key: cord-0998551-f4k5crvk
authors: Minichiello, Angela; Lawanto, Oenardi; Goodridge, Wade; Iqbal, Assad; Asghar, Muhammad
title: Flipping the digital switch: Affective responses of STEM undergraduates to emergency remote teaching during the COVID-19 pandemic
date: 2022-02-20
journal: Project Leadership and Society
DOI: 10.1016/j.plas.2022.100043
sha: 05bb8613fcb6547f75a80fffcfc0dca703ec2618
doc_id: 998551
cord_uid: f4k5crvk

The Corona Virus Disease-2019 (COVID-19) catalyzed a global shift to distance education known as an emergency transition to remote teaching (ERT). While prior research investigates students' experiences during traditional online learning, fewer studies examine students' affective responses (i.e., feelings, emotions) to those experiences, particularly when remote learning is unexpected and unplanned. To understand how science, technology, engineering, and mathematics (STEM) undergraduates responded affectively to the COVID-19 ERT, researchers generated open-ended survey data with 1340 undergraduates (253 female) in 27 courses across seven U.S. institutions. Using an inductive qualitative approach, researchers developed a three-tier thematic model to synthesize the self-reported reasons underlying participants’ affective responses to the COVID-19 ERT. Findings reveal a complex mix of positive and negative emotional responses among participants that included frequent occurrences of feelings of stress and uncertainty traced to a variety of external, internal, and contextual factors. Implications for STEM teaching practice are discussed.

Although the emergency transition to remote teaching (ERT) event-the unplanned and rapid shift to remote education-that occurred in the wake of the 2020 novel coronavirus pandemic was not a wholly new phenomenon (Czerniewicz et al., 2019; DiCarlo et al., 2007; Gardner et al., 2007; Wright & Wordsworth, 2013) , the scale and speed of this particular transition to remote instruction remains historically unprecedented (Winch et al., 2021) . The near immediate global transition was, as is often said in times of abrupt change, as if 'someone had flipped a switch.' In an instant, students everywhere were forced to respond to rapidly evolving mandates for online education and school and university closures.

These mandates required vast numbers of college-age students to physically relocate from residential campuses and, simultaneously, rearrange existing work and familial responsibilities or search for alternative employment opportunities at new destinations.

Although often conflated with online learning, an ERT event differs from traditional online instruction in several important ways. One prominent difference is the level of advance preparation each affords: while ERT is unexpected and unplanned, traditional online learning is deliberate and "wellplanned" (Hodges et al., 2020) . Understanding that effective online instruction is developed systematically over time, researchers (Means et al., 2014) have identified nine instructional dimensions (i.e., learning modality, instructional pacing, student-instructor ratio, pedagogy, instructor role, student role, communication synchrony, assessment, and feedback) essential to the design of effective online learning experiences. Each dimension contains multiple options; individually, options may be less effective than others, incompatible with others, or fixed or unavailable within a given university or disciplinary context. Judicious prior planning and coordination, therefore, are essential and distinctive hallmarks of traditional online learning that clearly distinguish it from unexpected and rapidly evolving ERT events.

Differences in how instructors prepare to teach remotely point to a second major difference between ERT and traditional online learning: the intention for the permanence of the online instructional design. Hodges and colleagues (2020) defined ERT as "…a temporary shift of instructional delivery to an J o u r n a l P r e -p r o o f alternate [remote] delivery mode due to crisis circumstances." During an ERT event, educators adopt and adapt remote teaching technologies and strategies to deliver courses or activities that a) would be taught using a different modality (i.e., face-to-face, blended, hybrid) if the crisis event did not occur and b) are expected to return to their original modalities after the crisis event passes (Hodges et al., 2020) . Without the intention of supporting remote instructional design in the long term, ERT educators make use of existing support systems and available remote technologies, whether they have used either before or not.

ERT educators may feel forced to adopt completely new or different instructional strategies (e.g., P/F assessment, open book exams, group assignments) that they had not previously considered in non-ERT settings. Thus, ERT stands in stark contrast to more traditional models of online education in which courses and activities are carefully designed and consistently assessed and improved with instructional permanence in mind.

Substantial prior research has investigated students' experiences during online learning; a substantial portion of this literature has focused on student satisfaction and retention within traditional online learning environments. Currently, however, there is evidence (see e.g., Pokhrel & Chhetri, 2021) of a small but growing body of empirical, practice-focused research that seeks to understand student and instructor experiences during the COVID-19 ERT. For example, researchers (Doucet et al., 2020) in U.S. K-12 contexts examined how instructors adjusted and adapted their teaching approaches during the COVID-19 ERT and meshed learning activities with the disciplinary (i.e., subject) and student (i.e., age) contexts of their classes for the purpose of providing recommendations for practice. The authors suggest following an early ERT strategy of 'Maslow before Bloom', which calls for safeguarding the physical, mental, and emotional well-being of students before making formal distance education a priority (Doucet et al., 2020, p. 8) . Across the globe, Sintema (2020) reported on COVID-19 ERT developments in Zambia where educators were expecting a marked drop in Grade 12 students' academic performance on STEM national exams due to reduced opportunities for students interactions with their peers and J o u r n a l P r e -p r o o f instructors due to mandatory school closures. The author reported educators' grave concerns reduced student performance would stall the ongoing rollout of a national STEM curriculum in that country.

In U.S. undergraduate education, Means et al. (2020) 2 reported results of a national, randomsample survey of 1008 U.S. college students. The authors found that the percentage of participants expressing dissatisfaction (i.e., selecting somewhat dissatisfied or very dissatisfied) with their learning increased from 12% to 40% after the switch to ERT. The most frequently cited challenge, which was reported by 42% of participants as a "major problem," was staying motivated to do well in their courses (Means et al., 2020, p. 12) . Other challenges, such as finding a quiet place to do online coursework (20%), fitting online coursework in with home/family responsibilities (17%), not knowing where to get help in the online course (16%), feeling too unwell (physically or emotionally) to participate in the online course (14%), and fitting online coursework in with paid work schedules (8%), were reported major problems by substantially fewer numbers of participants (Means et al., 2020, p. 12) . In all cases except one 3 , higher percentages of Black and Hispanic participants (higher than those of White participants) reported these challenges were major problems (instead of minor problems or non-problems) for them. This finding suggests that challenges associated with the switch to remote learning were not "uniformly distributed" (Means et al., 2020, pp. 12-13) or, rather, were not experienced uniformly by all participants.

Evidence of preferential experience of challenges to remote learning hints at socio-economic and institutional inequities, such as "digital inequality" (Czerniewicz et al., 2019, p. 18) , that is an area that should be directly addressed in future work.

Along with developing better understandings of the challenges that students faced during the COVID-19 ERT, a common theme underpinning the growing pandemic ERT literature is students' need for faculty expressions of care and compassion. Johnson et al. (2020, p. 16 ) examined survey data generated at 627 U.S. institutions during the early weeks of the pandemic and showed how faculty and administrators made deliberate and "progressive" changes (such as "lowering expectations," "ungrading," and "eliminating unnecessary work") to their teaching practices and policies in efforts to reduce students' (and sometimes their own) levels of anxiety and stress. The authors urged readers not to judge or criticize these emergency teaching practices, but rather to view them as necessities implemented to "support, care for, and enable students to succeed" (Johnson et al., 2020, p. 16 Together, these researchers highlighted use of flexible STEM teaching practices and policies that included asking students for feedback about course workload and schedule and altering assessments (e.g., assessing student learning using projects that enable autonomy over fixed exams) (Engineering Education Transformations Institute [EETI], 2020b) and providing leniency, removing time pressure from assessments, making accommodations (e.g. P/F grading), and increasing the remote accessibility of course materials (Gelles et al., 2020) .

Despite thirty years of research in traditional online learning and an expanding base of literature related to the COVID-19 ERT, there are limited empirical studies that focus on students' affective responses to remote learning experiences, particularly when remote learning is unexpected and unplanned. To add to the growing literature related to ERT experiences, this paper reports on qualitative analysis of open response survey data generated with 1340 (253 or 20.1% female; 5 or 0.4% nondisclosed gender; race/ethnicity data not collected) undergraduates enrolled in 27 U.S. STEM-related courses during the spring 2020 semester (when the switch to ERT occurred). Seeking to provide deeper understandings of students' emotional responses to ERT and their personalized reasons for their affective responses, this study was guided by two research questions:

In relation to their self-perceived abilities to succeed in the ERT learning environment, J o u r n a l P r e -p r o o f 1. What affective responses (i.e., emotions and feelings) did undergraduates report experiencing during the rapid transition to remote learning of the COVID-19 pandemic? 2. How did undergraduates describe reasons for their affective responses (i.e., emotions and feelings) during the rapid transition to remote learning of the COVID-19 pandemic?

By providing insights into the affective responses of STEM undergraduates and students' personalized reasons for these affective responses during the COVID-19 ERT, findings from this study will enable STEM educators, administrators, and staff to better prepare and equip students to adapt to, persevere during, and succeed amid future ERT events.

Research in students' affective responses during traditional online learning dates back to the 1990s. Boyd et al. (1998) reported that students experienced feelings of isolation in online learning environments due to the absence of face-to-face contact with other students and teachers. Considering education as a transaction between teachers and learners, Moore (1991 Moore ( , 1993 and Moore and Kearsley (2012) [both cited by Wheeler (2002) ] theorized student feelings of isolation to be influenced by the transactional distance that is inherent to distance education environments. While transactional distance (TD) is conceptualized as a "psychological and communication space, not a physical space, to be crossed, a space of potential misunderstanding between the inputs of the instructor and those of the learner" (Moore & Kearsley, 2012) , some researchers (Lennox et al., 2006; Willens, 2004) have posited that TD may also be influenced by large physical separations and, thus, may be greater for students who are geographically isolated from other actors in an online environment (e.g., rural or displaced students).

As an explanatory framework for how distance education works, TD manifests within online learning environments via dialogue (i.e., two-way communication) and structure. Dialogue represents communication between educators and students; structure represents how an online environment is designed to be flexible and supportive of the unique needs of remote learners (Lennox et al., 2006) . Close transactional distance (i.e., transactional presence) can be attained through accessible and always on twoway communication channels between instructors and students and has been found to be a significant J o u r n a l P r e -p r o o f predictor of student satisfaction and intentions to persist within online learning environments (Shin, 2002 (Shin, , 2003 . Increased transactional distance, as manifested through delayed and unclear responses from teachers, can cause students anguish (Hara & Kling, 2002) and, if responses are consistently delayed and unclear, frustration (Abrahamson, 1998) . Alternatively, closer TD is achieved by reducing structures that are limiting to students (i.e., practices and policies) to make courses less restrictive, more interactive, and more readily adaptable to the needs of remote learners (Horzum, 2015) . In this way, increasing dialogue and/or decreasing restrictive course structures favorably influence (i.e., reduce) TD.

The current study is the conceptually framed using three key facets of TD that manifest through dialogue and/or structure and have been linked to students' affective responses (i.e., emotions and feelings) during traditional online learning: social presence, interactions with technology, and the design of learning activities and supports. In the following sections, these three facets and their empirical connections to student affective responses in distance education are described.

Social presence is one of three key dimensions of the social constructivist Community of Inquiry (CoI) Model of learning in online and blended environments (Garrison, 2017) . Social presence, along with cognitive presence and teacher presence, has been identified as a strong predictor of student satisfaction within online learning experiences (Harasim, 2012) . Social presence is defined as students' ability to share their individual personalities and present themselves as 'real people,' socially and emotionally, into an remote community of learners (Garrison et al., 2000, p. 89) . Social presence is considered vital for increasing active student engagement because it helps students develop a sense of belonging and fosters teamwork and student interactions as a community of learners (Miller et al., 2020) .

Consequently, it is theorized that an absence or lack of social presence may contribute to students' feelings of isolation, disconnectedness, or loneliness and their eventual attrition from online learning environments (Boston et al., Spring 2011; Brindley et al., 2009 ).

Along with examining students' social presence among their peers, research has aimed at developing approaches for establishing and deepening interpersonal and emotional connections between all communicators within online learning environments. Researchers have found that all rewarding interactions, whether with peers or instructors, are apt to positively affect online students' satisfaction, learning outcomes, and social presence (Brinthaupt et al., 2011; Swan, 2001) . Others (Croxton, 2014; Dumas et al., 2013; Horzum, 2015) focused on the regulatory effects that external (e.g., time constrains, inflexible deadlines), internal (e.g., self-efficacy, task-value of the course), and contextual (e.g., feelings of social isolation, family-related issues) factors have on interactions between teachers and students and/or students and peers and how these factors dynamically play out within online environments.

Despite the rapid growth of information technology (IT), most notably the Internet, and society's increasing exposure to and confidence using technology, teachers and students continue to identify a lack of personal fluency using unfamiliar or infrequently used technology as a concern during online learning (Fu, 2013) . A lack of prior experience with technology, as well as the unexpected problems that arise when implementing known technology in new environments, hinders teachers' and students' abilities to navigate remote learning environments and can ultimately lead to frustrating and dissatisfying remote learning experiences.

For example, researchers have found that students with prior computer, software, and internet experience have higher positive perceptions of their online learning experiences than those with less computer experience (Wagner et al., 2002) . Feelings of frustration are often reported by students who experience technical problems with equipment, slow Internet connections, a lack of access to computers or compatible software, and/or a lack of computer skills (Schrum & Hong, 2002) Conversely, technology may also positively influence students' perception of the online classes (C. . For example, video posts and synchronous video conferencing may make students feel more connected. As technology continues to advance, facilitated communication and interaction through technology should be put to greater use to reduce student feelings of isolation. Fundamental changes in the ways society works J o u r n a l P r e -p r o o f and communicates may further work to change students' ways of thinking and knowledge-building and help dissipate feelings of loneliness and isolation in remote learning environments.

For students, self-discipline and intrinsic motivation are known to promote successful and meaningful online learning. For instructors, thoughtful learning plans and use of appropriate e-pedagogies are vital considerations for facilitating successful knowledge-building among students working remotely.

Just as important as the choice of online pedagogy for instructors, however, is the level of quality and consistency of the remote learning materials they develop and provide to students within online learning environments. Researchers (Boyd et al., 1998; Swan, 2001) found that students consider high quality instructional materials essential to their success in online learning environments; students were more satisfied and more positive about their remote learning experiences if remote learning materials used consistent processes, presentation features, and procedures throughout the course.

In this emergent, empirical study of undergraduate learners in STEM courses, we employed a cross-sectional, mixed-method survey research approach. Our goal was to identify in real-time, examine, and describe STEM undergraduates' affective responses, and associated rationales, to the unanticipated transition to remote learning that occurred during the spring 2020 semester.

Shortly after the COVID-19 ERT event began in mid-March 2020, an online survey was developed, face-validated, and further refined to meet the purpose of the study and to improve the readability of the survey items . The resultant survey comprised 13 questions (i.e., 10 multiplechoice/multiple-answer and 3 open-ended text entry) as shown in Table 1 

Course Features to Learning 1) Select online course features that contributed positively to online learning experience 2) Select online course features that contributed negatively to online learning experience 3) Select online course features that had no effect on online learning experience 4. Feelings related to Success in Online Course 1) Select and/or input the feelings you experienced in relation to your capabilities to succeed in the online course (i.e., motivated, uncertain, safe, scared, confident, isolated/alone, anxious, depressed, comfortable, stressed, independent, empowered, supported, other) 2) Input the reasons why you had the feeling(s) selected or provided above 5. Change in Feelings during Online Course 1) Select how your feelings changed during the online course (i.e., grew more positive, grew more negative, did not change) 6. Effective Learning strategies used during Online Course 1) Input what you did to adapt to the new online course 2) Input effective learning strategies you used in the new online course

As part of a larger study, this paper reports on the findings of an inductive, qualitative analysis of participant responses to two survey questions that assessed participants feelings related to their capabilities to succeed in their courses remotely after the COVID-19 ERT event (Table 1 , area of assessment #4).

In this qualitative study, the researchers developed and assigned codes to interpret participants'

open-ended textual responses to survey questions that asked about their emotional responses to the COVID-10 ERT event (Table 1 , area of assessment #4). In doing so, the researchers adopted a socialconstructivist (i.e., interpretivist) theoretical perspective (Glesne, 2016; Koro-Ljungberg & Douglas, J o u r n a l P r e -p r o o f 2008; Lincoln et al., 2011) . Using this perspective, the researchers assumed that human reality is a social construction and that people "…experience the world around them in different ways" (Jawitz & Case, 2009, p. 152) . The constructivist paradigm aligns with the study's purpose to examine how undergraduates recognized and described their unique affective responses (emotions) during the ERT.

The 

Using procedures approved by the research teams' university-sponsored Institutional Review

Board and the web-based survey tool Qualtrics, an online survey was administered to 1340 students enrolled in 27 unique courses at seven institutions of higher education near the end of the spring 2020

semester. Due to the rapid development and evolving nature of the pandemic and consequently the study, convenience sampling (Creswell, 2014) institutions, and one non-doctoral granting institution; and three institutions located in the eastern United

States and four institutions located in the western United States.

Participants were recruited from the following undergraduate courses: engineering (19 courses), mathematics and statistics (3 courses), technical communication for engineers (2 courses), and social sciences (3 courses). All 27 courses were designed and initially taught using a face-to-face approach and then rapidly transitioned-most within a one-week period-to remote learning formats near the mid-point of the spring 2020 semester. Following the transition, all courses were taught remotely using unique varieties of online learning features, such as asynchronous video-lectures, live synchronous remote lectures, virtual labs, and online office hours. All students in each course were invited by their instructors to complete the online survey before the course final exam. Decisions to provide incentives in the form of course extra credit varied among the course instructors and were not regulated by the researchers.

A total of 1340 students responded to the online survey. Prior to conducting data analyses, the entire set of survey responses were evaluated for completeness; incomplete and/or irregular responses were removed. After discarding all incomplete and/or irregular responses, 1237 responses remained and were considered during subsequent analyses. The data used in this work (i.e., responses to survey items shown in Table 2 , #4), consisted of 1237 text-based inputs that addressed the context and the personal, social and/or institutional reasons for the emotional responses reported.

Demographic data for the 1237 participants are presented in Table 2 . Data show that women were represented in the sample at approximately the same level (20%) as women are represented in U.S. engineering programs (20%) (American Society for Engineering Education, 2020). We compare our data to women in engineering since most of the participants were taking engineering courses and, therefore, we assume that they were pursuing engineering degrees. The sample was skewed toward more advanced J o u r n a l P r e -p r o o f (i.e., 71% were juniors or seniors) and higher performing students (89% reported having a CGPA above 3.00). Sixty percent of participants reported having online learning experience. To prepare for qualitative data analysis, the research team consulted the Merriam-Webster online dictionary (merriam-webster.com) as a guide to develop a set of common definitions for the 13 feelings that were provided as potential responses in the survey. Definitions and their wordings were discussed, revised, and agreed upon by the research team and then face-validated by another faculty member, who was not part of the research team, with expertise in professional communications and technical writing.

Researchers also assigned a valence (i.e., positive, or negative) to each emotion as it related to participants' perceptions of their abilities to succeed in the online course. For example, feeling confident was considered an indicator of participants' positive perceptions of their abilities to succeed in the online course, while feeling anxious was considered an indicator of participants' negative perceptions of their abilities to succeed in the online course. The list of common definitions is provided in Table 3 .

J o u r n a l P r e -p r o o f To begin data analysis, we calculated the frequency counts of each emotion (i.e., Table 1 , area of assessment #4, survey item #1) and categorized them as either positive (i.e., comfortable, confident, empowered, independent, motivated, safe, supported) or negative (i.e., anxious, alone/isolated, depressed, 4 Positive or negative emotion with respect to participants' perceptions of their abilities to succeed in an online learning environment 5 Merriam-Webster.com substituting "relief or encouragement" for "comfort" 6 Merriam Webster.com substituting assisted or helped for "supported" 7 Merriam-Webster.com substituting "tension" for "stress" 8 Merriam-Webster.com substituting "feelings of nervousness that makes [one] unable to relax" for "tension" scared, stressed, uncertain) based on valence assignments shown in Table 3 . Next, we conducted a qualitative thematic analysis of the text-based responses that participants provided as reasons for their affective (i.e., emotional) responses. The goals of the thematic analysis were to (1) understand the overarching rationales that participants gave for experiencing each emotion and (2) look across emotions to understand which emotions occurred concurrently or due to similar events or situations.

To prepare the textual data for the thematic analysis, we iteratively analyzed 1273 responses.

Because participants were able to select/input multiple emotions in the survey but were provided only one textbox to input (all of) their reasonings, textual responses often contained information related to several emotions. Therefore, we segmented the textual responses into excerpts (i.e., individual reasons as possible) and then each excerpt was individually and interpretatively linked, or coded, to one or more of the 13 emotions, using Table 3 as a codebook. To retain analytic continuity while accounting for the inherent variation in researchers' interpretations, each excerpt was coded to emotions independently by three of the five members of the research team. Once each excerpt was independently coded by three researchers, the three researchers met virtually via ZOOM to discuss coding choices and resolve differences. Only those excerpts coded with 100% rater agreement (after researchers met to resolve any interpretative differences) were carried forward into the qualitative thematic analysis.

Of the 1273 textual responses, we coded 1193 excerpts (Table 4 ) to the 13 emotions and carried forward to the thematic analysis. To complete the thematic analysis collaboratively, each of the fivemember research team was assigned one to three of the 13 emotions. We made the assignments based on the number of excerpts coded to each emotion. Care was taken to distribute data excerpts equally (as possible) across the research team to mitigate bias by ensuring that any single researcher did not have inordinate interpretive influence on the findings (Table 4) . Excerpt assignment resulted in the following distribution of data for thematic analysis among the research team: Author 1-368 excerpts; Author 2-207 excerpts; Author 3 -277 excerpts; Author 4-145 excerpts;

Author 5-196 excerpts.

Once the excerpts were assigned, the research team conducted a joint qualitative thematic analysis (Saldaña, 2021) . According to (V. , qualitative thematic analysis is appropriate for research questions related to people's lived experiences and the factors and social processes that underpin these experiences. Each researcher began thematic analysis by grouping like excerpts (reasons) within each emotion together into categories and then labeling or "thematizing" each category with a short descriptive phrase (Brinkmann & Kvale, 2015) . During this initial grouping phase, the research team met to discuss how data should be interpreted considering the codebook (Table 3) and research questions. When first pass groupings within all emotions were complete, the researchers met several times in small groups (two or three researchers) to "present" their groupings within individual emotions to the other researcher(s). Conversations related to data interpretation, groupings, and labeling of the categories that occurred within the small groups helped to propagate and integrate individual researcher interpretations among the larger group and provide time and space for researchers to think more deeply about their data and developing categories. Iterative and "cyclical" (Saldaña, 2021) analytical passes with small group presentations continued until all researchers felt satisfied with their categorizations within each emotion.

Next, the research team met as a one large group to review and refine categorizations within each emotion and to work toward the integration of the individual categories into superordinate categories, or themes, that applied across all emotions. This work took several iterations and required several research team meetings. When the research team reached agreement on an initial framework based on the emotion stressed which had the largest number of excerpts (368) and initial categories (7), the first author completed integration of the remaining 12 emotions into the framework. Individually (one by one) and

using an iterative process, the researcher mapped data and categories of the other 12 emotions to the framework. At times the data and categories neatly fit within the framework and at other times changes to the superordinate categories (which then had to be propagated back through the emotions that had already gone through the process) or additions of superordinate categories was required. In the end, 8

superordinate ordinate categories and 23 subcategories were developed that encompassed the data excerpts (reasons) coded to the 13 emotions.

This study is limited in at least four ways. First, all data generated and analyzed in this study were self-reported by the participants. Although data generation was conducted during the COVID-19 ERT event in mid-to-late spring semester 2020, it is possible that participants' affective responses to the ERT could have shifted or changed prior to or during data collection. To some extent, the survey design helped mitigate this limitation by asking participants to explain why they were experiencing the emotions they reported. The act of explaining the reasons for their emotional response may have encouraged participants to think more deeply and carefully about their response selections.

Second, the timespan of online learning examined in this study is considered short in that it lasted substantially less than one full semester. Thus, the affective responses of the participants, especially the responses for those who had no prior online learning experience, may have been unsteady, volatile, or more extreme during this time as participants rapidly adjusted to the online learning environment than they may have been otherwise. In addition, prior online learning experience may have mediated the affective responses of some, but not all, participants in ways that this study did not discern.

Next, due to the emergent nature of the research design as the ERT was unfolding, the demographic data generated for this study was limited in scope; demographic data that was collected did not capture information needed to determine STEM underrepresented status (i.e., race and ethnicity) of the participants and the selection options provided for gender identification were binary in nature (i.e., male, female, prefer not to disclose). Ultimately, the lack of robust intersectional demographic data limited the researchers' ability to examine if and how participants' affective responses differed along intersectional axes. However, research team efforts to sample a variety of courses offered at different types of institutions and at different institutional locations (i.e., R1, R2, teaching-focused, public, landgrant, private, HBCU, and U.S. eastern/western regions) helped to mitigate this limitation and ensure that the data examined in this research represented substantial diversity of student experience (and thus was inclusive of a wide range of emotional responses to those experiences) during the COVID-19 ERT event.

Last, participants were asked to identify their affective responses based on their perceived abilities to succeed in their ERT courses. Participants were provided a list of 13 feelings to select from and an open text box to input additional/other feeling(s) not listed. While all participants selected from the same list of 13 "common" feelings, participants were not provided the definitions of these feelings.

Therefore, the possibility exists that participants selected feelings based on (varied) personal understandings of what those feelings meant, rather than a common understanding across all participants and the researchers.

Findings from this study are presented in the following order: findings related to the affective responses (i.e., emotions and feelings) that participants reported experiencing in relation to their perceived abilities to succeed in the COVID-19 ERT environment (Research Question 1) are presented first;

findings related to the reasons participants reported for having these affective responses (Research Question 2) are presented last.

To identify the affective responses (i.e., emotions and feelings) that the participants reported, we examined the participant selection count of each emotion ( survey item #4) as shown in Table 5 . Considering all selections by all participants (i.e., the responses to Table 1 , assessment area #4, item survey #1), we found that 40% of the total selections corresponded to positive emotions and 60% of the total selections corresponded to negative emotions. Of the positive emotions selected, independent (8.9%), motivated (6.9%), confident (6.7%), and comfortable (6.2%) were selected most often and empowered (2.2%) was selected least often. None of the positive emotions individually accounted for more than 9% of the total emotion selections. Of the negative emotions selected, uncertain (16.5%), stressed (14.9%), anxious (11.2%), and isolated/alone (9.1%) were selected most often and scared (3.9%) was selected least often. Uncertain, stressed, anxious, and isolated each accounted for more than 9% of the total selections. The emotions empowered (2.2%), safe (4.2%), supported (4.9%), depressed (4.3%), and scared (3.9%) each accounted for less than 5% of the total responses.

Considering the number of participants who selected each emotion, the most frequently selected positive emotions (i.e., independent, motivated, confident, comfortable) were each selected by approximately 22-31% of participants. The most frequently selected negative emotions (i.e., uncertain, stressed, anxious and isolated/alone) were each selected by approximately 32-59% of participants.

Remaining emotions (empowered, safe, supported, depressed, scared) were each selected by less than 18% of participants.

To understand how participants explained the reasons underpinning their affective responses, we first characterized the number of the textual excerpts (reasons) (i.e., Table 1 , assessment area #4, survey item #2) coded to each positive and negative emotion and then conducted a qualitative thematic analysis of the excerpts. In this section we describe the characterization of the excerpts first and then discuss the findings from the thematic analysis.

The breakdown (i.e., positive emotion or negative emotion) of the textual excerpts coded to the 13 emotions is provided in Table 6 . Of 1193 total excerpts, 461 (38.6%) excerpts described reasons why participants experienced positive emotions and 732 (61.4%) excerpts provided participants' rationales for having negative emotions. We note this breakdown is approximately equal to rates that participants selected positive (40%) and negative (60%) emotions; this finding provides a degree of confidence that participants were able to express (and researchers were able to interpret) rationales for experiencing positive and negative emotions to the same extent. The breakdown of excerpts within each emotion (Table 6) , however, often differed from the emotion selection count percentage (Table 5) . For example, the emotion stressed received 14.9% of the total selections but was coded to 30.8% of the excerpts. Similarly, the emotion independent received 8.9%

of the total selections but was coded to 4.9% of the excerpts. This finding may suggest that 1) without being provided definitions of the emotions, participants may have been able to describe and provide rationales for some emotions more easily than others, and/or 2) while most participants selected more than one emotion, participants may have felt some emotions more strongly and focused on describing the reasons for having those emotions to a greater extent than the others.

Emotions comfortable (140 (11.7%) excerpts), supported (87 (7.3%) excerpts), and confident (86 excerpts or 7.2%) were the positive emotions with the most coded excerpts and the emotions stressed (368 excerpts or 30.8%), uncertain (169 excerpts or 14.2%), and isolated (110 excerpts or 9.2%) were the negative emotions with the most coded excerpts. The remaining seven emotions (empowered, independent, motivated, safe, anxious, depressed, and scared) were each described by a 5% or less of the total excerpts. In other words, there were approximately four times more excerpts describing why participants felt stressed (368 excerpts or 30.8%) than why they felt confident (86 excerpts or 7.2%).

The coded excerpts were developed into eight superordinate themes, with a combined 23 subcategories, that describe participants' self-reported reasons for the emotions (either positive or negative) they experienced during the COVID-19 ERT. Themes and their associated subcategories were then grouped as external (institutional or social), internal (personal), and contextual factors that influenced, both positively and negatively, the unique affective responses of the participants (Figure 1 ). Table 7 . 

As shown in Table 7 ., external factors played a substantial role in participants' experience of both positive and negative emotions.

Somewhat surprisingly, and in several ways, the multi-faceted changes catalyzed by the ERT helped participants feel comfortable, confident, and safe. Participants who described themselves as adaptive, self-directed learners prior to the ERT expressed that they were comfortable with and confident about the rapid switch to remote learning. For these participants, the pace and unplanned nature of the switch had no real affect because they already possessed well-developed skills for directing their own learning. Others felt comfortable during the rapid transition to remote learning based on their belief that the change in learning environment would result in lowered expectations from instructors for student performance. In addition, the change of physical locations for learning to a place where learning could take place in one's own home helped participants feel safe. As one participant noted, " I can get most things done on my own time and from the comfort of home."

For others, the ERT changes induced negative feelings. The newness of remote learning, its rapid pace and unplanned nature, and the disruption it caused to personal work routines and schedules J o u r n a l P r e -p r o o f combined to produce substantial amounts of stress and uncertainty. Participants wrote how "the learning environment feels chaotic," "my day-to-day life has no structure," and "I lost all sense of routine which is a big part of my personal success." Some participants described difficulties making schedules for remote learning, especially if instructors did not post lectures and materials in accordance with the face-to-face class schedule. Changes to physical learning environments also influenced negative feelings. Some participants struggled with disruptions and poor work environments at home, while others missed the routine of physically attending class at a place distinct from home. As one participant wrote, "I sign up for physical lectures to force some structure into my learning." Last, students felt stressed when their perceptions that course expectations would be lowered were not met. Participants described how, "The expectations for this class we're still associated with the regular in person attendance format, so the workload became overwhelming." Some wrote how "Teachers weren't necessarily making class easier"

and a few commented that they "felt it was unfair" that teachers were purposefully "making things more difficult like taking away points for what can only be called as being petty or if assignments are late they don't make accommodations."

The inherent characteristics and requirements of remote learning, including its general level of difficulty, workload, anywhere and anytime nature, and need for self-direction and time management and organizational skills, carried implications that influenced participants' affective responses positively and negatively. For some, these characteristics combined to enable positive feelings of comfort, empowerment, independence, and motivation. By and large, participants accredited their positive feelings to the flexibility, in terms of place, time, and schedule, that remote learning affords. As one participant wrote, "I know that I am a motivated person and will work at my own pace. I really enjoy being able to set my own schedule and get things done as fast as I want and be able to work ahead."

Participants' perceptions about the difficulty and workload of remote learning and the lack of constraints on the time and place for learning, however, energized negative emotions including stress, anxiety, and uncertainty. Many participants expressed how they felt remote learning encompassed more J o u r n a l P r e -p r o o f work and took longer. One participant noted this writing, "It feels like I have a lot more work than I used to. I don't, at least I don't think I do, but the amount of time I spend on schoolwork feels so much longer when I'm doing it at home." Another participant wrote, "My workload tripled having to learn everything myself with still the same number of assignments." Several participants considered engineering content to difficult to learn remotely, simultaneously comparing remote learning to self-teaching. One participant described feeling stressed by writing, "I felt like it was 100% on me to teach myself all the material" and another wrote that "engineering topics should not be self-taught." Several participants described how the need for constant self-teaching degraded their motivation for learning, which in turn amped up their feelings of anxiety and stress. Participants also struggled with time management and knowing how to organize their work, especially across multiple remote courses. One participant recounted, "Nothing had a time value. I didn't have set dates or times when I would do things and so I would do a lot of them all at once and then forget about the next ones and [had to] scramble to do those before the deadline."

The design and delivery of the ERT remote courses also influenced both positive and negative affective responses among the participants. Interestingly, remote course design and delivery was the single factor in our model that was reported influenced the way participants felt supported. Remote course design and delivery helped participants feel supported and comfortable through the implementation of adequate, well-designed, and open communication channels, which were commended as being especially helpful during the one-week transition period. Providing high-quality course materials (e.g., video lectures, lecture notes) and support resources (virtual office hours, teaching assistant support, and peer help available through working groups) were also common reasons that participants wrote made them feel supported.

Alternatively, remote course design and delivery influenced feelings of stress and uncertainty via "unengaged teaching" and poor instructor communication that led to uncertainty about due dates and course expectations. Alternatively, some participants defined poor communication as overcommunication writing, "There are too many emails to keep track of everything. So, the information is J o u r n a l P r e -p r o o f there, but can be overloaded." A lack of structure and/or inconsistencies in policies and procedures within a single course and across courses in the same degree program was also noted as a cause of stress and uncertainty. Participants expressed how "Every class and professor is [sic] running things differently and it can be hard to keep up with everything." Group projects were often mentioned as being stressors, and some participants noted that group projects "were still pursued despite being much less practical when done remotely." Group projects were considered stressful when there was poor communication between group members, when some group members did not participate, and when there was difficulty accessing resources in closed maker spaces and laboratories that were needed to complete projects. Some participants noted course technology issues as being stress-inducing, such as when communication tools didn't work properly, the internet became unreliable, needed software didn't work on home computers, and instructors struggled to use course technology properly or effectively.

Internal factors comprise personal abilities, actions, and attitudes that are unique to everyone.

Internal factors that were identified as reasons for participants' affective responses include interactions, perceptions of abilities or outcomes, ERT-life integration, and other. An "other" category was added to account for excerpts that related pre-existing mental health conditions as reasons for feeling depressed.

The mapping of the four internal factors to each emotion via the number of excerpts is shown in Table 8 . 

Overall, participants cited internal factors as reasons for experiencing positive and negative emotions.

Lack of human interactions was predominantly reported as a reason for feeling negative emotions such as isolated, stressed, and uncertain. One participant summed up these sentiments by writing, "All I do every day is sit quietly by myself, trying to stay afloat with all my online classes." Other participants described how they felt the combined effects of isolation, stress, and uncertainty writing, "…engineering is hard and the group work atmosphere was mostly gone, even with the use of Webex chats." Others stated that the "…disconnect from the other students adds so much stress and uncertainty about the quality of work I am able to produce." Still others noted feeling uncertainty and stress due to difficulties interacting online and not being able to get their usual or desired amount of reassurance and reinforcement from regular interactions with instructors and teaching assistants.

Participants' perceptions of their abilities and potential outcomes were described as reasons for feeling both positive and negative emotions. On the one hand, participants described feeling confident and comfortable about their abilities to succeed due to their prior online learning experiences or high selfefficacy for learning gained during previous STEM courses. Others described how they felt confident that their strong performance in the course thus far would carry them through the remainder of the semester.

Some participants described how they felt comfortable and confident because they possessed certain attitudes, such as a strong work ethic and a positive mental attitude, that they could count on to help them stay motivated in the ERT environment. Having confidence and being comfortable while learning in a remote environment also led some participants to experience feelings of independence; as one participant wrote, "I enjoy putting the work in on my own time, especially when I am able to work ahead."

On the other hand, self-perceptions about their personal remote learning abilities or potential outcomes led other participants to feel anxiety, stress, and uncertainty about "how doing everything J o u r n a l P r e -p r o o f online would affect understanding and performance." Many described that it was "much harder to grasp the material on my own" or how they were anxious and stressed that they "would forget about something and then have it impact my grade negatively." One participant summed up how his self-perceptions affected his emotions writing, "Online is not my style and my grades reflect that right now. It's very frustrating and stressful which makes me do worse."

Challenges integrating ERT with life were also cited as reasons for having negative feelings such as stressed. Participants described having to find new jobs, revise working schedules, or work overtime as the pandemic wore on. These situations increased participants' stress levels and took time away from studying. Others wrote about "coming home and trying to figure out how to learn" with young children at home, too. Increased personal responsibilities at home, and having to be home while studying, led many participants to reduce the time they spent on schoolwork , further adding to their stress.

Contextual factors comprise influences of an individual's surroundings and social milieu. One contextual factor was identified as pandemic-related concerns. The mapping of the contextual factor to each emotion via the number of excerpts is shown in Table 9 . (Means et al., 2020) . While both studies identified trends of more undergraduates feeling negative toward/less satisfied with their STEM-related courses during the ERT, both studies also identified smaller groups of undergraduates who reported feeling more positive toward/satisfied with their STEM courses during ERT. These findings highlight the complexity, contextuality, and individuality of students' affective responses and is an area for future research.

In our study, uncertainty and stress were the most frequently reported affective responses to the ERT. More than one-half of participants reported experiencing each of these emotions as they transitioned to and participated in remote learning. Additionally, more participants described reasons why they experienced uncertainty and stress than they described reasons for experiencing any other emotion. The ability and/or willingness of participants to describe their experiences of uncertainty and stress in greater frequency and detail may hint at how strongly participants experienced these emotions. These results come at a time when recent reports describe increasing depression, anxiety, and suicidal ideation among college students (Danowitz & Beddoes, 2020; Duffy et al., 2019) . Concurrently, there are increasing calls to shift "engineering stress culture" from "one of suffering to one of thriving" (Jensen, 2021) , and to

integrate new knowledge about non-cognitive and affective factors (e.g., stress, social support, mindfulness) into definitions and measures of STEM student success (Krest et al., 2020) . Our findings add to the conversation regarding the need for STEM educators to take compassionate action to reduce anxiety, stress, and uncertainty among undergraduates, particularly during future ERT events.

Our resultant three-tiered, thematic model provides educators with a conceptual way to consider how external (i.e., institutional and faculty controlled), internal (i.e., student controlled), and contextual factors influence STEM undergraduate emotions during ERT events. Looking across the three factors, we can identify concrete actions that STEM educators can take to reduce stress and uncertainty among students during an ERT :

1) Consider ERT course requirements and expectations and whether they can be adapted. (This includes the need for projects and how projects can be adequately supported remotely). Keep in mind that students may be expecting requirements and expectations to lessen and that assignments may take longer if students are working alone remotely. Clearly communicate expectations to students.

2) Consider that students schedules may be in flux and infuse flexibility into courses to the maximum extent possible. If moving to asynchronous online course formats to maximize flexibility, communicate with students how that an asynchronous format differs from a synchronous style course (they may not know). Consider that students may be keeping their personal study schedules intact, since they are taking multiple courses remotely, and expect course material postings and assignments to adhere to original class schedule.

3) Consider that students may struggle with task management and organization, particularly when there are no synchronous meetings when they are reminded of assignment dates and requirements. Consider sending out frequent but measured messages or announcements, perhaps once per week, to help keep students on track and ensure they are not deluged with messages. 4) Be available online and setup communication mechanisms for students to communicate in small groups with each other. 5) Organization is key. Clear and simple instructional design is likely to keep students engaged.

Consider integrating with other courses in the same program so that exams don't overlap and policies and procedures are consistent as possible across courses. While most participants reported feeling uncertain and stressed during the ERT, reports from smaller groups of participants who felt more comfortable, confident, and independent during the ERT provide hope for the future and makes us ask, 'What was different?' Apart from appreciating the flexibility of remote learning and well implemented course design and delivery, participants' positive feelings during the ERT came from internal factors such as their previous online learning experiences and positive perceptions of themselves as self-directed learners who could self-teach and stay positive amid trying times. These skills and mindsets, however, aren't easily activated through activities that we can simply add to our ERT courses and assume students will absorb.

As 21 st century educators, we are now obliged to build up our knowledge of educational technology, LMS course design, and e-pedagogy to prepare our skills, abilities, and courses for future ERT events, we are also beholden to develop students into resilient, self-regulated learners. As one participant wrote, "It seemed unfair to have the same expectations for us and our learning online when we were not prepared to take classes online and the professors could not necessarily do everything that would help us online." Given the individuality and contextuality of experiences during an ERT, it seems clear that no ERT course can fill the needs of every STEM student; educators won't be able to do everything to help students once an ERT starts. Our data suggests that most of our participants were not prepared for the COVID-19 ERT and that, most likely, it wasn't fair. Our findings make us think that part of doing everything we can, as educators, to reduce students' stress and uncertainty in future ERTs necessitates doing more now in terms of providing opportunities for students to gain online learning experience at low risk and developing our non-ERT courses and activities to train students to be selfdirecting , self-regulating, adaptable learners.

J o u r n a l P r e -p r o o f Student's affective responses to their educational environment play a substantial role in their ability and desire to learn and succeed. Our study, which examined the affective responses of STEM undergraduates across 27 U.S. institutions during the outbreak of the COVID-19 pandemic, indicates that most participants reported experiencing negative emotional reactions, including anxiety, uncertainty, stress, and isolation, when considering their abilities to succeed in the ERT learning environment. At the same time, a smaller group of participants reported feeling comfortable, confident, and independent in the same learning environments. A three-tiered thematic model representing the reasons participants experienced positive and negative emotions was developed. The model showed that while both internal and external factors influenced participants' feelings of stress and uncertainty, primarily internal factors supported participants' feelings of comfort, confidence, and independence as a remote learner. Findings suggest that, along with compassionate course design and delivery that can be used to increase student feelings of support and decrease student stress during ERT events, STEM educators can help shift students' affective responses before the next ERT by providing more online learning opportunities and proactively acting to train STEM students as self-directing and self-regulating learners.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Issues in interactive communication in distance education

Engineering and Engineering Technology by the Numbers

Comprehensive assessment of student retention in online learning environments

Do distance students get value for their HECS dollar?

Creating effective collaborative learning groups in an online environment. International Review of Research in Open and Distributed Learning

Interviews: Learning the craft of qualitative research interviewing

What the best online teachers should do

Comparing asynchronous and synchronous video vs. text based discussions in an online teacher education course

Thematic analysis

Research design: Quantitative, qualitative, and mixed methods approaches

The role of interactivity in student satisfaction and persistence in online learning

Online teaching in response to student protests and campus shutdowns: Academics' perspectives

A snapshot of mental health and wellness of engineering students across the western United States

Maintaining the educational mission of the Louisiana State University School of Medicine in the aftermath of Hurricane Katrina

Thinking about pedagogy in an unfolding pandemic (An independent report on approaches to distance learning during COVID-19 school closures

Trends in mood and anxiety symptoms and suiciderelated outcomes among US undergraduates

Stressors and supports for adult online learners: Comparing First-and Continuing-Generation

Responding to the COVID-19 Crisis: Making a Change through

Responding to the COVID-19 Crisis: Making a Change through

ICT in education: A critical literature review and its implications

Student a airs and Hurricane Katrina: Contextual perspectives from five institutions of higher education in New Orl

E-Learning in the 21st Century: A Community of Inquiry Framework for research and practice

Critical inquiry in a text-based environment: Computer conferencing in higher education. The Internet and Higher Education

Compassionate flexibility and self-discipline: Student adaptation to Emergency Remote Teaching in an Integrated during COVID-19. Education Sciences

Becoming qualitative researchers: An introduction

Students' difficulties in a web-based distance education course: An ethnographic study

Learning theory and online technologies

The Difference between Emergency Remote Teaching and Online Learning

Interaction, structure, social presence, and satisfaction in online learning

Communicating your findings in engineering education: The value of making your theoretical perspective explicit

The time is now to build a culture of wellness in engineering

U.S. Faculty and Administrators' Experiences and Approaches in the Early Weeks of the COVID-19 Pandemic

State of qualitative research in engineering education: Meta-analysis of JEE articles

Examining the importance of non-cognitive and affective (NCA) factors for engineering student success

Getting connected: Learning from external early childhood education student perceptions of their study experiences

Paradigmatic controversies, contradictions, and emerging confluences, revisited

Learning online: What research tells us about whether, when and how

Suddenly online: A national survey of undergraduates during the COVID-19 pandemic (Digital Promise

Online learning: Practices, perceptions, and technology

Editorial: Distance education theory

Theoretical principles of distance education

Distance education: A systematic view of online learning

A literature review on impact of COVID-19 pandemic on teaching and learning. Higher Education for the Future

The coding manual for qualitative researchers

Dimensions and strategies for online success: Voices from experienced educators

Beyond interaction: The relational construct of 'Transactional Presence

Transactional presence as a critical predictor of success in distance learning

Effect of COVID-19 on the performance of grade 12 students: Implications for STEM education

Virtual interaction: Design factors affecting student satisfaction and perceived learning in asynchronous online courses

An evaluation of student satisfaction with distance learning courses. 18th Annual Conference on Distance Teaching and Learning

Student perceptions of learning support in distance education

The issue of equal access to computer-mediated learning in distance education

Operation Warp Speed: Projects responding to the COVID-19 pandemic

Teaching through 10,000 Earthquakes: Constructive Practice for Instructors in a Post-Disaster Environment

The authors would like to thank those Professors who volunteered to recruit participants in their courses: BLIND FOR REVIEW Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.J o u r n a l P r e -p r o o f

• Open-response survey data was generated with 1340 (20% women) undergraduates enrolled in 27 STEM courses across seven U.S. universities during the COVID-19 emergency transition to remote (ERT) learning • More negative (60%) than positive (40%) emotional responses to the ERT were reported• Negative emotions were more often related to external factors including online course design, instructor actions, and the rapid pace of the change • Positive emotions were related to internal factors such as prior online learning experience, personal resiliency, and an ability to self-direct during learning J o u r n a l P r e -p r o o f

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Nothing to declare J o u r n a l P r e -p r o o f