ABSTRACT
Background and Objectives: High self-efficacy may reduce emotional and physiological stress responses in the context of an examination. The present study investigated how these stress responses develop on an exam day, and sequential indirect effects between self-efficacy, threat appraisals, stress responses and performance.
Design and Methods: The sample comprised 92 students (46 women). Self-efficacy, threat appraisals and state anxiety were assessed on a control day one week before an oral exam. Additionally, anxiety was assessed three times on the exam day. Salivary cortisol samples were collected at all time points.
Results: Pre-exam anxiety and cortisol decreased until grades were announced. For both responses, greater levels were related to a steeper decline. However, changes in anxiety and cortisol were unrelated. Self-efficacy was negatively related to threat appraisals and anxiety on the control day. Greater threat appraisals were associated with higher pre-exam anxiety and a steeper anxiety decrease on the exam day, which in turn, was related to better performance.
Conclusions: High levels of self-efficacy may reduce threat appraisals and anxiety in the lead up to an exam, which are related to the intensity and decline of anxiety on the exam day. A steeper decline of anxiety may be beneficial to performance.
Disclosure statement
No potential conflict of interest was reported by the authors.
ORCID
Tobias Ringeisen http://orcid.org/0000-0002-4863-5262
Stephanie Lichtenfeld http://orcid.org/0000-0003-3485-9078
Sandra Becker http://orcid.org/0000-0002-3212-8784
Notes
1 Aside from the control day, we initially planned to measure both stress responses at intervals of 30 min on the exam day: 30 min before, right before, directly after the exam (which lasted about 30 min) but before announcement of the grade, and another 30 min later after announcement of the grade to the examinee. The ethics commission of the university, however, considered an assessment right before the exam as potentially performance-hindering for students and required us to omit the measurement point right before the exam.
2 As bootstrapped confidence intervals are not available with MLR estimation, we used the confidence intervals from an analogous model with ML estimation.