Improving student attendance: does it improve student learning? (The scholarship of teaching and learning).

Abstract

This study explores whether students with better class attendance perform better on assessment instruments. The relationship between test performance and student learning was also assessed. Students' attendance records were compared to their test scores on objective tests. The students also demonstrated their knowledge of course content on a comprehensive question that was unannounced so students could not study for it. The results show that student test performance is better when class attendance is better and that students with better attendance demonstrate more knowledge of the course material.

**********

There is a continuing discussion in the academic community around improving student learning. Related to this there is also considerable discussion about the importance of class attendance. This concern for attendance, and how to ensure students are maximizing their learning, is the foundation for this study.

The first question we asked ourselves was: If we improve attendance will students perform better on tests and learn more? We also questioned if we get an accurate assessment of student learning from tests and similar assessment tools routinely used in classes? While it is tempting to answer the second question with a resounding yes, there seems to be little evidence to support such an answer. This questioning of our assumptions led us to develop specific research questions as indicated below.

Many faculty in upper division courses seem to have experienced problems with students who do not appear to be prepared for advanced course material. The normal reaction to such a discovery is typically to question how the students made it through the lower level core or foundation course. One possible explanation for these observations is that students only memorized certain information or concepts long enough to pass a test, but they did not learn them. Another explanation might be that the student missed class on the day that certain topics were covered and the test that included that material did not ask every possible question because there was a limited time period allotted for the exam. This time constraint compelled the professor to ask a limited number of questions. Thus, the student "learned" enough to pass the test.

The emphasis on assessment by accreditation bodies suggests the academic community is continually grappling with this issue of assessing student learning. As noted by Rynes and Trank (1999) pedagogical research needs to begin assessing learning. Such evaluations need to go beyond grades and test scores to assess understanding (Frost and Fukami, 1997). This study evaluates the relationship between class attendance and student learning beyond just performance on standardized tests. That evaluation specifically assesses whether students who have better test scores also have a better understanding of course material. For the purpose of this study we are using student learning to imply an ability to understand course material, use that understanding in subsequent courses, and demonstrate that understanding even when the student is not given the opportunity to "cram" for an exam. We perceive learning as different from test performance in that learning is the acquisition of lasting knowledge and understanding of a subject.

The research on the effects of student attendance on learning has previously focused exclusively on the relationship between student attendance and test performance. In one of the few published studies of student attendance and performance Durden and Ellis (1995) report on a study of economics students. They conclude that absences did not negatively affect student performance on nationally standardized tests until they exceeded four classes. They further found that the negative effects on performance increased as absences increased. Thus, they conclude attendance does matter for academic performance and the effects of absences are nonlinear. Romer (1993) also found that attendance made a significant contribution to student performance, even when controlled for motivation. Both of these studies were conducted in economics courses. However, Park and Kerr (1990) examined effects of attendance on student performance in a money and banking course. They found that attendance was a determinant of student performance. These studies, the only ones we could find in a thorough search of business and education literature, clearly suggest there is added value to students who attend class. If students who attend class outperform those who are repeatedly absent, it seems reasonable to conclude that something is occurring as a result of student attendance. Therefore, we hypothesized that students who attend class regularly, not only perform better on tests but also learn more.

Thus, our conclusion leads to the primary objective of this study. If the ultimate goal is to increase student learning, can learning be improved as a result of improving class attendance? A review of research on behavior modification found that, on average, performance improved 17 percent when behavior modification techniques were used (Stajkovic and Luthans, 1997). Their findings led us to investigate whether behavior modification techniques could bring about improved attendance, i.e. a change in attendance behavior. Applying a behavior modification approach in an effort to significantly improve attendance as the first step to improving performance and learning was found to be successful (Ledman and Kamuche, 2000). This study extends that research to assess the impact of improved attendance on student performance and to assess the relationship between student test performance and the student's ability to demonstrate an understanding of the topic, as suggested by Frost and Fukami (1997). The three hypotheses we developed are given below.

H1: Students who attend class more frequently will perform better on identical tests developed by the professor.

H1a: Students who miss more than one class will perform less well on tests than students with one or fewer absences.

H2: Students who attend class more frequently will learn more.

H3: Students' test performance will correlate with their ability to demonstrate an understanding of the topic.

All of these hypotheses, except 1a, are meant to imply a linear correlation.

Method

Sample

The sample in this study is the students in the same faculty member's classes over two academic years. The typical enrollment in the classes was thirty students per section with two or more sections being taught each semester, thus yielding sample sizes of 118-170. No control for GPA was attempted since the sample size was so large.

Procedures

Attendance records were maintained during the study years. As previously demonstrated by Ledman and Kamuche (2000), student attendance was significantly improved with the use of incentives for excellent attendance and punishment for excessive absences. To test student performance, the students in the course used for this study were given the same test for all years of the study. Since that was an objective test in a quantitative subject (statistics) the possibility of grading bias was minimized. The performance of students on the tests was compared to their attendance.

The second part of this study assessed the relationship between test performance and learning. The students in the most recent semester were given an additional question on the final exam. That addition was not announced in advance so students did not know to study for it. The same bonus question was used in all classes and was developed by the professor with the intent of assessing whether the students retained any understanding of the subject. The question was based on text and lecture material. Therefore, the bonus question was one that required students to demonstrate specific knowledge of course content from throughout the semester. The unannounced question also required students to demonstrate their ability to apply the course concepts and clearly explain their use of those concepts. The bonus question was graded without knowledge of the student's identity since they identified their answers only by a numeric code. The students were given an incentive to do their best on the question by providing bonus points on their final exam grade for correct answers. Since the final exam represented thirty percent of the course grade the incentive was substantial.

Data Collection

College policies required faculty to maintain complete and accurate attendance records. These records and the professor's grade books served as the data sources for this study.

Data Analysis

This study is not intended to discover causation, but instead, is intended to explore whether a linear relationship exists between: student attendance and test performance; student attendance and learning; and student test performance and learning. Therefore, hypotheses one-three were tested using correlation analysis to measure the strength of the linear relationships. Hypotheses one and two were also tested using t-tests to determine if there was a difference in the mean test performance between and within treatments. Hypothesis 1a was tested using Mann-Witney U to compare student test performance against attendance. This test was used as a comparison of two sample sets--average test scores of students with one or fewer absences versus students with more than one absence.

Results

Table one shows the results of the correlation analysis for hypotheses one, two and three. The results were significant for all three hypotheses at the .001 level.

See issue's website <http://rapidintellect.com/AEQweb/spri02.htm>

The correlation between student test performance and number of absences was 0.90. That correlation suggests a very strong linear relationship between the number of student absences and their test performance. The correlation between student attendance and student learning was 0.80, again suggesting a strong linear relationship between student attendance and their ability to demonstrate learning of the course material. The correlation between student learning and student test performance was weaker that the previous two relationships but still significant. That correlation was 0.68. The t-tests comparing mean test performance between and within treatments showed no significant differences. The ANOVA tables below are provided to show the test of significance for the correlation analysis are given in Table Two.

See issue's website <http://rapidintellect.com/AEQweb/spri02.htm>

The results for hypothesis 1a revealed that student test performance declined significantly after more than one absence. The results of the Mann-Witney U test are shown in Table Three.

See issue's website <http://rapidintellect.com/AEQweb/spri02.htm>

Discussion

The focus of this study was on improving student learning. Presumably all faculty members are concerned about how their teaching contributes to students learning. We have undertaken to explore whether student attendance is related to learning from our teaching, without regard to techniques used. Previous studies have demonstrated that student attendance can be improved (Ledman and Kamuche, 2000) using behavioral management techniques. This study provides clear evidence that faculty desire for students to attend class is well founded. The extremely high correlation between student attendance and test performance clearly suggests that student attendance is quite important to student performance. This is the first study we are aware of that has investigated the degree of correlation between student absences and test performance. Clearly we cannot identify whether absences cause poor performance or if the worst students are more inclined to miss more classes. However, visual review of our class grade records and the lack of a one hundred percent correlation suggest at least some high performing students are inclined to be absent and some poorer performing students are present for every class.

Further study of the correlation between absences and student performance is needed in different course environments, especially in light of the findings for hypothesis 1a. Our findings suggest that student absences have a greater impact than previously found by Durden and Ellis (1995). They found that performance was not negatively impacted until after four absences where we found significant performance deterioration after more than one absence. That difference could possibly be explained by the differences in the courses. Durden and Ellis studied economics students while our sample is drawn from statistics courses. It may be that the extent of the detrimental test performance effects of student absences is partially a factor of the kind of course. More studies with different types of courses are needed to improve our understanding of how student performance is impacted by absences in, for instance, liberal arts, mathematics, social science, or fine arts courses.

Our results for hypothesis two are also a strong endorsement of the idea that class attendance is important. These findings clearly suggest that class attendance is important to learning. Following the suggestions of Frost and Fukami (1997) and Rynes and Trank (1999) we have made an attempt to begin to study whether students are learning by assessing that learning by means other than test scores. That we found a very high correlation between student learning and student attendance provides additional support for our findings in hypothesis one that student attendance is important for student performance. Our results also clearly suggest that attendance is important for student learning. Improved student learning is the rationale often used to justify attendance requirements. The argument is typically made that students who attend class regularly will learn more. In fact, that rationale is the foundation of the bonus point reward system the authors use in their classes. We argue that "giving" bonus points is our way of recognizing learning that is likely occurring that we are unable to fully measure with our regular assessment tools.

The results of hypothesis three at first appear to be the most surprising. While we still found a significant correlation between student test performance and students' ability to demonstrate that they learned the course material, that correlation was substantially lower that the previous hypotheses. This result, however, is precisely what we expected and provides some empirical support for our premise and the suggestion of Frost and Fukami (1997) that test scores are not a sufficient measure of learning. Clearly more studies are needed to confirm that tests are not a sufficient measure of learning. The fact that we found a higher correlation between student attendance and learning than we did between student test performance and student learning presents significant challenges for future pedagogical research. Our data clearly support the suggestion of Frost and Fukami (1997) and Rynes and Trank that we move beyond grades and test scores to assess learning.

Figure one is an illustration of the relationship between attendance, test performance, and learning based on this study. The figure illustrates that both test performance and learning are improved when students miss fewer classes. However, the link between student attendance and learning is slightly weaker while the link between student test performance and learning is weaker still.

See issue's website <http://rapidintellect.com/AEQweb/spri02.htm>

It should be noted again that our findings were based on a sample of students in statistics courses. The quantitative nature of these courses is such that they lend themselves to objective tests with precise answers. Further studies are needed to verify that our findings are consistent across disciplines. Such studies could present substantial challenges when the subjects in question are more abstract and tests are more subjective. Issues such as consistency in grading of subjective tests and measuring learning will need to be carefully considered in the research design.

Additionally, the question of retention is an important extension of this study. There needs to be research that explores how well students retain what they seem to have learned. For example, do students who appear to have learned more retain knowledge from a pre-requisite course and thereby perform better in subsequent courses? How long does that retention last? Will students who take subsequent courses in the term immediately following a pre-requisite do better than students who allow a term to pass between a course and its pre-requisite? The answers to these and similar questions will greatly add to our understanding of student learning and what improves it.

References

Durden, G. C. and Ellis, L. V. (1995). The effects of attendance on student learning in principles of economics. AEA Papers and Proceedings, 343-346.

Frost, P. and Fukami, C. V. (1997). Teaching effectiveness in the organizational sciences: Recognizing and enhancing the scholarship of teaching. Academy of Management Journal, 40 1271-1281.

Ledman, R. E. and Kamuche, F. (2000). Improving student attendance to improve student learning. Proceedings of the National Business and Economics Society, San Diego, CA.

Park, K. H. and Kerr, P. M. (1990). Determinants of academic performance: A multinomial logit approach. Journal of Economic Education, 21(2) 101-111.

Rynes, S. L. and Trank. C. Q. (1999). Behavior science in the business school curriculum: Teaching in a changing institutional environment. The Academy of Management Review, 24 808-824.

Romer, D. (1993). Do students go to class? Should They? Journal of Economic Perspectives, 7(2) 167-74.

Stajkovic, A. D. and Luthans, F. D. (1997). A meta-analysis of the effects of organizational behavior modification on task performance, 1975-1995. Academy of Management Journal, 40 1122-1149.

Robert E. Ledman, Morehouse College, GA Felix Kamuche, Morehouse College, GA

Robert, Assistant Professor of Management, received his Ph.D. in Management from Virginia Commonwealth University. His research interests focus on improving student performance and learning, labor relations, and diversity. Felix, Lecturer in Statistics and Decision Sciences, received his Ph.D. from the University of North Texas. His current research interests are statistical methods and pedagogy.