer with an Internet connection. If necessary, equipment is provided by CentERdata (the agency). The recruitment of new panel members is done through a random sample of landline numbers of candidates. The panel is designed to be representative of the Dutch population. We demonstrate that respondents use the “middle means typical” heuristic only in a polar point scale. When numbers 1 to 5 are added to the polar point format and with fully labeled scales, the effect of spacing between response options disappears. In a second experiment we show that respondents are confused when options are presented in an inconsistent order and do not follow the “left and top means first” heuristic. In a third experiment we find that inter-item correlations are higher in a polar point scale when items are presented on a single screen instead of separate screens. The effect gets smaller when numbers 1 to 5 are added to the polar point scale. In a fully labeled scale no effect is found. In a fourth experiment, we found no evidence that respondents use the “up means good” heuristic: the use of a decremental or incremental scale did not result in different answer distributions. In a fifth experiment, we show that respondents use the “like means close” heuristic: the adding of numbers that differ both in sign and value (-2 to 2) and the adding of different shadings of red and green color affects respondents’ answers. The effect of different numbers was apparent in both a polar point and fully labeled format, while the effect of color was only apparent in the polar point format.

Access/Direct link

Conference Homepage (abstract)

Year of publication2009
Bibliographic typeConferences, workshops, tutorials, presentations
Full text availabilityFurther details

Web survey bibliography - General Online Research Conference (GOR) 2009 (54)

  • 1
  • 2
  • 1
  • 2