Abstract
Background
Few studies have examined the effects of later class start times in college, even though developmentally driven sleep phase changes persist into the mid-20s.
Objective
We hypothesized that sleepiness would be higher in an 8 a.m. versus 10 a.m. section of Introductory Psychology, and grades, engagement, enjoyment, attendance, assignment completion, ease of waking and staying awake, and student evaluations of teaching (SETs) would be lower.
Method
Eighty-two students enrolled in an 8 a.m. (n = 39) or 10 a.m. (n = 43) section of Introductory Psychology reported their GPA and completed the Stanford Sleepiness Scale, Likert-scale items from the College’s SET form, and other questions.
Results
Students in the 8 a.m. section reported lower GPA, class grade, and assignments completed; more sleepiness; and a harder time waking up and staying awake for class. Several findings held when controlling for GPA.
Conclusion
Results extend previous research and suggest the importance of offering later classes in college. Limitations include a small sample size, quasi-experimental design, and use of self-report data. Future researchers should test whether findings replicate with other kinds of classes and in other subjects.
Teaching Implications
To maximize student success and well-being, later courses should be offered at the college level whenever feasible.
Beginning with the onset of puberty, adolescents experience a reliable and consistent change in sleep/wake patterns that shifts them towards later bedtimes and later wake times (Roenneberg et al., 2004). These circadian rhythm changes are hormonally based and often mean that teens’ biological needs clash with socially- and academically imposed sleep-wake times (Kelley et al., 2017; Wahlstrom et al., 2014). This clash is a major factor leading most adolescents to get less than the medically recommended 8–10 hours of sleep per night (Hirshkowitz et al., 2015). Teen sleep deprivation contributes to many adverse outcomes, including symptoms of anxiety and depression (Shochat et al., 2014), risky behaviors (Shochat et al., 2014), poor grades (Saxvig et al., 2012), and school absenteeism and tardiness (Hysing et al., 2014). Later school start times may help to alleviate many of these problems. Compared to students who attend schools with earlier start times, students who attend schools with start times between 8:30 a.m. and 9:00 a.m. get more sleep (Neuroth et al., 2021) and show higher motivation, reduced sleepiness and better mood (Owens et al., 2010), greater focus (Minges & Rede, 2016), decreased tardiness and disciplinary issues (Thacher & Onyper, 2016), and other positive outcomes. As such, most major professional organizations recommend a start time of 8:30 a.m. or later for middle and high schools (e.g., American Academy of Pediatrics Adolescent Sleep Working Group, 2014).
Nearly all the research on school start times and the adolescent sleep phase shift is at the middle and high school levels. However, developmentally driven sleep changes persist into one’s mid-20s (Roenneberg et al., 2004). The prevalence of delayed sleep phase syndrome in college students may be as high as 24.90% (Samaranayake et al., 2014), a rate 4 times that of the general population. Overall, only 11.00% of college students report good sleep quality, with the majority endorsing fatigue in the morning (Buboltz et al., 2001). College students with regular sleep deprivation and daytime sleepiness have lower grade point averages (GPAs), impaired mood and learning, and risk of academic failure (Hershner & Chervin, 2014). Both self-reported sleep problems (Hartmann & Prichard, 2018) and poor sleep, as measured by wearable activity trackers (Okano et al., 2019) predict worse grades. These and other studies suggest that early college classes might have similarly negative effects on undergraduates’ alertness, focus, attendance, and performance as we see in younger adolescents. Indeed, research suggests that college students would prefer that university activities start about 2 hours later than is typical (Norbury & Evans, 2019).
Only two published studies of which we are aware have directly examined the impacts of later class times at the university level. Diette and Raghav (2017) studied class grades for college students randomly assigned to morning versus afternoon sections of the same courses over a 9-year period at a highly selective American liberal arts college. Across 115,610 observations, students assigned to morning sections earned lower course grades than students assigned to the same classes later in the day. This was especially true for 8 a.m. and 9 a.m. sections and for male students. Similarly, Carrell and colleagues (2011) examined academic achievement in 6165 first-year students at the United States Air Force Academy, who were randomly assigned to courses starting before 8 a.m. or later in the day and found that students taking the earliest classes performed worse in all courses taken on that day, not just in the early course. Taken together, these studies suggest that early-morning college classes can have a significant negative impact on academic achievement.
On the other hand, at least one study has suggested that later college class times may be associated with lower student grades and certain other negative outcomes. In a sample of 255 undergraduates at a northeastern American liberal arts college, Onyper and colleagues (2012) found that students choosing later class times got more sleep, reported less daytime sleepiness, and had lower absenteeism. However, these students were also at greater risk for alcohol misuse, which ultimately impeded academic success and resulted in lower grades. Clearly, the question of whether and how early college classes help, or harm students is unresolved.
One potentially consequential way in which college student performance and engagement can be indirectly reflected is in student evaluations of teaching (SETs). Recent research suggests that SETs are problematic in many ways. SETs can encourage lenient grading and ineffective teaching (Stroebe, 2020) and they are positively correlated with students’ expected grades (Johnson et al., 2013). Further, students are not good judges of their own learning (Carpenter et al., 2020), and SETs can be biased by instructors’ race, ethnicity, national origin, and gender (Fisher et al., 2019; Wang & Gonzalez, 2020). Accordingly, some colleges have reduced the emphasis they place on SETs in tenure, promotion, and other evaluation processes, but they remain in widespread use. If the time at which a class is offered impacts the feedback that students provide on SETs, it would be critical for evaluative bodies to take this into account when reviewing faculty teaching portfolios.
In the current study, we add to the developing research on college class start times by testing whether the same Introductory Psychology course, taught by the same instructor in the same semester, yields different student outcomes when offered at 8 a.m. (before the generally advised 8:30 a.m. start time) versus 10 a.m. (the mid-morning time that some research suggests is best for adolescent learning; Dikker et al., 2020). We examined between-section differences in grades, course engagement and enjoyment, sleepiness, ease of waking up and staying awake for class, attendance, assignment completion, and SETs. We hypothesized that sleepiness would be higher in the earlier section, while grades, engagement, enjoyment, attendance, assignment completion, ease of waking and staying awake, and SETs would be lower.
Method
Participants
Participants were 82 undergraduate students enrolled in an 8 a.m. (n = 39) or 10 a.m. (n = 43) section of Introductory Psychology in Spring, 2021 at a private, mid-Atlantic, American small liberal arts college. Fifty participants (60.98%) identified as female, 31 identified as male, and 1 identified as non-binary. Most participants (n = 54, 65.85%) were freshmen, followed by sophomores (n = 19), juniors (n = 6), and seniors (n = 3). Age ranged from 18-22 years (M = 19.13, SD = 0.97). To preserve anonymity, we did not measure race; the typical makeup of the college is about 73.00% White, 8.00% Hispanic, 5.00% Black, 5.00% Asian, 4.00% multi-race, and 5.00% unknown or other.
Procedures
The Lafayette College Institutional Review Board approved all study procedures. Participants were drawn from the 104 students enrolled in the 8 a.m. section (n = 50 total students, 78.00% participation) and 10 a.m. section (n = 43 total students, 81.13% participation) of Introductory Psychology in Spring, 2021. Students chose which lecture section to enroll in during the course registration period. The sections were taught by the same instructor, and identical course material was covered at the same pace in both sections. All quizzes and exams took place on the same dates. Due to disruptions caused by the COVID-19 pandemic, students viewed pre-recorded lectures and completed assigned readings on their own time. They then attended a small-group, virtual (Zoom), live review session for that unit on a Monday, Wednesday, or Friday from 8 a.m. - 8:50 a.m. (section 1) or 10 a.m.-10:50 a.m. (section 3). Days were consistent throughout the semester and were pre-assigned based on alphabetical order, with some adjustments for student preference. Small groups ranged in size from 16-19 students. In addition to attending a large-group meeting at the beginning, midpoint, and end of the semester, each student was expected to attend a small-group, live meeting once per week. On the last day of each small group meeting (Monday 5/10/21, Wednesday 5/12/21, and Friday 5/14/21), the course instructor began class by verbally introducing the study and providing in the Zoom chat a link to the online consent form. The instructor then placed students in individual breakout rooms for 10 minutes so that the instructor did not know who was completing the survey. This process was repeated in the large-group meeting on the last day of the semester (5/19/21). Students who elected to take part in the study completed a Qualtrics survey that included the consent form, study questions, and a debriefing form. For their participation, individuals could provide their student identification number to enter a lottery into a drawing for one of 3 $25 prizes.
Materials
Participants completed the one-item Stanford Sleepiness Scale (1 = Feeling active, vital, alert, or wide awake to 7 = No longer fighting sleep, sleep onset soon, having dream-like thoughts; Hoddes et al., 1972), the five Likert-scale items from the College’s end-of-semester SET form (1 = Strongly disagree to 5 = Strongly agree, 1 = Poor to 5 = Excellent, 1 = Strongly discourage to 5 = Strongly encourage, as appropriate), and single-item Likert-scale assessments of course enjoyment (1 = Not at all to 5 = A great deal), ease of waking up for class, and ease of staying awake during class (1 = Extremely difficult to 5 = Extremely easy). They also reported their current cumulative grade point average (GPA), current lecture grade (distributed ahead of time, 0 - 100), and estimated live class sessions missed (1 = None to 5 = Four or more), level of engagement/participation (1 = Poor to 5 = Excellent), completion of assigned readings, and percent of recorded lectures viewed (1 = Less than 20% to 6 = 100%). Questions were presented in the same order for all participants. Survey questions are available in this paper’s online Open Science Framework (OSF) project (see Wenze & Charles, 2022).
Results
We used principal components analysis (PCA) to aid in decisions about data reduction. Outcome variables all correlated at least .30 with at least one other item and communalities were all above .30, suggesting good factorability. PCA indicated a strong factor including the 5 SET items, the question about class enjoyment, and the question about engagement and participation; this factor, which we call “Overall course rating” had an eigenvalue of 4.88 and explained 37.53% of the variance. The scree plot leveled off after this first factor, but a second, third, and fourth factor had eigenvalues just over one and explained between 8.00 and 13.00% of the variance. Therefore, using both factor loadings and judgments about underlying theoretical constructs, we combined questions about ease of waking up and staying awake for class into a second factor (“Ease of wakefulness”), we combined questions about completed readings and recorded lectures viewed into a third factor (“Assignment completion”), and we retained missed classes and the Stanford Sleepiness Scale (Hoddes et al., 1972) as single items.
Between-group comparisons were conducted using Mann-Whitney tests for ordinal data (e.g., class year, Likert scales) and independent samples t tests for interval data (e.g., GPA, lecture grade). Since maximum potential sample size was limited by course enrollment, a priori power analyses were not conducted. Post hoc power analysis using G*Power3 (Faul et al., 2007) indicated that we achieved adequate power for testing between-group differences for many of our key questions. For example, using a two-tailed Mann-Whitney test, a large effect size (d = 0.80), and an alpha of .05, we achieved a power of 0.94 to detect between-group differences in wakefulness. However, using a two-tailed t test, a medium effect size (d = 0.50), and an alpha of .05, achieved power was only 0.61 to detect between-group differences in grades.
Most (n = 78, 95.12%) participants reported that they were in the same time zone (Eastern Standard) as classes were held for most of the semester. Participants completed most surveys (n = 74; 90.24%) during small-group class meetings (5/10/21, 5/12/21, or 5/14/21); the remaining 8 surveys were completed during the final, large-group class meeting (5/19/21). Surveys took participants an average of 295.60 seconds (4.93 minutes, SD = 120.33 seconds). There were no missing datapoints except for current lecture grade; 4 participants declined to report this value. No data were excluded from analyses.
Between-Group Differences in Baseline and Outcome Variables.
aDescriptive statistics for categorical and ordinal variables are presented in Supplemental Table 1 in this paper’s online Open Science Framework (OSF) project (see Wenze & Charles, 2022).
GPA and Class Section as Simultaneous Predictors of Outcome Variables.
Current sleepiness, ease of wakefulness, and assignments completed: 8 a.m. versus 10 a.m. section.
Discussion
In the current study, students in an early-morning section of a college-level Introductory Psychology course had lower GPAs and class grades and reported completing fewer assignments, feeling sleepier, and having more difficulty with wakefulness than students in a mid-morning section of the same course. The association between class start time and difficulty with wakefulness held when controlling for baseline GPA, and the associations with assignment completion and current sleepiness became marginal. Although preliminary, these results are consistent with previous studies demonstrating the benefits of later class times for middle and high school students (e.g., Minges & Redeker, 2016; Owens et al., 2010; Thacher & Onyper, 2016), and suggest that such conclusions may extend to the college level. To maximize student success, universities should consider these results when creating class schedules (Hershner & Chervin, 2014). Pressures to maximize use of classroom space and to avoid conflicts with sports practices, rehearsals, clubs, and committee meetings often mean that universities cannot entirely avoid the early-morning scheduling block. Assessing students’ circadian preferences and, whenever possible, assigning class times accordingly could be one solution (e.g., Evans et al., 2017).
In contrast, controlling for section, GPA was a better predictor of grades and overall course rating, and GPA independently predicted assignments completed. Low-achieving students might struggle in their classes, regardless of the time of day. Study strategy interventions, which can improve grades among low-performing students in college science classes (e.g., Deslauriers et al., 2012), are recommended. Further, given evidence that student engagement in learning predicts grades, persistence, and degree completion (Svanum & Bigatti, 2009), use of classroom incentives to increase participation (Chylinski, 2010) is advised.
Contrary to expectations, class start time was not significantly associated with overall course rating in the current study. However, this relationship was in the expected direction, and we note that annual evaluations, tenure decisions, and more are typically based on visual inspection of raw numerical values rather than formal statistical analyses. Further, controlling for course section, GPA and overall course rating were positively associated. This finding is consistent with prior work (Richardson & Williams, 2021) and with research highlighting numerous problems with the use of SETs in higher education (e.g., Esarey & Valdes, 2020). Evaluative bodies should use caution when considering SET scores.
Strengths of this study include the fact that sections were matched on course instructor, semester, and class content, and the fact that most participants were first- and second-year college students; research suggests that, compared with juniors and seniors, freshmen and sophomores are at particular risk for sleep phase delay (Lund et al., 2010; Roenneberg et al., 2004). Further, although it might have obscured some potentially significant findings, data collection during a semester when students were only expected to attend a synchronous class meeting once per week offers a conservative test of whether start time is associated with student outcomes. Indeed, it is noteworthy that between-group differences emerged even with relatively limited real-time, live course instruction.
This study also has significant limitations. Our sample size was small, and the study was underpowered to detect some anticipated effects. The research design was quasi-experimental; students were free to enroll in the section of their choosing, rather than being randomly assigned. Students who are less aware of the potentially detrimental effects of taking early-morning classes might have chosen to enroll in the 8 a.m. section, perhaps in a bid to free up their schedules later in the day. Alternatively, some students who were less motivated or organized – and therefore later to complete their course enrollments - might have been locked out of the later section, which tends to be preferred and therefore fill faster during course enrollment. Controlling for GPA only partially mitigates these issues because GPA reflects many non-classroom factors (e.g., mental health, substance abuse, socioeconomic status; Bolin et al., 2017; Bruffaerts et al., 2018) that we did not measure. It is also possible that findings would have been more robust if we had been able to compare outcomes across sections that differed by more than 2 hours or if it had been possible to compare an 8 a.m. section to an afternoon section; one study combining survey methodology and a neuroscience-based theoretical model determined that 11 a.m. or 12 p.m. may be the optimal class time for maximizing student performance in college courses (Evans et al., 2017). Other limitations include a reliance on self-report data and lack of assessment of student circadian preference.
Despite these limitations, this study contributes to the limited research examining relationships between class start times and student engagement, achievement, and well-being at the college level. Future research should replicate these preliminary findings using an experimental design and objective behavioral, academic, and physiological data, such as repeated, real-time assessment of student sleepiness, weekly assessment of assignment completion and class participation, final course grades, and/or wrist actigraphy. Future studies should also measure important potential confounds, such as mental health, substance abuse, and socioeconomic status, as well as test whether findings hold for other kinds of classes (e.g., seminar, lab, studio), in other subject areas, and in courses that meet face-to-face and/or more than once per week. Findings might also differ by class year. Finally, these findings should be replicated now that COVID-related academic disruptions have abated. Research suggests that adolescents delayed their bedtimes and waketimes and shifted towards evening chronotypes during the pandemic (Genta et al., 2021). Further, morning-oriented students generally fared better during lockdowns and remote learning than evening-oriented students (Staller et al., 2021). It is therefore possible that some of the negative effects of early class start times might be mitigated following a return to pre-pandemic social and academic rhythms.
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.
Open Practices
Data Availability Statement
The raw data and analysis code used in this study are not openly available but are available upon request to the corresponding author. Data collection and analysis were not pre-registered. Study materials and a supplemental table are available in this paper’s online Open Science Framework (OSF) project at 
.