Design factors in mouse-tracking: What makes a difference?

Abstract

Investigating cognitive processes by analyzing mouse movements has become a popular method in many psychological disciplines. When creating mouse-tracking experiments, researchers face many design choices-for example, whether participants indicate responses by clicking a button or just by entering the button area. Hitherto, numerous different settings have been employed, but little is known about how these methodological differences affect mouse-tracking data. We systematically investigated the influences of three central design factors, using a classic mouse-tracking paradigm in which participants classified typical and atypical exemplars. In separate experiments, we manipulated the response indication, mouse sensitivity, and starting procedure. The core finding that mouse movements deviate more toward the nonchosen option for atypical exemplars was replicated in all conditions. However, the size of this effect varied. Specifically, it was larger when participants indicated responses via click and when they were instructed to initialize the movement early. Trajectory shapes also differed between setups. For example, a dynamic start led to mostly curved trajectories, responses via click led to a mix of straight and Bchange-of-mind^trajectories, and responses via touch led to mostly straight trajectories. Moreover, the distribution of curvature indices was classified as bimodal in some setups and as unimodal in others. Because trajectory curvature and shape are frequently used to make inferences about psychological theories, such as differentiating between dynamic and dual-system models, this study shows that the specific design must be carefully considered when drawing theoretical inferences. All methodological designs and analyses were implemented using open-source software and are available from https://osf.io/xdp7a/.

Evidential Strength of Intonational Cues and Rational Adaptation to (Un‐)Reliable Intonation

“…Cursor acceleration was linearized and cursor speed was slowed down (to 1,400 sensitivity) using the CursorSense© application (version 1.32). Slowing down the cursor ensured that motor behavior was recorded as the acoustic signal unfolded, resulting in a smooth trajectory from start to target (Kieslich, Schoemann, Grage, Hepp, & Scherbaum, 2019).…”

Using mouse cursor tracking to investigate online cognition: Preserving methodological ingenuity while moving towards reproducible science

“…Such methodological variation is a consequence of a developing approach without agreed standards and has been important in sampling the range of potential paradigms that might be employed. Nevertheless, recent analyses suggest the methodological specifics of mouse-tracking paradigms (i.e., design features) influence the strength of the relationships observed between experimental manipulations and mouse-tracking outcomes (Grage, Schoemann, Kieslich, & Scherbaum, in press;Kieslich, Schoemann, Grage, Hepp, & Scherbaum, 2019;Scherbaum & Kieslich, 2018;Schoemann, Lüken, Grage, Kieslich, & Scherbaum, 2019). We feel the time has come for the field, to scrutinize and advance its methodological ingenuity and to develop new standards for mouse-tracking research which facilitate strong reproducible findings that can more easily be combined across studies (Morey et al, 2016;Munafò et al, 2017;Nosek et al, 2015).…”

“…We reviewed and synthesized recent findings on design features in mouse cursor tracking (Grage et al, in press;Kieslich, Schoemann, et al, 2019;Scherbaum & Kieslich, 2018;Schoemann, Lüken, et al, 2019) that indicate a considerable influence of design factors on cognitive effects as measured by mouse cursor tracking.…”

Mouse-Tracking

“…However, several data sets, particularly those collected with cameras (e.g., Friedman, Brown, & Finkenbeiner, 2013) or a robotic handle (e.g., Burk et al, 2014) showed relatively homogenous sets of smoothly curved trajectories that might be best described by a single cluster. Other important moderators are characteristics of the study design (for discussions, see Kieslich & Henninger, 2017;Kieslich, Schoemann, Grage, Hepp, & Scherbaum, 2018;Scherbaum & Kieslich, in press;Wulff, Haslbeck, & Schulte-Mecklenbeck, 2018). Figure 5 gives an illustration of the trajectories produced by four hand-picked studies, including the study by Spivey et al (2005; Figure 5a), a mouse-tracking study using a similar methodological setup as Spivey et al (Koop & Johnson, 2013; Figure 5b), a mouse-tracking study using a different methodological setup than Spivey et al (Scherbaum et al, 2010; Figure 5c), and a hand-tracking study using a robotic handle (Burk et al, 2014; Figure 5d).…”