upa - home page JUS - Journal of usability studies
An international peer-reviewed journal

Response Interpolation and Scale Sensitivity: Evidence Against 5-Point Scales

Kraig Finstad

Journal of Usability Studies, Volume 5, Issue 3, May 2010, pp. 104 - 110

Article Contents


Results

The dependent measure in this study was the frequency of respondent interpolations, defined as a response outside the bounds of the values inherent to the Likert items presented to participants. Interpolation was counted as an all-or-nothing event, e.g., responses such as 3.5, 3 1/2, and between 3 and 4 were all counted as equivalent interpolations.

The data were first analyzed as a function of total interpolations, regardless of their source. Some participants demonstrated a predisposition towards interpolating and did so more than once during the course of the survey. Standard questionnaire replies where the participant did not interpolate were scored as discrete responses. These results appear below in Table 1.

Table 1. Interpolations vs. discrete responses

Table 1

A Fisher’s Exact Test run on these data revealed that the 5-point Likert items elicited a significantly higher number of interpolations than the 7-point items (p < .01). These data were also analyzed from another perspective to control for the effects of multiple interpolations by the same participant. Instead of focusing on the number of interpolations that occurred throughout the usability testing, the total number of participants engaged in interpolation was the metric of interest. For this analysis, the participants themselves were coded as either interpolators (committing one or more interpolations) or discrete responders (committing no interpolations). Table 2 illustrates the findings.

Table 2. Interpolators vs. discrete responders

Table 2

Once again, a Fisher’s Exact Test was employed and a significant difference emerged (p < .01), showing that participants were more likely to interpolate in the 5-point condition.

Previous | Next