Abstract
While colleges and universities grapple with delivering instruction face-to-face during the pandemic, there is still a lot to learn from remote teaching experiences. The present study aimed to predict self-reported learning during the first year of the pandemic. Building on previous scholarship on the topic, we focus on the moderating effects of self-efficacy, and the mediating effects of coping styles on the relationship between stress and self-reported learning experiences. We also included self-perceptions of class effort, the instructor, and changes in class, personal, professor, and health behaviors. Students (N = 272) in Introductory Psychology classes participated in an online survey as part of a class research requirement. Analyses demonstrated that self-efficacy predicted differences in many measures associated with learning and predicted learning over and above all other variables entered in a hierarchical regression. The relationship between stress and learning was mediated by coping, but not moderated by self-efficacy. These results suggest student beliefs about their ability to perform online are important to learning outcomes, but coping mechanisms mediate the relationship of stress and learning. Especially while teaching during pandemic times using different modalities, instructors will do well to directly address students’ perceptions of their own ability and build self-efficacy.
Introduction
Teaching and learning have changed significantly since the onset of the COVID-19 pandemic (Gurung & Plaza, 2023). The wear and tear of the last few years continues to add up with expected consequences. Faculty report students struggle to engage with material (Fox et al., 2020) and students report difficulty staying motivated (Means et al., 2020). While many higher education institutions returned to in-person instruction in Fall 2021, some colleges and universities continued using remote learning. The emergence of highly infectious strains of COVID-19 (e.g., Delta, Omicron) required many institutions to utilize remote learning for short periods throughout 2022. It is clear that higher education will utilize remote learning again in the future (Daniela & Visvizi, 2022). Our experiences have taught us much that may help shape our future efforts to provide modified modalities in the event of emergency or otherwise. Indeed, a growing body of scholarship on the subject provides significant insight on this matter (Daniela & Visvizi, 2022; Gurung & Plaza, 2023).
There are a number of investigations into key elements of the teaching-learning equation. Studies assessed different approaches to asynchronous learning, student study habits, objective and subjective learning, satisfaction with education, and barriers to learning (Gonzalez-Ramirez et al., 2021; Gurung et al., 2022; Keržič et al., 2021; Kim et al., 2021a; Schreiber, 2021), all of which were impacted by the transition to emergency remote teaching modalities during the COVID-19 lockdowns. In this study, we build on previous work to take a more nuanced look at pandemic learning.
Inspired by calls to take a holistic approach to examining learning (e.g., pedagogical ecology; Daniel & Poole, 2009), utilize a systemic approach (e.g., Learning Sciences; Sawyer & Dunlosky, 2019), and investigate moderators and mediators of learning (Gurung & Hackathorn, 2018), especially in the unique learning context of pandemic learning, we selected a diverse set of behaviors and attitudes suggested to be important in past research. In light of the increased stress experienced during the COVID-19 lockdowns and the proven role of coping in reducing stress (Gruenewald & Wang, 2019; Nuere et al., 2022), we were especially interested in the roles of both stress and coping on student learning outcomes. Thus, we focused both on student stress and their subjective reports of learning, and measured student study behaviors and student perceptions of their courses and their instructors. We tested potential moderators (e.g., self-efficacy) and mediators (e.g., coping) of the stress and perceived learning relationship.
The Role of Self-Efficacy
Self-efficacy, the personal belief that one can accomplish a task, has long been a significant variable in psychological research, particularly concerning learning (Bandura, 2013; Kim et al., 2021b). As suggested by expectancy value theory, believing one can succeed has a significant effect on motivation, in that students will focus more on activities that they expect to succeed in (e.g., Wigfield & Eccles, 2000). Self-efficacy's role in motivation is especially evident in students’ willingness to accept technology enhanced learning (Gurung & Stone, 2020; Rosli & Saleh, 2022). Remote learning was a new modality for most students and faculty, but self-efficacy remains a key variable in predicting learning outcomes.
Recent research indicates the self-efficacious beliefs of students based on educational delivery method (e.g., remote vs. face-to-face), predicts final exam scores. A study conducted with 649 students in April 2020 examined live, synchronous classes (Gurung & Stone, 2020). Researchers measured learning objectively using class exam scores and factored in key contextual features such as class behaviors, attitudes towards the class, and perceptions of the instructor. Students who believed they would do well online achieved higher exam scores and reported greater skill development than students who believed they would only do well in face-to-face classes. This specific self-efficacy, termed Modality-Based Self-Efficacy (MBSE; Gurung & Stone, 2020), also predicted the degree of changes in learning behaviors during the pandemic. The present study extends this work.
A key goal of the present study was to again focus on the effects of self-efficacy while also factoring in key personal factors, including those mental health factors most likely to be exacerbated by conditions of the pandemic. Based on the aforementioned research, we expected the belief of success in online classes to moderate the relationship between pandemic stress and perceived learning in remote courses. Additionally, in order to better ascertain the ways in which personal behaviors around mental health factors would impact perceptions of learning, we also chose to focus on coping.
Stress, Coping, and Learning
The relationship between college students’ coping strategies and learning outcomes has been especially strained during the COVID-19 pandemic (Bathallath & Brahimi, 2022). In numerous studies, college students reported that the abrupt transition to online learning, prolonged time spent in an online learning environment, and technological concerns have led to increased frustration, anxiety, and stress about school (e.g., Browning et al., 2021; Lederer et al., 2021). As effective coping strategies provide students with a buffer against major stressful events (Gori et al., 2020), the ramifications of having poor/underdeveloped coping strategies can result in increased levels of psychological stress (Yang et al., 2020, 2021); arguably this may be the case for individuals who employ maladaptive or negative coping techniques.
Given the natural relevance of coping behaviors and facing stress in the context of pandemic education, it is important to include both these variables in studying learning (Bamber & Schneider, 2022). Experimental designs show acute stress increases cognitive-effort avoidance (Bogdanov et al., 2021), findings mirrored in classroom studies. For example, students reported significant health disruptions, and increases in anxiety and mindless tech use while reporting significant decreases in motivation and the ability to focus (Giuntella et al., 2021; Hicks et al., 2021). In another study, students moving to remote learning decreased their social connections with peers, professors, and the college community (Gonzalez-Ramirez et al., 2021). Healthy habits connected to exercise and eating during the remote portion of the term were similarly negatively impacted. The pandemic clearly influenced how students coped, which influenced how they functioned and felt, suggesting the role of coping as a mediator.
Important Correlates of Learning
In addition to measures of stress, coping, and learning, we selected variables shown to relate to students’ academic performance. In particular, research has established that study behaviors, ratings of the professor, and student and instructor classroom behaviors all predict learning (Dunlosky et al., 2013; Gurung et al., 2022; Richmond et al., 2021). For example, an instructor's behaviors can also play important roles in student learning, especially in influencing student effort (Geier, 2022). In addition to student study techniques and their views of their instructor's behaviors, a holistic view of student learning must also include their health behaviors (Schmid et al., 2021). It is reasonable to suggest that such behaviors would be significantly impacted by the changes imposed on society and the learning environment by the pandemic.
In addition, we also included variables such as personality (e.g., the Big Five) which are often predictive of behaviors in psychological research (Gosling et al., 2003). While the use of some of the variables we included is primarily exploratory and their predictive power during the pandemic is yet to be established, research shows they each can play important roles in learning. In pedagogical research, study designs often have to focus on a small set of variables due to the difficulties of using active classrooms versus a lab setting. This often results in only few potentially confounding variables being measured. We explicitly planned to expand our focus beyond our key research questions to allow for examinations of more relationships. Additionally, having a better picture of the associations between key factors when studying a still novel form of learning modality (remote learning) will enhance future pedagogical planning. Consequently, we included numerous measures such as personality in our design.
The Present Study
This study had four main research questions using key variables from previous studies of learning:
What are the associations between key factors (stress, coping, personal characteristics, study behaviors, and perceptions of the class, professor, and learning) during remote learning? Does self-efficacy moderate the relationship between stress, coping, and learning? Does coping mediate the relationship between stress and learning? Does self-efficacy predict self-reported experiences of learning and class quality, beyond personal factors and class ratings?
We predicted that self-efficacy would moderate the stress-perceived learning relationship and predict a unique portion of variance in perceived learning and related experiences, while coping would be a significant mediator.
Method
Participants
Undergraduate students (N = 308) at a large west coast university started the survey, but 36 (11.7%) did not proceed past the consent page (2% progress). Of the remaining 272 students, 218 (70.8%) were female, 46 (14.9%) male, 1 reported as transgender, and 1 self-identified. All students were enrolled in Introductory Psychology classes and earned credit through their participation in this study. The majority of the students were in their first year (29%), with the rest in their second year (21%), third year (15%), fourth year (13%) and other (4%). The majority of the sample reported a cumulative college grade point average (GPA) in the range of 3.01–4.00 (67%). We did not collect age, ethnicity, or other demographic information to limit indirect identifiers as requested by the IRB. Students were compensated for participation with research credit.
Materials
We measured four major categories of variables: student learning experience, pandemic behavioral changes, class and instructor perceptions, and personal characteristics (stress, coping, self-efficacy, and personality). Descriptive data and internal reliability information for all measures are listed in Table 1.
Descriptive Statistics and Internal Reliability for Major Study Variables.
Note:
Learning Experience
We used four measures of learning experience. We measured student perceptions of learning by having students rate how much they learned in all their classes for the current term. We provided space for five classes and a scale ranging from 1 (none at all) to 5 (a great deal). We also asked students how much they enjoyed all their classes using the same scale. Given classes vary in quality and student enjoyment, we did not expect ratings to be correlated and, consequently, did not compute internal reliability statistics for these measures.
We measured Course Quality with six items (e.g., Canvas material). Students used a 5-point scale ranging from 1 (terrible) to 5 (excellent). We measured class attitudes and perceptions of classes during the term by asking students how they felt toward their classmates, instructor, and learning engagement, as well as their related experiences of happiness and optimism relating to their learning experiences, using a scale ranging from 1 (none at all) to 5 (a great deal).
Pandemic Behavior Changes
We used items modified from previous research to measure change in four major categories. Students indicated the extent to which each of 14 behaviors (e.g., class preparation) changed during the pandemic to create a measure of Class Behavior Change, using a scale ranging from 1 (much lower) to 7 (much higher). Seven items measured how Personal Behavior Change (e.g., television watching) occurred during the pandemic with a scale ranging from 1 (much less) to 5 (much more). Nine items measured how Health Behavior Change (e.g., biking, yoga) occurred during the pandemic with a scale ranging from 1 (much less) to 5 (much more). We measured perceptions of Professor Behavior Change using 10 items from the Teacher Behavior Checklist (Keeley et al., 2006), using a scale ranging from 1 (much lower) to 7 (much higher).
Class and Instructor Perceptions
Items derived from the American Psychological Association's Introductory Psychology Initiative measured student and instructor classroom behaviors (Richmond et al., 2021). We measured student class effort by asking how often they viewed slideshows, took notes, attended class prepared, interacted in small groups, asked classmates a question, and asked instructors a question. Students used a 5-point scale ranging from 1 (never) to 5 (always).
To measure instructor effort, we asked students to think of their favorite class's instructor and rate them on five items: communicates the goals of the course, communicates the purpose of course activities, measures learning in different ways, respects students that have diverse backgrounds, and creates a supportive environment using a 7-point Likert-type scale from 1 (strongly disagree) to 7 (strongly agree).
Personal Factors
Students reported their self-efficacy for online learning by selecting where they learned best: in online classes, in-person classes, or no difference between the modalities. Students fell into two clear groups. The first group (n = 151) who reported they did not learn well in online classes were coded to have low self-efficacy for online learning. Students who reported being able to learn equally well online and in-person were coded as having high self-efficacy for online learning (n = 103). Only 13 students said they did not learn well in in-person classes and were merged into the previous category.
We measured perceptions of stress using the Perceived Stress Scale (PSS; Cohen et al., 1983). This ten-item scale asked students to identify the degree to which they experienced a number of different stressful feelings (e.g., How often have you felt that things were not going your way?). Students responded using a scale ranging from 1 (never) to 5 (often).
We assessed coping styles using the COPE Inventory (Carver et al., 1989). Students indicated the extent they used each of 60 different behaviors (e.g., I talk to someone about how I feel) using a scale ranging from 1 (I usually don’t do this at all) to 4 (I usually do this a lot). We created a positive coping and negative coping composite as suggested by past research (e.g., Kapsou et al., 2010) summing subscales primarily positive (e.g., seeking support) or negative (e.g., substance use) in nature.
Finally, we used a short measure of personality, the Ten-Item Personality Inventory (TIPI, Gosling et al., 2003). This ten-item measure includes two items to assess each of the five major aspects of personality: conscientiousness, agreeableness, neuroticism, openness, and extraversion. Five items are reverse scored and participants used a seven-item scale ranging from 1 (disagree strongly) to 7 (agree strongly). The authors caution against calculating reliability.
Procedure
We posted a link and invitation to a Qualtrics survey on the department participant pool (SONA software). Students had different studies to select from to satisfy a class research requirement. Those who picked the study first read an informed consent form. The entire survey, including the order of presentation and measures not used in this report, is available on https://osf.io/zhwtf/?view_only=e66ebc5968214358adefa627cf8236e0. The study took approximately 20 minutes.
Results
Associations Between Factors Related to Learning
We first explored the associations between variables in the study that have previously been independently associated with perceived learning. Correlations are shown in Table 2. We found evidence supporting most previously established relationships showing an important network of linkages between variables. Students with higher levels of Class Effort reported learning more, r(247) = .20, p = .002, and had more positive Class Attitudes, r(247) = .19, p = .003. We found similar positive relationships between Instructor Effort and all learning experiences including enjoyment of class. For example, the students’ rating of the instructor was significantly correlated with the students’ Class Attitudes, r(254) = .33, p < .001.
Correlations Between Key Study Variables.
*p < 0.05.
**p < 0.01.
A multivariate analysis of variance (MANOVA) tested if students’ self-efficacy related to changes in their learning using the four learning composites. Means and standard deviations for all main variables separated by level of self-efficacy are available on the OSF page. We found significant main effects for self-efficacy on learning, Hotelling's Trace F(4,248) = 20.90 , p < .001, η2 = .25, reflecting a small effect size (Cohen, 1988). Tests of between subject effects showed a significant main effect for self-efficacy on Learning, F(1,252) = 33.74, p < .001, η2 = .12, Enjoyment, F(1,252) = 35.98, p < .001, η2 = .13, Class Quality, F(1,252) = 13.90 , p < .001, η2 = .05, and Class Attitudes, F(1,252) = 75.31 , p < .001, η2 = .23. Students who saw both modalities as equivalent for their learning did better on all counts.
Does Self-Efficacy Moderate the Relationship Between Stress, Coping, and Learning?
We tested moderation using a hierarchical regression model with self-efficacy as a reference variable and moderator for each independent variable of interest. We included an interaction term for self-efficacy*stress, self-efficacy*positive coping, and self-efficacy*negative coping. Self-efficacy did not show a moderating effect on the relationship between stress and coping, and learning. The first step of the model predicted 22% of the variance in learning, F(4,241) = 16.87, p < .001, with self-efficacy (b = .29, SE = 0.08, p < .001, positive coping (b = .21, SE = 0.01, p = .001) and negative coping (b = −0.28, SE = 0.01, p < .001) all significant. Self-efficacy was not significant in the second step, and neither stress nor any of the interaction terms were significant.
It is possible that the single regression model does not capture the nuance of the relationship between these variables. To test if self-efficacy moderates the stress-learning relationship with coping as a significant mediator, we utilized Hayes’ PROCESS Model 5 to examine self-efficacy as a moderator while negative and positive coping were held as mediators. The conditional process overall model showed self-efficacy was not a significant moderator (b = -.55, SE = 0.67, p = .42, CI = −1.877 to 0.777), however positive coping (b = .028, SE = 0.01, p < .001, CI = 0.012–0.044) and negative coping (b = −0.043, SE = 0.01, p < .001, CI = −0.061 to −0.024) both predicted learning. Importantly, the style of coping had a commensurate impact on learning such that when negative strategies were employed, learning outcomes were more negative (and vice versa).
Does Coping Mediate the Relationship Between Stress and Learning?
To focus on the effects of coping exclusively, we conducted a hierarchical regression model utilizing Model 4 PROCESS macro version 3.5 (Hayes, 2018) in SPSS to test if positive and negative coping are mediators of the stress and learning relationship. The overall model showed no significant direct effect between stress and learning (b = -.000, SE = 0.12, p = .99, CI = −0.024 to 0.024), once positive coping (b = .03, SE = 0.01, p < .001, CI = 0.014–0.048) and negative coping (b = −.048, SE = 0.01, p < .001, CI = −0.07 to −0.03) were accounted for; coping predicted learning, thus mediating the relationship between stress and learning, as hypothesized. Additionally, positive coping (b = .39, SE = 0.08, p < .001, CI = 0.24–0.54) and negative coping were predicted by stress (b = 0.49, SE = 0.07, p < .001, CI = 0.35–0.62).
Does Self-Efficacy Predict Unique Variance in Self-Reported Experiences of Learning?
We used a hierarchical multiple regression analysis to test if self-efficacy significantly predicted self-reported learning, and if it did so more than personal characteristics and behaviors. We entered five personality variables, perceived stress, and positive and negative coping in the first block. We then entered student behaviors and ratings of the professor as a block in a second step. Finally, we entered self-efficacy in the third step.
Step 1 predicted a significant portion of the variance, R2 = .16, F (8, 237) = 5.56, p < .001. Positive coping (β = .19, p = .004), negative coping (β = -.23, p = .003), and perceived stress (β = -.19, p < .019) were significant predictors of learning. Step 2 accounted for an additional 6% of the variance, F (2,235) = 9.21, p < .001, with instructor ratings significant (β = .23, p < .001). Step 3 accounted for an additional 4% of variance, F (1,234) = 13.30, p < .001, with self-efficacy a significant predictor of learning (β = .22, p < .001).
Discussion
This study examined changes in student learning behaviors during the first year of the COVID-19 pandemic, with a focus on a potential moderator (self-efficacy) and mediator (coping) of the stress and perceived learning relationship. We found significant positive relationships between self-efficacy, stress, coping and perceived learning, especially in attitudes toward class and ratings of instructors. Similarly to recent work (Bathallath & Brahimi, 2022; Nuere et al., 2022), this study provides further evidence that perceived stress and coping are significantly related to learning experiences, evidenced by our correlational results, and in the role of self-efficacy and coping in perceived learning (Gurung & Stone, 2020). These results provide a strong incentive for more robust examination of potential interventions related to these factors, particularly as the COVID-19 pandemic continues.
As predicted, we found significant variation in our measures as a function of self-efficacy, which also predicted a unique portion of variance in learning experiences, consistent with previous pandemic studies of learning (Gurung & Stone, 2020). We found that self-efficacy predicted perceived learning, but we did not find evidence suggesting that self-efficacy moderates the relationship between stress and coping. This non-significant finding is in contrast to the strong predictive value of self-efficacy predicted by previous work (Bandura, 2013). It is possible that pandemic-related stress was strong enough to minimize the effect of student beliefs about their success in remote modalities with coping mechanisms. In fact, comparing the descriptive data on the PSS against typical scores for individuals in the same age range from before the pandemic, support this possibility. Although psychological research suggests self-efficacy, in many domains, is responsible for the type and valence of student strategies when managing highly stressful situations, it may not be able to compete with coping as a moderator between stress and learning (Bathallath & Brahimi, 2022).
Coping was significantly related to perceived learning in every analysis, suggesting students could benefit from additional opportunities designed to develop positive approaches to coping. Strongly supporting recent research (Nuere et al., 2022), our results suggest useful coping strategies can influence students’ perceptions of their educational experiences. They also suggested that employing negative strategies were associated with negative outcomes, while positive strategies were associated with positive outcomes. This is another reason that we should attempt to provide students with more opportunities to learn positive coping mechanisms, particularly for populations most vulnerable to negative strategies (Siira et al., 2022).
Coping was also related to a number of other student experiences and behaviors. These results imply that coping mechanisms are important for socio-emotional factors related to students’ connection with school and classes, happiness, and overall optimism about their education, while student alignment within modality is crucial to performance-based measures such as learning outcomes. More research is factoring in the role of student behaviors and experiences (Richmond et al., 2021) and it clear this approach provides greater insight into what can aid student learning.
This study replicated and extended previous work to provide an understanding of the multiple changes in attitudes and behaviors in response to learning during the COVID-19 pandemic (Gonzalez-Ramirez et al., 2021; Keržič et al., 2021; Kim et al., 2021a; Schreiber, 2021). Teaching and learning has continued to change throughout the COVID-19 pandemic, and it is likely that higher education will continue to utilize remote methods of learning, both to address safety concerns in possible future global health threats, but also to provide opportunities for students who may be otherwise limited in attending regular face-to-face classes (Daniela & Visvizi, 2022). By taking into consideration the high levels of stress that students self-report regarding living and learning through a global pandemic, we are more certain that factors related to learning outcomes are correlated with coping and other understudied variables in teaching and learning research.
Limitations and Future Directions
There are three major limitations to this study. First, all data are cross-sectional, preventing tests of causality. While PROCESS modeling can be misinterpreted to address causality by virtue of the way models are drawn and temporal precedence, this is not the reality. Both perceived stress and self-efficacy can be strong self-fulfilling prophecies, and this cannot be captured without measurements taken at multiple points during the term.
A second limitation is the lack of objective measures of learning within the study. Relying on self-reported data does not provide a full picture of learning in the COVID-19 pandemic. It might capture more subjective experiences of students’ learning outcomes, which may give greater insight into the effect of stress and self-efficacy on their experiences of learning. Our measure of self-efficacy was only one item and additional work using multi-item scales would provide a more valid measure of the concept. Finally, we did not collect demographic information regarding age, ethnicity, and other personal identity factors, which may have resulted in lack of representation across the sample. The information collected suggests the sample is somewhat lopsided, with the majority of the population being female students in their first or second year of college. These demographic data certainly impact the magnitude of the generalizability of these findings.
Future research should first use more robust measures of self-efficacy. Our measure was a single item question and it is possible that students conflated online learning and remote learning. A multi-item measure will increase the construct validity of the self-efficacy measure. Research also needs to include the collection of demographic data to show differences across race, ethnicity, and other social groups, and the addition of objective measures of learning. It would be prudent to also consider using a less stressful time (i.e., just regular amounts of stress) as the pandemic caused an unusually high amount of stress, which resulted in the PSS potentially overshadowing any effect of self-efficacy.
Students are no strangers to stress throughout college, with or without COVID-19. These factors have a unique relationship across modalities that future research should explore in more detail to give a clearer picture of the mechanisms that drive strategy selection. Follow-up studies can re-examine these factors outside of the context of the pandemic and provide evidence of the importance of modality to students. It is clear that students do not always get what they want with respect to modality even during times of relative normalcy, and the impact of modality in those times should continue to be examined fully. These are just a few more avenues to explore as we continue to mount evidence that success begins with believing one can even try.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
