We build upon our current efforts to describe and measure Haptic Experience (HX) using scale development.
Haptic feedback is very experiential, contributing to user and player experience. However, existing metrics give little insight into how and why haptics contributes to peoples’ experience, and little direction for hapticians to improve their designs. Our group has been developing a scale to measure haptic experience (HX). So far, we have developed the items (questions) and found evidence for this five-factor model through exploratory factory analysis. However, prior work has only been tested with vibrotactile feedback.
We continued with the process of scale development to assess how the HX model fits different devices and applications. We collected data from five devices: Haply 2diy, Ultraleap Stratos Explore, Oculus Quest 2, TanvasTouch, and 3D Systems Touch. Each participant interacted with one device in a coffeeshop at the University of Waterloo campus for five minutes, then completed the 22-item questionnaire previously developed. We collected two samples, the first (N=291) for exploratory factor analysis, and the second (N=112) for confirmatory factor analysis.
We analyzed the responses using exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). For EFA, we looked at several different models, with 3-5 factors (as indicated by the Scree plot). The most interpretable model that met all recommended criteria only had 4 factors, different from prior results with vibrotactile feedback. We labelled the four factors as Realism, Expressivity, Harmony, and Involvement; the first three are extremely similar or identical to vibrotactile feedback, while involvement is a new factor that needs elaboration.
When we conducted CFA and analyzed scale reliability, all metrics and criteria were met except for sub-scale Cronbach alpha values for Expressivity and Harmony. As such, we believe we have strong evidence for this four factor structure, but the questions are not ready for deployment as a questionnaire. The low reliability scores could be affected by the either low number of items (i.e., only 2-3 items per factor), or require more attention from participants.
Our main takeaway is that there appears to be a structure for HX that is generalized across different modalities, but that it is slightly different from vibrotactile experience. We hypothesize that there might be other related, but slightly different, structures for other modalities. We are also optimistic that, with a little refinement, we could soon have a reliable questionnaire.
Resources
Recorded talk: