Paper Summary

Direct link:

Electrodermal Activity and Self-Report Measures: Converging and Independent Evidence of Emotions' Impact on Exam Performance

Sun, April 19, 8:15 to 9:45am, Virtual Room


Theoretical Framework
Prior research within Control Value Theory, primarily utilizing survey methods, has demonstrated relations between self-efficacy, achievement emotions, and performance (Pekrun, 2006). Advances in electrodermal activity (EDA) measurement devices allow researchers to measure students’ objective physiological arousal in authentic educational settings (Harley, Jarell, & Lajoie, 2019). These measures allow exploration of moment-to-moment affective states. These measures do not, however, allow researchers to identify specific discrete emotions students experience or the beliefs that underlie the arousal. We argue that combining self-report and physiological measures can allow emotion researchers to understand better the relationship between affective states and exam performance.
In this study we examine the predictive power of students’ self-reported state emotion (nervousness), control beliefs (self-efficacy), emotional regulation, and physiological arousal (EDA) on performance.

We recruited 69 (52 identified as male, 59 as White/Caucasian) participants from an engineering statics course at a public land-grant research university in the United States. We collected data during a 23-question multiple-choice practice exam. Participants were asked to record their efficacy and nervousness after reading but before answering each question on the practice exam (Figure 1). During the entire duration of the exam period, EDA signals were collected from each participant via an Empatica E4 wrist sensor (Empatica, Boston, MA). The phasic mean EDA was then extracted per duration of each exam question. Performance was measured as correct (1) or incorrect (0) for each of the 23 multiple-choice questions (Table 1). Self-efficacy and nervousness were both assessed using single item ratings on 10-point scales. Emotional regulation (Gross & John, 2003) was assessed, both cognitive reappraisal and expressive suppression strategies.

Analytic Procedures
Multilevel logistic regression models were applied to the data obtained over the course of the study period. A multilevel modeling approach was adopted as each participant provided a rating of their efficacy and nervousness after reading each item stem and had a mean EDA score for each item (i.e., 23 observations on each measure for each of the 53 participants).

Estimation of the unconditional model resulted in a sampling error adjusted performance rate of 74.0% (95% CI = 56.3%, 97.4%) and evidence of statistical variation across participants (τ00 = 0.82), 2 (52) = 163.561, p < .001. Table 2 presents the conditional model results. At level-1, efficacy was positive and EDA was negatively related to item performance. At level 2, mean exam performance was positively related to mean efficacy, negatively related to mean EDA, emotional reappraisal, and mean nervousness. Together, the predictor variables accounted for 42% of the variation in exam performance.

EDA and self-report are aligned with regard to item and exam performance, EDA and nervousness have independent between-person effects. This suggests that physiological and self-report measures cannot only provide converging evidence, but also provide complementary findings that more fully represent the students’ affective states.