Making sense of confusion: Relating performance, confidence, and self-efficacy to expressions of confusion in an introductory physics class

Although confusion is generally perceived to be negative, educators dating as far back as Socrates, who asked students to question assumptions and wrestle with ideas, have challenged this notion. Can confusion be productive? How should instructors interpret student expressions of confusion? During two semesters of introductory physics that involved Just-in-Time Teaching (JiTT) and research-based reading materials, we evaluated performance on reading assignments while simultaneously measuring students’ self-assessment of their confusion over the preclass reading material ($N=137$; $N_{\text{fall}}=106$, $N_{\text{spring}}=88$). We examined the relationship between confusion and correctness, confidence in reasoning, and (in the spring) precourse self-efficacy. We find that student expressions of confusion before coming to class are negatively related to correctness on preclass content-related questions, confidence in reasoning on those questions, and self-efficacy, but weakly positively related to final grade when controlling for these factors ($\beta =0.23$, $p=0.03$).

Figure 1

Before class, students are asked to respond to this three-part confusion question regarding the reading material that they just completed as part of a reading assignment. The second part differs depending on whether students respond yes or no in the first part, as shown. The third part is free response.

Figure 2

Fitted linear regression models explaining variation in confusion index by selected variables of student learning and engagement. Models $2f$, $2s$, and $4s$ are displayed here, where the letters $f$ and $s$ represent fall and spring semesters, respectively. The bars represent the 95% confidence interval around each of the mean values; if the confidence interval does not overlap the “zero line,” then the difference is statistically significant at the $p=0.05$ level.

Figure 3

Difference between each student’s average confusion index and the confusion index predicted by regression model $3s$ (which includes the reading assignment correctness, confidence in reasoning, and self-efficacy during the spring semester), binned according to final grade. The vertical axis displays the difference in confusion index; although the absolute scale of confusion index ranges from 0 (not at all confused) to 2 (very confused), the difference between actual and predicted values can be positive or negative. The horizontal axis displays final grades binned roughly according to letter grade. The error bars represent the standard error of the mean of binned values. The numbers $N$ of students in each bin are as follows: 60 or lower ($N=2$), 60–70 ($N=3$), 70–80 ($N=14$), 80–90 ($N=41$), and 90–100 ($N=28$).