Abstract
Does attending lectures improve student performance? Using novel attendance data, we examine statistically the relationships between attendance and performance for first-year and third-year students. The relationship is moderately positive: very high attendance is significantly associated with an improvement in performance over very low attenders of between 5.3 and 12.8 per cent, depending on circumstances. Then, we provide qualitative evidence from in-depth interviews with students about their views and motives regarding lectures. We find a range of reasons why attendance may be less than complete, and conclude that attendance is related to performance in complex ways.
Acknowledgements
The authors gratefully acknowledge the insights they received from the student interviewees, the reviewers, data from Julia Bland and statistical advice from Dr Andrew Folkard.
Notes
1 ‘Lectures’ refers here to hour-long, lecturer-focussed teaching events, interspersed with varying amounts of student activity. These are distinct from ‘seminars’, which are principally student-focused and include more interactivity.
2 The Lancaster Environment Centre was formed in 2009 as a result of the merger of three academic departments of Lancaster University: Biological Science, Environmental Science and Geography.
3 Two or three essays are completed in the first year, and markers are aware of their students' identities.
4 The 24 interviewees are grouped as follows. Year 1, good attendance: James, Peter, Lesley, Anna, Jennifer, Lisa. Year 1, poor attendance: Alan, Simon, Stuart, Eamon, Emily, Charlie. Year 3, good attendance: Henrietta, Sarah, Rosie, Maggie, Toby, Alice. Year 3, poor attendance: Nadine, Jeremy, Ursula, Richard, Liam, Phil.
5 The lower classes are inflated in order to give statistically significant results later on, but this does not affect comparison of the within-class differences illustrated in Figure .
6 We use Pearson's correlation to ‘establish the strength of the relationship between two variables and the probability that the relationship has occurred by chance’ (Berman-Brown & Saunders, Citation2008, p. 89). This is why each test produces both a Pearson's correlation result, which is between − 1 to denote a perfect negative relationship and +1 to denote a perfect positive relationship, and a significance value.
7 Interpretations of what makes a ‘strong’ or ‘weak’ Pearson's correlation vary widely. We take a result between ± 1.0 and 0.7 to indicate a strong relationship, ± 0.7 and 0.3 to indicate a moderate relationship and ± 0.3 and 0.0 to indicate a weak relationship, but this classification, like any other, is necessarily subjective.
8 The classes used are the same as those in Figure , except that the bottom three classes were merged into one due to low counts.
9 ANOVA is an abbreviation of ‘analysis of variance’. It is used to determine ‘the likelihood of three or more distinct groups being different’ (Berman-Brown & Saunders, Citation2008, p. 88).
10 While ANOVA tests examine whether three or more means are different from each other, Tukey tests examine whether individual means are different from each other. They are generally used in conjunction with ANOVA tests and they compare all possible pairs of means in a sample.
11 The acronym LUVLE, which appears in several of the quotations, stands for Lancaster University Virtual Learning Environment, a VLE developed locally and on which PowerPoint files and other materials are stored for students to access during their course.