Online learning involves students attending classes remotely either synchronously via videoconferencing platforms or asynchronously via online learning management systems. Regardless of the specific online learning environment, students need to have reasonably high expectations of their ability to complete academic work online, a sense of current and future value in undertaking the work, and not view it as insurmountable or too costly in effort required. These factors align with the subconstructs of the expectancy-value theory, which posits that students will be more motivated to pursue goals if they perceive that the value outweighs the cost (Beymer et al., 2022; Wigfield et al., 2021).

Under the expectancy-value theory, expectancy is a person’s belief in their ability to succeed in future tasks in a given domain (Eccles and Wigfield, 2020; Wigfield and Eccles, 2000). Value is a measure of their perception of the importance, enjoyment, and utility gained from engaging in specific tasks (Kosovich et al., 2015). Value and expectancy are positively related, domain-specific variables with multiplicative effects (Kosovich et al., 2015; Lauermann et al., 2017). However, their connection may diverge across education levels, e.g., weakening as academic difficulty increases, but strengthening in areas of preferred focus as students progress in their coursework (Loh, 2019). Cost, defined as one’s perception of the price of engaging in a task (Beymer et al., 2022), is negatively associated with expectancy and value (Perez et al., 2019), and student outcomes (Flake et al., 2015). Future interest is the perceived long-term utility of engaging in and succeeding in a given task, which correlates positively with expectancy and value, but negatively with cost (Goldman et al., 2022). Undergraduates who do not perceive much value in taking a course tend to exert less effort than those placing higher value on the course (Cole et al., 2008). First-generation undergraduates tend to perceive higher cost than their non-first-generation counterparts and also show lower academic achievement, yet these two factors have not been causally linked (Goldman et al., 2022).

Other research has emphasized the importance of students’ sense of self-efficacy in predicting their success in online coursework (Alqurashi, 2016). Academic self-efficacy is defined as beliefs in one’s ability to organize, implement, and accomplish specific academic tasks (Bandura, 1994; Doménech-Betoret et al., 2017). As students achieve academic success, their perceptions of their abilities tend to increase (Schunk and DiBenedetto, 2020), which may lead to greater persistence and higher course grades (Wright et al., 2013). At the college level, academic self-efficacy has been linked with academic resilience (Cassidy, 2015; Stephen et al., 2020), especially in academic contexts that demand self-regulated study such as massive open online courses (Alamri, 2022). Expectancy-value beliefs and academic self-efficacy are related. Both refer to students’ belief in their own ability; however, their conceptual definitions are different. Expectancy-value beliefs involve one’s expectation of success in an upcoming task, while academic self-efficacy beliefs refers to one’s ability to perform that task (Wigfield et al., 2021; Gao et al., 2008).

In the context of online coursework, it is also imperative for students to minimize distractions to remain focused and engaged in their studies (Greenhow et al., 2022; Salhab and Daher, 2023). Hence, the present study also examined students’ self-reported ability to pay attention and maintain focus while doing academic work in relation to course outcomes. Attentional control plays a critical role in students’ ability to self-regulate while studying, and it is positively associated with academic outcomes (Rueda et al., 2010). With regard to learning conditions, attentional control is thought to play a greater role in individualized, remote learning environments as compared to traditional, in-person instructional settings with enhanced social support for learning (Biwer et al., 2021; Nesher-Shoshan and Wehrt, 2022). To date, there is a paucity of research linking attentional control with expectancy-value beliefs and academic self-efficacy. However, university students report being less motivated and less able to regulate their attention when learning online compared to in person (Biwer et al., 2021), suggesting a positive association. Researchers using a proxy for attentional control (P3b, derived from electroencephalography) reported a significant positive association between students’ self-efficacy and P3b amplitude, suggesting that students with higher self-efficacy showed higher levels of attentional control (Themanson and Rosen, 2015).

In addition to examining students’ self-reported expectancy-value beliefs, academic self-efficacy, and self-reported attentional control in relation to course outcomes, we considered students’ reading comprehension as a critical skill for online coursework. Reading is fundamental at all levels of education (Elleman and Oslund, 2019; Perin, 2013); however, according to the Nation’s Report Card, it remains a grave problem. Since 1992, the average U. S. student has scored below proficiency in reading from elementary to secondary levels of education (National Assessment of Educational Progress, 2022). Alarmingly, fewer than half of U.S. adults between the ages of 16 and 74 demonstrate reading abilities above the 6th grade level (Rothwell, 2020). Though 62% of high school graduates enroll immediately in college (National Center for Education Statistics, 2023), 63% of them score below proficiency-levels in reading (National Assessment of Educational Progress, 2022). This is troubling because strong reading comprehension is necessary for college coursework and predictive of grades (Clinton-Lisell et al., 2022; Talwar et al., 2023). Undergraduates with a history of reading difficulties tend to have lower self-efficacy than their counterparts and earn lower grades (Bergey et al., 2018). More generally, less-skilled students tend to be outperformed by their peers, making it difficult for them to close the skills gap and elevating their risk of dropping out (Kalbfleisch et al., 2021; Pinkerton, 2010).

The present study used the composite persistence model (Rovai, 2003) as a framework for identifying factors associated with learning outcomes in an online Introductory Psychology course. Course attrition and dropout rates tend to be highest among first-year students taking general education courses like Introductory Psychology, making this course suitable for examining academic persistence (Windham et al., 2014). Further, at the time of this study (Fall 2021), students were still experiencing the repercussions of the COVID-19 pandemic and its disruptive effect on education, specifically for high-school students making the transition to college. We used structural equation modeling (SEM) to explore factors associated with course outcomes in fully online course sections. Aligning with the composite persistence model, we focused on self-reported expectancy-value beliefs about the course, academic self-efficacy, and attentional control and assessed students’ reading comprehension as predictors of their success in an online course. These factors have been linked to course outcomes in prior research (e.g., Gravelle et al., 2023, 2024; Gurung and Stone, 2023) but have yet to be considered jointly in a structural model.

Course outcomes assessment data were collected from 16 sections of an online Introductory Psychology course taught at a non-selective, minority-serving public institution in the northeastern United States in Fall 2021 (Institutional Review Board classification: exempt). Following best practices for replicability, the following materials are publicly available in an Open Science Framework repository (Roberts et al., 2025): course syllabus, the Qualtrics online assignments (PDF and QSF file for implementation), a de-identified datafile, R analysis script, and Supplementary tables. For an updated course curriculum with deployable Qualtrics assignments, see Brooks et al. (2025) and Zapparrata et al. (2025).

Twelve course sections had regular enrollments (M = 41.3 students, SD = 9.2) and four had large enrollments (M = 111.3 students, SD = 9.7). All sections followed a uniform syllabus for a 15-week semester, with links to course materials posted to a learning management system. All sections held synchronous meetings once or twice weekly on Zoom. During the Zoom meetings, students were advised to turn cameras off to preserve Internet bandwidth, as per institutional policy. In contrast, instructors had cameras on at all times.

Asynchronous course features included a free online textbook (Diener Education Foundation, 2020), weekly multiple-choice quizzes and periodic multiple-choice tests on textbook modules, bi-weekly online assignments emphasizing psychology as a data science, a role-play activity and discussion board on research ethics, and rubric, template, and instructions for recorded student presentations on psychological disorders. The bi-weekly online assignments were developed using Qualtrics survey software to align with the textbook modules. The Qualtrics assignments included TED talks by prominent psychological scientists, instruction on using the college library website and Google Scholar to find empirical research articles, practice in reading scientific abstracts, and exercises in using Excel for data analysis and manipulation. Each assignment was designed to take approximately 1 hour to complete and included various open-response and multiple choice questions; the assignments are publically available in our Open Science Framework repository (Roberts et al., 2025).

Of the 1,090 enrolled students, 142 (13.0%) did not complete the first Qualtrics assignment containing the measures reported in this paper. An additional 8 students (0.7%) did not complete any of the course outcome measures apart from the first Qualtrics assignment; these students were dropped from the analytic sample. We have no further information about these students. Otherwise, students who completed the first Qualtrics assignment and at least some of the other course outcome measures were included in the sample. The final analytic sample comprised 940 students (M age = 19.3 years, SD = 3.9, range = 16–51). Of these students, 93 students (9.9%) withdrew at some point during the semester. Students self-reported their race/ethnicity using non-mutually exclusive categories; see Table 1. Most students (78%) were in their first semester of college. About half (46.4%) were among the first generation of their families to attend college.

Students completed measures of expectancy-value beliefs, academic self-efficacy, self-reported attentional control, and reading comprehension in the first Qualtrics assignment. As a data quality check that students were reading the items, we examined variability in responses for scales with reversed scored items (i.e., expectancy-value beliefs, academic self-efficacy). Students who failed to vary responses across all items had scores imputed using the mice package in R (van Buuren and Groothuis-Oudshoorn, 2011).

As an initial step in model-building, we constructed a measurement model to examine how the three latent variables (expectancy-value beliefs, academic self-efficacy, and course outcomes) fit together in a structural model. All three latent variables had excellent model fit indices, establishing their validity. Ratings taken at the start of the semester indicated that students generally held high expectations of success in the course, high perceptions of immediate and future interest in the course and believed that costs associated with the course were reasonably low. They also possessed generally high levels of academic self-efficacy, reporting moderate-to-strong beliefs in their reading, research, technology, and information literacy abilities. Expectancy-value beliefs and academic self-efficacy were positively correlated, which is appropriate given the connection between beliefs of success and future task performance (Wigfield and Eccles, 2000; Wigfield et al., 2021). Surprisingly, however, only expectancy-value beliefs were positively related to course outcomes, and the magnitude of the relation was small.

Regarding associations between intrapersonal beliefs and student skills, both expectancy-value beliefs and academic self-efficacy showed positive relations to self-reported attentional control, while only expectancy-value beliefs showed a positive relation to reading comprehension. These results align with prior research linking academic motivation in pre-college students with staying on task (Wu, 2019) and reading performance (Geng et al., 2023). The association between academic self-efficacy and perceived attentional control aligns with research identifying associations between self-efficacy and cognitive measures of attentional control in experimental tasks (e.g., flanker task; Themanson and Rosen, 2015). The finding that perceived attentional control exhibited stronger correlations with both belief variables than with reading comprehension might be attributed to the self-report nature of the former measures and students’ generally optimistic views of their abilities at the start of the semester. While a cognitive measure of attentional control, such as the flanker task, may have produced different results, we considered a self-reported measure of students’ perceived ability to pay attention while studying online to be more relevant in the context of online coursework.

Regarding the effects of intrapersonal beliefs and student skills on course outcomes, only student skills predicted course outcomes. Students’ outcomes were twice as strongly associated with the performance-based reading comprehension test than with their self-reported attentional control. This difference might reflect students overestimating their ability to maintain attention while studying online over a 15-week semester. The finding that a performance-based measure was more predictive of course outcomes than any of the attitudinal measures is in keeping with other research findings. For example, Hood et al. (2012) reported that prior knowledge of course-related material (i.e., past performance) predicted over 20% of the variance in course grades in a psychology statistics course, whereas expectancy-value beliefs explained only 2% of additional variance. Similarly, Zapparrata et al. (2025) found that prior knowledge of statistical concepts predicted course outcomes in an Introductory Psychology course to a greater extent than academic self-efficacy, though the self-efficacy measure was statistically significant in some of their analyses (e.g., quiz scores). Here, the knowledge and attitudinal measures collectively accounted for 5.4–11.8% of the variance in course outcomes. Taken together, these and other findings (e.g., Zakariya, 2021) suggest strong continuities between past and future academic success, with attitudinal measures likely reflecting students’ educational experiences rather than yielding a direct effect on future performance.

In the present study, the SEM explained a mere 6.6% of the variation in course outcomes. When we examined previous estimates from Gravelle et al. (2024), which included measures of academic self-efficacy, self-reported attentional control, and reading comprehension, we similarly found that these three factors alone explained just 2.1% of the variance in quiz scores, 4.4% of the variance in test scores, and 2.2% of the variance in Qualtric assignments completed; see Supplementary Table 11. Note that the Gravelle et al. (2024) dataset came from the same institution as the present study and examined Introductory Psychology course outcomes in Fall 2020 when classes were fully online due to the COVID-19 pandemic. Taken together with the present SEM analysis, the results suggest that individual factors, even when well-established, may account for relatively little variance in online course outcomes. Here it is important to note that the students may have been experiencing social isolation, health and mental issues, and other negative consequences of the pandemic (Filho et al., 2021). Though they seemed to be largely motivated, expressed confidence in their academic abilities, and seemed adequately skilled at the start of the semester, those factors seemed to play a limited role in determining their course performance. Student motivation tends to be higher at the beginning of the semester and declines as the semester progresses (Darby et al., 2013). In Fall 2021, this trend may have been more pronounced as the after-effects of the COVID-19 pandemic were still resonant, and the steady diet of classes via Zoom may have exacted an earlier toll on many students, with attention and zeal quickly giving way to boredom and fatigue.

The present study is representative of educational research examining effects of individual-level attributes on academic performance: Though many intrapersonal, non-cognitive factors may predict student outcomes, such as grades and GPA, previous models indicate that they collectively account for less than 20% of the variance in performance (Abdulwahid et al., 2022; Forjan, 2017; Gravelle et al., 2023). Looking forward, we must consider limitations to this study and how we can mitigate them in future research. Future work should consider how self-report measures of motivation, self-efficacy, and attentional control might change as the semester progresses. That is, administering measures more frequently instead of once at the start of the semester might catch students closer in time to when they are at risk of giving up. Unfortunately, however, as students withdraw from participating, it becomes increasingly difficult to obtain measures of their course-related attitudes. One potential approach is to adopt analytic approaches that allow researchers to estimate when students will stop participating or withdraw from the course (e.g., survival curve analyses; Ameri et al., 2016) to better understand factors influencing attrition.

Notably, in creating the SEMs for the present study, we had to drop 13% of the enrolled students because they failed to complete the first homework assignment containing the reported measures and an additional 1% who did not submit any quizzes, tests, or other assignments. Consequently, we cannot rule out the possibility that students’ intrapersonal beliefs might play a greater role in determining their initial engagement in an online course as opposed to their sustained engagement throughout the semester. Relatedly, our students tended to have relatively high expectancy-value beliefs and academic self-efficacy at the start of the semester, and they tended to do quite well in the course overall. Significant effects of self-efficacy and expectancy-value beliefs on learning outcomes have been observed in mathematics and statistics courses (Cherney and Cooney, 2005; Hoegler and Nelson, 2018; Hood et al., 2012; Huang and Mayer, 2019). Thus, had we examined the performance of students in a more difficult science or math course with a wider grade distribution, we may have found intrapersonal beliefs to play a larger role in explaining online course outcomes. Additionally, research involving high school students has reported larger effects of expectancy-value beliefs on academic achievement than we observed (Doménech-Betoret et al., 2017), suggesting that effects may be attenuated in higher education settings where students self-select to enroll in specific courses.

Another concern is that we examined attentional control using a self-report measure, as opposed to using a behavioral measure. While students’ perceptions of their ability to focus on school work predicted online learning outcomes in previous study (Gravelle et al., 2024), future research should attempt to measure attentional control in a more objective way. Similarly, by focusing specifically on expectancy-value and self-efficacy as intrapersonal constructs, we failed to consider whether other beliefs (e.g., growth mindset or mindfulness) might influence engagement and success in online coursework. Hence, we should be cautious about overstating our findings regarding the minimal effects of students’ intrapersonal beliefs on online learning outcomes.

Emerging scholarship makes a more strenuous, compelling argument for psychologists to direct more of their efforts beyond the level of individual-level factors and focus more on systemic factors associated with educational outcomes and other social issues (Chater and Loewenstein, 2023; Thomas, 2021; Zengilowski et al., 2023). As evidence of the imbalance, researchers cite the preponderance of studies focusing on individual-level factors and the relatively small number exploring effects of social and systemic factors in relation to student success. This opens up an interesting line of inquiry, possibly more appropriately studied qualitatively at some future juncture. Along these lines, future work should consider how social-cultural and contextual factors (e.g., discrimination and social injustice, food and housing insecurity, and family and work obligations) influence students’ sense of agency and efficacy at the undergraduate careers and, ultimately start of their undergraduate careers and, ultimately, their academic success (Osher et al., 2018; Stetsenko, 2017).