Key Predictor of Student Success: Why Attendance Matters Most in Challenging Times

Should college faculty track class attendance for their courses? As students face a number of challenges in attending class, such as feelings of anonymity and facelessness in large first-year courses (Oldfield et al. 2019; Romer 1993), or logistical issues like unreliable transportation or conflicts due to work, tracking class attendance may seem harmful rather than helpful. In the midst of the pandemic, with a large proportion of college courses being offered online or in a hybrid fashion, tracking attendance may seem even more time-consuming and cumbersome than in prior semesters, and maybe even feel punitive at a time when students are balancing increased financial instability, food insecurity, lack of access to technology or internet, and/or mental health issues.

At the same time, tracking who comes to class could help faculty and staff identify these issues earlier and prevent students from falling through the cracks.

As higher education reassesses how students learn, both in-person and online, and how we provide a well-rounded education, considerations of tracking class attendance should be at the forefront of these discussions. With a few exceptions, the literature is clear on the relationship between class attendance and performance: attending class lectures has a strong relationship with course performance and degree completion (Arulampalam et al . 2012; Cohn and Johnson 2006; Durden and Ellis 1995; Gump 2005; Marburger 2006; Stanca 2006). In this White Paper, we review the literature on the relationship between class attendance and performance, and how both faculty and staff can use the tracking of class attendance to identify resources students require, and thereby help to ensure a more holistic approach to student success.

“With a few exceptions, the literature is clear on the relationship between class attendance and performance: attending class lectures has a strong relationship with course performance and degree completion.”

There are a variety of reasons students skip class. Some of these reasons have to do with unavoidable circumstances such as a family emergency or illness, or the need to quarantine due to exposure to someone who is ill. There may also be personal reasons, such as lack of confidence, lack of motivation, feeling shy or uncomfortable in group settings, or being unprepared for the class session (Barlow and Fleischer 2011). Perceived teaching quality is also a prominent reason (Barlow and Fleischer 2011), with studies of students citing boring or unengaging lectures as a reason students do not attend class. Students might also presume the teacher does not notice or care if they are absent (Friedman et al. 2001). Students could also assume that the material covered in the class session is just a repeat of that in the textbook and online materials, and thus do not see utility in attending in person (Barlow and Fleischer 2011; Stripling et al. 2013). While some of these reasons for class absence may be unavoidable, others can be addressed through clear expectations for attendance, the reasons why attendance is important for course outcomes, and providing course content in an engaging way beyond repetition of what is in the course textbook.

It is also possible that students act as “rational actors” in their own best interests, knowing when attendance in a class session works to their benefit, or if that time could be better spent elsewhere. However, it is not entirely clear this is the case. In response to letters to the edit or regarding his seminal study (Romer 1993), Romer addressed the “rational actor” argument. Romer (1994), who advocated for at least experimenting with mandatory attendance policies (rather than recommending them outright), said, “in deciding whether to go to class, students are deciding whether to incur a current cost in return for a distant and uncertain benefit. Individuals may not be fully rational in making this kind of intertemporal tradeoff” (p. 215).

Numerous academic peer-reviewed research articles across many disciplines and institution types suggest that attendance does increase learning and course outcomes for students (Arulampalam et al. 2012; Cohn and Johnson 2006; Durden and Ellis 1995; Marburger 2006; Stanca 2006). In a metaanalysis of 52 published articles and 16 unpublished works regarding the relationship between class attendance and grades in college, Credé and his colleagues at the State University of New York at Albany show that class attendance has a stronger relationship not only with individual class grades, but also overall GPA. Race/Ethnicity & Gender Subgroups

“Class attendance is a better predictor of college grades than any other known predict or of academic performance, including scores on standardized admissions tests such as the SAT, high school GPA, study habits, and study skills” (Credé et al. 2010: 272).

This benefit is statistically and substantively significant too: other authors have seen a 2-4% boost to test scores when students attend lectures (Lin and Chen 2006; Marburger 2001). Intuitively, the relationship between class attendance and class performance might make sense. When students attend class they are presented with crucial information not covered in the textbook, and are presented with information from the textbook in a variety of engaging ways (Credé et al. 2010). But determining with research whether this relationship exists across the wide diversity of colleges and universities in the United States and the world and the wide diversity of college majors, courses, and teaching styles is tricky. Unsurprisingly, it can be difficult to truly capture whether a student’s attendance in class improves learning and class performance, or whether such a relationship is more a reflection of a motivated student’s desire to learn. Characteristics of students that predict class attendance – internal motivation, cognitive ability, amount of effort – also predict class performance. Therefore, research in this area may be subject to omitted variable bias (Stanca 2006).

Researchers have attempted to correct for this bias in a number of smart and creative ways. For instance, Chen and Lin approached the question of class attendance through a randomized experiment (Chen and Lin 2008). The authors estimated the impact of attending class for those who go to class through selectively lecturing on course material. Across two sections of the same class, a faculty member covered some material in Class A but not Class B and vice versa to see if exposure to material in class could be traced to performance on questions on the exams regarding that material. While all material on the exams were included in the textbook and lecture slides for each section (posted to the class website after class), each class received slightly different lectures. The authors then tracked how students did on exam questions covered and not covered in their sections.

The authors conclude that there was no difference in exam performance on questions covered in both sections. But when focused on questions covered in one section but not the other, “the effect of attending lectures corresponds to a 9.4 percent to 18.0 percent improvement in exam performance for those who choose to attend classes” (Chen and Lin 2008: 225). By controlling the material in this way, the authors could avoid the endogeneity issues that plague other research with variables such as motivation and effort.

Similarly, Daniel Marburger (2001) tracked which days students missed a class and the days certain exam questions were covered in class. He too found that those who were absent on certain days were much more likely to answer exam questions covered on those days incorrectly. Missing class reduced the overall class average score by 2.3 percent. Even across the grade distribution in a problem-based learning environment, attendance positively impacts credit completion (Bijsmans and Schakel 2018).

J. R. Rodgers (2001) compared OLS, fixed-effects (FEM), and random-effects models (REM) to determine if absence in an introductory statistics course had a negative effect on class performance. Finding that OLS estimates were biased upwards because of omitted variable bias, Rodgers determined that, using the FEM model, class attendance does positively predict performance, but that this effect is greater for those not chronically absent, upwards of a letter grade for some. Consequently, this and other research led Credé and colleagues to conclude that class attendance and student characteristics such as motivation and effort each separately influence academic performance (Credé et al. 2010).

Daniel Marburger (2001) tracked which days students missed a class and the days certain exam questions were covered in class. He too found that those who were absent on certain days were much more likely to answer exam questions covered on those days incorrectly.

The authors conclude that there was no difference in exam performance on questions covered in both sections. But when focused on questions covered in one section but not the other, “the effect of attending lectures corresponds to a 9.4 percent to 18.0 percent improvement in exam performance for those who choose to attend classes” (Chen and Lin 2008: 225). By controlling the material in this way, the authors could avoid the endogeneity issues that plague other research with variables such as motivation and effort.

Similarly, Daniel Marburger (2001) tracked which days students missed a class and the days certain exam questions were covered in class. He too found that those who were absent on certain days were much more likely to answer exam questions covered on those days incorrectly. Missing class reduced the overall class average score by 2.3 percent. Even across the grade distribution in a problem-based learning environment, attendance positively impacts credit completion (Bijsmans and Schakel 2018).

J. R. Rodgers (2001) compared OLS, fixed-effects (FEM), and random-effects models (REM) to determine if absence in an introductory statistics course had a negative effect on class performance. Finding that OLS estimates were biased upwards because of omitted variable bias, Rodgers determined that, using the FEM model, class attendance does positively predict performance, but that this effect is greater for those not chronically absent, upwards of a letter grade for some. Consequently, this and other research led Credé and colleagues to conclude that class attendance and student characteristics such as motivation and effort each separately influence academic performance (Credé et al. 2010).

While concerning, this evidence of the negative effect of absenteeism on course grades should not be our only focus. While course grades are one way to measure a student’s learning, not all learning can be adequately measured on a test or a paper. Much of our focus as faculty is on in-person discussions and conversations about the material.

“Knowledge is not just about reading the book and being tested on it; it is the conversations we have about that knowledge that give it meaning, and the classroom setting is of ten the primary— and perhaps the only—place where intellectual dialogues happen” (Hassel and Laourey 2005: 5). In addition, many faculty try to create a community within the classroom, which can be difficult when parts of that community are absent, and can make course discussions for the students who remain less stimulating and impactful (Barlow and Fleischer 2011).

Another reason to track class attendance that has become even more salient in college courses today is the negative mental health impacts of the coronavirus pandemic and the social distancing requirements that most colleges have enacted. Rather than track attendance just as a way to encourage students to come to class, many faculty have in the past and continue to track attendance to keep tabs on students (McIssac 2020).

In the absence of a mandatory attendance policy, tracking attendance can at the very least give faculty (and the advisors they work with) an opportunity to see if students are disengaging not just from their class but from the institution as a whole.

If a student misses a few course sessions in a row without notifying the professor ahead of time, the professor can reach out to the student to see if everything is okay, or if that student needs referral to academic, financial, and/or mental health support inside or outside the institution. Advisors noticing student absences in multiple courses could do the same.

This type of outreach by faculty and advisors signals to the student that there is someone who cares for them at the college, increasing their feelings of connectedness and their motivation to come to class.

Does this research about the positive benefits of class attendance mean that faculty should require it for their courses? The research regarding whether or not mandatory attendance policies increase student performance is not as clear-cut as the literature that tries to connect class attendance with class performance. Karen St. Clair (1999) argues attendance shouldn’t be mandatory, suggesting students could lose a sense of control over their education. Leaving it up to students to decide whether they attend class could encourage them to be more active when they do come, or at the very least exercise autonomy and take responsibility for their own education.

Other research suggests that, when applied to students doing poorly in a class at midterm, mandating class attendance can improve final exam performance (Dobkin et al. 2010). Credé and colleagues’ (2010) reanalysis of Kooker (1976) suggests that the policy may particularly help lowperforming students as well. On the other hand, Bratti and Staffolani in their study of an Italian university found that when class attendance is mandatory, students might reduce their self-study time in response (Bratti and Staffolani 2013). They posit that this could have a negative effect if students reduce their effort outside of class. However, this article assumes self-study and lecture attendance are equally good at increasing exam performance, which could be an untenable assumption (especially considering the results of Chen and Lin (2008) and Marburger (2001)), and is based on retrospective student self-report of attendance and time spent studying, which could be biased upwards and thus underestimate the effect of both.

Lastly, while Marburger (2006) finds that a mandatory attendance policy reduced class absences and improved student performance on the exam, he concluded that it wasn’t the policy itself that improved the material, as students who missed the material in both the policy and no policy conditions missed exam questions at the same rate. It was that the policy induced attendance, which increased performance. If faculty could induce attendance without a mandatory policy, they likely will achieve the same results.

Research is clear that students who are absent less from class perform better on exams and have higher overall GPAs, even when controlling for student motivation and effort. However, it is unclear from the research at this point regarding what it is about attendance that improves student course outcomes. Faculty could be using the time in class to cover concepts that are unclear in the readings, or maybe signaling the importance of a topic for exams in discussing the topic (Arulampalam et al. 2012). Or there could be an indirect effect of class attendance through class (Kim et al. 2019).

Regardless of the mechanism, it is clear that class attendance is important for student success, and tracking class attendance can help address issues that prevent Student Success before students disengage completely, especially amidst the increasing challenges of a pandemic.

As members of a community of higher education professionals, tracking attendance is not only important for class outcomes but also for overall student success. If a student is missing class, they may also be missing others as well (Arulampalam et al. 2012). Watermark takes information about students – including course attendance and grades – and uses this information to predict students who are at risk of not completing a course or dropping out altogether. The more faculty collecting this information and including it in Watermark, the greater chance an academic advisor can see a red flag for a student and intervene before the absences become problematic. Using Watermark, faculty can both encourage students to attend class and collect information on attendance to help predict overall student performance and increase student course and college completion.

Watermark connects to an institution’s SIS system to capture attendance data and incorporates this information into Watermark’s Student Success & Engagement. Attendance information within Student Success & Engagement allows for staff to have the necessary transparency to help inform the potential intervention opportunities. Attendance information can trigger automated early alerts within Student Success & Engagement allowing staff to perform proactive outreach and provide the necessary support to students. Attendance tracking within Student Success & Engagement makes it easy to capture student attendance, and more importantly, connect that information to learner analytics that recognize students at-risk of failing a class. The data collected through institutions that leverage Watermark confirms the conclusions in the research. For instance, at one institution, students missing at least one course meeting are 9-12% less likely to pass the course than students who attend all class meetings. This difference is both statistically and substantively significant. Class absences also negatively impact college persistence and graduation: students missing more than two class meetings within any course are between 5-6% more likely to drop out than other students.