The Impact of Students’ Tolerance of Uncertainty on Their Learning

Abstract

Background. A key aspect of statistical thinking is “accepting uncertainty” (Wasserstein et al., 2019). However, learning to accept uncertainty may not be so easy for everyone. Psychologists describe a trait known as Tolerance of Uncertainty (TU) as uncertainty arising from situations that are unfamiliar or not easily resolved (Gerrity et al., 1990). Evidence suggests that TU is a domain-stable trait. Thus, we might expect that there is little we could do (especially in the short term) to change students’ TU. Furthermore, general TU has been shown to be related to general anxiety, suggesting that TU may also be related to statistics anxiety, a known predictor of students’ learning. Here, we ask if statistical TU may be related to students’ learning as either an antecedent or mediator of statistics anxiety (and its strongly covaried factor, attitudes towards statistics). 

We ask this question as a possible explanation for the lack of measurable impact of interventions aimed directly at targeting statistics anxiety and attitudes towards statistics in hopes of increasing students’ learning. If students’ statistics anxiety and attitudes towards statistics are impacted by domain-stable traits, then an intervention aimed at impacting students’ learning through these affective traits will also need to address their underlying causes.

Methods. We collected data from 744 students across three sections of an introductory statistics class at our university. We measured TU using the Multiple Stimulus Types Ambiguity Tolerance Scale (MSTATS; McLain, 2009), statistics anxiety using items from the Statistics Anxiety Rating Scale (STARS; Cruise et al., 1985), and attitudes towards statistics using items from the Survey of Attitudes Towards Statistics (SATS; Schau et al., 1995). Students’ learning was measured by final exam grades – we report the psychometric qualities of that assessment to justify its use. Relationships between constructs were analyzed using a Structural Equation Model.

Findings. Note, the current phase of research is at initial results – we will have final results at the time of presentation. Preliminary results suggest that TU is correlated with students' learning as an antecedent of statistics anxiety. The correlation between TU and statistics anxiety is .26, and the correlation between statistics anxiety and students’ learning was .47. The total effect of TU on students’ learning was .17.

Implications For Teaching and For Research. That a domain-stable personality trait underlies correlates of students’ learning raises several questions for both research and teaching. Interventions aimed at improving students’ learning by impacting affective traits need to address more fundamental and deep-seated personality characteristics such as TU. Furthermore, an uncertainty-intolerant student may thrive in a traditional introductory statistics curriculum presented as procedural routines. Yet, the same student may struggle in a curriculum focused on a conceptual understanding of statistics. These results should factor into the way we transition towards Wasserstein et al.’s (2019) “world beyond p < .05”.