Conference Object

Presentation - Using Computer Mouse Tracking for Stress Measurement? An Online Study

Author(s) / Creator(s)

Freihaut, Paul
Göritz, Anja
Rockstroh, Christoph
Blum, Johannes

Abstract / Description

Computer mouse tracking offers a simple and cost-efficient way to gather continuous behavioral data and has mostly been utilized in psychological science to study cognitive processes. The present study extends the potential applicability of computer mouse tracking and investigates the feasibility of using computer mouse tracking for stress measurement. First empirical results and theoretical considerations suggests that stress affects sensorimotor processes involved in mouse usage. The underlying processes, however, are complex and the empirical evidence sparse. Therefore, our research is exploratory with the goal of finding a meaningful relationship between stress and computer mouse usage. We present data from a between-participant field experiment with N = 994 participants. The online setting allowed capturing mouse usage in a natural setting, while the experimental design fostered internal validity. In the experiment, participants worked on four different prototypical mouse usage tasks (point-and-click task, drag-and-drop task, slider task and target-following task) in a high-stress or low-stress condition. Stress was manipulated with a self-developed stress manipulation task as well as a threatening versus neutral framing of the study’s purpose. In the manipulation check, participants in the two conditions reported small, but consistently different stress levels on multiple self-report measures. The relationship between stress and mouse usage was explored with multiple analysis: (1) We calculated 17 different mouse usage features representing temporal, spatial and task specific mouse usage information in each mouse usage task and compared each feature between the stress conditions using frequentist data analysis. (2) We used the mouse usage features to predict a participant’s experimental condition using a machine learning classification approach. (3) We collapsed the two groups and used the mouse usage features to predict participant’s stress using a machine learning regression approach. (4) We transformed the raw mouse usage data into images of the mouse usage during each task and used the images as the input for machine learning classification of the stress condition and regression on reported stress. None of these analytical approaches revealed a clear and systematic relationship between stress and mouse usage. These findings question the feasibility of using straightforward computer mouse tracking for generalized stress measurement. More generally, our research discloses challenges but also points out promises when working with unobtrusive sensor data to capture feeling states.

Persistent Identifier

Date of first publication

2021-05-18

Is part of

Research Synthesis & Big Data, 2021, online

Publisher

ZPID (Leibniz Institute for Psychology)

Citation

Freihaut, P., Göritz, A., Rockstroh, C., & Blum, J. (2021). Presentation - Using Computer Mouse Tracking for Stress Measurement? An Online Study. ZPID (Leibniz Institute for Psychology). https://doi.org/10.23668/PSYCHARCHIVES.4810
  • Author(s) / Creator(s)
    Freihaut, Paul
  • Author(s) / Creator(s)
    Göritz, Anja
  • Author(s) / Creator(s)
    Rockstroh, Christoph
  • Author(s) / Creator(s)
    Blum, Johannes
  • PsychArchives acquisition timestamp
    2021-05-07T10:15:36Z
  • Made available on
    2021-05-07T10:15:36Z
  • Date of first publication
    2021-05-18
  • Abstract / Description
    Computer mouse tracking offers a simple and cost-efficient way to gather continuous behavioral data and has mostly been utilized in psychological science to study cognitive processes. The present study extends the potential applicability of computer mouse tracking and investigates the feasibility of using computer mouse tracking for stress measurement. First empirical results and theoretical considerations suggests that stress affects sensorimotor processes involved in mouse usage. The underlying processes, however, are complex and the empirical evidence sparse. Therefore, our research is exploratory with the goal of finding a meaningful relationship between stress and computer mouse usage. We present data from a between-participant field experiment with N = 994 participants. The online setting allowed capturing mouse usage in a natural setting, while the experimental design fostered internal validity. In the experiment, participants worked on four different prototypical mouse usage tasks (point-and-click task, drag-and-drop task, slider task and target-following task) in a high-stress or low-stress condition. Stress was manipulated with a self-developed stress manipulation task as well as a threatening versus neutral framing of the study’s purpose. In the manipulation check, participants in the two conditions reported small, but consistently different stress levels on multiple self-report measures. The relationship between stress and mouse usage was explored with multiple analysis: (1) We calculated 17 different mouse usage features representing temporal, spatial and task specific mouse usage information in each mouse usage task and compared each feature between the stress conditions using frequentist data analysis. (2) We used the mouse usage features to predict a participant’s experimental condition using a machine learning classification approach. (3) We collapsed the two groups and used the mouse usage features to predict participant’s stress using a machine learning regression approach. (4) We transformed the raw mouse usage data into images of the mouse usage during each task and used the images as the input for machine learning classification of the stress condition and regression on reported stress. None of these analytical approaches revealed a clear and systematic relationship between stress and mouse usage. These findings question the feasibility of using straightforward computer mouse tracking for generalized stress measurement. More generally, our research discloses challenges but also points out promises when working with unobtrusive sensor data to capture feeling states.
    en
  • Publication status
    unknown
    en
  • Review status
    unknown
    en
  • Citation
    Freihaut, P., Göritz, A., Rockstroh, C., & Blum, J. (2021). Presentation - Using Computer Mouse Tracking for Stress Measurement? An Online Study. ZPID (Leibniz Institute for Psychology). https://doi.org/10.23668/PSYCHARCHIVES.4810
    en
  • Persistent Identifier
    https://hdl.handle.net/20.500.12034/4247
  • Persistent Identifier
    https://doi.org/10.23668/psycharchives.4810
  • Language of content
    eng
  • Publisher
    ZPID (Leibniz Institute for Psychology)
    en
  • Is part of
    Research Synthesis & Big Data, 2021, online
    en
  • Dewey Decimal Classification number(s)
    150
  • Title
    Presentation - Using Computer Mouse Tracking for Stress Measurement? An Online Study
    en
  • DRO type
    conferenceObject
    en
  • Leibniz institute name(s) / abbreviation(s)
    ZPID
  • Visible tag(s)
    ZPID Conferences and Workshops