Understanding student attendance in business schools: An exploratory study

International Review of Economics Education 17 (2014) 120–136

Contents lists available at ScienceDirect

International Review of Economics Education journal homepage: www.elsevier.com/locate/iree

Understanding student attendance in business schools: An exploratory study Andrew Mearman a, Gail Pacheco b,*, Don Webber a, Artjoms Ivlevs a, Tanzila Rahman a a b

Department of Accounting, Economics and Finance, University of the West of England, Bristol, UK Department of Economics, Auckland University of Technology, Auckland, New Zealand

A R T I C L E I N F O

A B S T R A C T

Article history: Received 19 August 2014 Received in revised form 9 October 2014 Accepted 26 October 2014 Available online 6 November 2014

There is considerable literature indicating that class attendance is positively related to academic performance. However, the narrative on what influences students’ decisions to attend class is scant. This article examines why students choose not to attend class through the use of a survey distributed to first year undergraduates. Regression results point to three main reasons for reduced attendance rates: (i) alternative sources of information; (ii) valuing attendance low on the priority ladder; and (iii) timing/scheduling constraints. The most significant driver of greater attendance levels was attitudinal differences amongst students, and in particular, students with extrinsic achievement motivations with regard to their education. ß 2014 Elsevier Ltd. All rights reserved.

JEL classification: A12 A13 A14 A22 Keywords: Student attendance Survey Virtual learning environment

1. Introduction Students skipping class appears to be an increasingly common phenomenon, and its prevalence is worrying due to potential negative impacts on not only the student, but also their peers, teachers, and even wider society. With regard to the negative impact on the student, there is a wealth of empirical evidence to support the notion that increased class attendance results in higher academic achievement (Caldas, 1993; Cohn and Johnson, 2006; Devadoss and Foltz, 1996; Gatherer and

* Corresponding author. Tel.: +64 9 9219999x5708. E-mail address: [email protected] (G. Pacheco). http://dx.doi.org/10.1016/j.iree.2014.10.002 1477-3880/ß 2014 Elsevier Ltd. All rights reserved.

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

121

Manning, 1998; Kirby and McElroy, 2003; Lamdin, 1996; Marburger, 2001; Newman-Ford et al., 2008; Paisey and Paisey, 2004; Park and Kerr, 1990; Rodgers, 2001; Romer, 1993; Schmulian and Coetzee, 20111; Woodfield et al., 2006). For example, Paisey and Paisey (2004) and Newman-Ford et al. (2008) document a strong positive correlation between attendance and academic performance at universities in Scotland and Wales respectively. Of course, correlation does not prove causation, and attendance may simply reveal a student’s underlying attitude/motivation for education,2 which then acts as the main driver of performance. If this is the case then further research is needed to understand the drivers of attendance rates, especially if there are other traits that deserve attention. One such trait may be gender: recent evidence from Woodfield et al. (2006) found clear differences in attendance rates by gender at the University of Sussex and their results highlight far higher absence rates for male undergraduates relative to their female counterparts. Their study also provides evidence which corroborates the view that a student’s attendance rate explains a significant degree of variance in academic performance, even after controlling for the influence of personality and cognitive ability indicators. Similar gender differences were also found by Clifford et al. (2011). While existing literature seems consistent in the conviction that attendance rates positively influence students’ performance (Caldas, 1993; Lamdin, 1996; Rau and Durand, 2000; Romer, 1993), it is inconsistent with respect to the reasons for low attendance, thus resulting in difficulties for those attempting to design policy to raise attendance rates. Romer (1993) suggested that high rates of absenteeism reflect students’ perceptions that teaching quality is poor and thus the belief that attendance would lead to little ‘academic gain.’ However, Woodfield et al. (2006) found that more than half of the students they surveyed were concerned about the work they missed following absences, indicating their belief that there were potential gains to be made from increased attendance. Overall, Woodfield et al. (2006) present evidence which suggests that absence is explained by a lack of application and conscientiousness of the student. Two recent trends in the UK may act as opposing forces with respect to the attendance of their students. First, substantial increases in student fees may create added incentive for students to attend class because each class foregone has a higher average cost. In contrast, advances in learning technology and the increased willingness of universities to utilise this technology – perhaps driven by a perceived need to satisfy paying customers – create structures in which students are more likely to elect not to attend class. This paper investigates the student attendance puzzle via application of a survey to first year students in a business school located in the UK. The remainder of this paper proceeds as follows: the next section presents a brief discussion of relevant literature. Section 3 outlines the data and provides details of the survey utilised. Section 4 contains the descriptive statistics and results of the empirical analysis. Conclusions and further directions for this research are provided in Section 5. 2. Literature review 2.1. Why attendance matters As indicated earlier, there is a multitude of evidence in support of a positive relationship between attendance and student performance. Collecting data from four U.S. universities, Devadoss and Foltz (1996) find that even after controlling for other influences that might reasonably be expected to influence performance, a student that attended all classes is likely to achieve a grade 0.45 points higher (representing an increase of three letter grades, e.g. a B- to A-) on average than a student who only attended half of their classes. Marburger (2001) attempts to capture cause-and-effect by compiling a panel data set on 60 students in an introductory microeconomics class at a U.S. University, recording which specific class periods students missed over the semester, and relating class content to specific exam multiple choice questions. Using probit analysis, Marburger estimates that absenteeism 1 This particular study investigated attendance patterns in an accounting class in South Africa, and while the authors find a positive and significant link between attendance and academic performance, the relationship is weak. 2 Recent work by Andrietti (2014) makes use of proxy variables to capture the effect of unobservable student traits (which may be potentially correlated with attendance) and still find a positive and significant impact of attendance on academic performance.

122

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

from the relevant class increased the probability of an incorrect response for the question at hand by as much as 14%. Employing panel data methodology that allows heterogeneity among students with regard to unobservables such as intelligence and motivation that can affect both attendance and performance to be controlled for, Rodgers (2001) finds attendance to have a small, but statistically significant effect on the academic performance of undergraduate students in an Australian introductory statistics course. Finally, Cohn and Johnson (2006) find that even after taking into account individual heterogeneity, class attendance is positively associated with academic performance for students in an introductory economics course. Investigating the potential for a reverse relationship between attendance and performance where low-performing students become less motivated to attend class, the authors present evidence that attendance at earlier segments of the course removes any effect that earlier test scores might otherwise have on subsequent attendance. In addition to the positive effect on academic performance, attendance is also often associated with development of important soft skills and personal development, such as responsibility, good work habits, and improved social skills (Bean, 1985; Cohn and Johnson, 2006). Just as importantly, attendance is also associated with student engagement and thus higher retention rates – an important consideration not just for the student, but in many instances for the state, as higher education is costly, and in many countries throughout Europe (including Ireland, Wales, and the UK) heavily subsidised (see Barlow and Fleischer, 2011; Bennett, 2003; Kelly, 2012). Finally, poor class attendance incurs a hefty opportunity cost in terms of the education facilities and resources that might otherwise have been put to use elsewhere, and results in lost income for the university (Bennett, 2003). It also conceivably impacts negatively on staff morale and the other students whose group work is disrupted by non-attendance of fellow group members (Bennett, 2003; Friedman et al., 2001; Brauer, 1994; Longhurst, 1999; Wyatt, 1992). 2.2. The determinants of attendance: what we (don’t) know so far Extant literature stresses the heterogeneity of students and their disparate motivations to attend class. Early research in this area by Laurillard (1979) argues that study strategies (and thereby attendance) are contingent on context, which implies that the structures created by the university and the individual tutor/lecturer will affect student behaviour. Broadly, research efforts have focussed on four over-arching factors: the practicalities of attending class, such as conflicting pressures for a student’s time (e.g. Cohn and Johnson, 2006; Gysbers et al., 2011; Kelly, 2012; Kottasz, 2005; Paisey and Paisey, 2004; van Blerkom, 1992); course characteristics, such as time of day and lecturer quality (e.g. Devadoss and Foltz, 1996; Dolnicar et al., 2009; Friedman et al., 2001; Gysbers et al., 2011; Newman-Ford et al., 2008; Paisey and Paisey, 2004; Romer, 1993; Wyatt, 1992); student attributes and inherent traits, such as learning style, personality, and gender (e.g. Cohn and Johnson, 2006; Crede et al., 2010; Dolnicar et al., 2009; Friedman et al., 2001; Gump, 2006; Paisey and Paisey, 2004; Woodfield et al., 2006); and, finally, student motivations and perceptions, such as the perceived value in attending class (e.g. Billings-Gagliardi and Mazor, 2007; Cohn and Johnson, 2006; Dolnicar, 2004, 2005; Gysbers et al., 2011; Romer, 1993; Woodfield et al., 2006). The practicalities involved in attending class continue to receive attention in the literature. Conducting an online open-ended engagement survey on students from the science school at the University of Sydney, Gysbers et al. (2011) find that logistics are the main reason for nonattendance, such as time-table clashes and transport difficulties, along with extra-curricular activities such as sport and work commitments. The time pressures experienced by students in higher education are also documented by other researchers, the most commonly cited being: travel distance/difficulties (Cohn and Johnson, 2006; Kelly, 2012; Longhurst, 1999); family and work commitments (Paisey and Paisey, 2004); and course-work due (Kottasz, 2005; Paisey and Paisey, 2004; van Blerkom, 1992). Class characteristics also appear to play an important role and several studies have found that the day and time of classes is important, although they disagree as to which days and times are best. For instance, Devadoss and Foltz (1996) find that students prefer classes held on Mondays, Wednesdays, and Fridays, and those held between 10 am and 3 pm. In contrast, while they find similar results with

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

123

regard to the time of day,3 Paisey and Paisey (2004) find that Monday classes are the least attended, while Newman-Ford et al. (2008) find no time relationship and Friday’s to be the least commonly attended. Other class characteristics that have been documented as positively influencing attendance are the quality of the lecturer – often defined as how engaging and experienced a lecturer is (Devadoss and Foltz, 1996; Dolnicar et al., 2009; Gysbers et al., 2011; Romer, 1993), and the difficulty of the paper content – in terms of both subject matter (e.g. math) and level of study (Devadoss and Foltz, 1996; Dolnicar et al., 2009; Romer, 1993).4 Meanwhile, there are no consistent findings with respect to class size, which Friedman et al. (2001) show size to be negatively related to attendance; while Devadoss and Foltz (1996) find the reverse. Most recently, attention has also turned to the introduction of the virtual learning environment, as an important aspect of class characteristics, and whether their existence can impact negatively on lecture attendance rates. For instance, Friedman et al. (2001) find that putting class content up on-line negatively affects attendance rates. In contrast, however, Gysbers et al. (2011) find that very few students prefer online resource materials, or find it equivalent, to attending a face-to-face lecture (just 6%), with students alluding to the fact that on-line resources cannot substitute fully for the live experience (e.g. non-verbal cues are missing). Rather, in contrast to fears that students will substitute live lectures for electronic resources if these are made too freely available, students appear to use online materials as a supplementary learning tool to attending lectures, enabling students to review lecture content at their own pace (see also Billings-Gagliardi and Mazor, 2007; Copley, 2007). All we can say with any certainty with regard to the role of virtual learning approaches on class attendance patterns is that more research is required to understand the exact nature of this relationship. The third strand of literature regarding determinants of attendance is that of the role of student attributes. Some researchers have argued that learning styles are the best predictors of the learning process (Kolb, 1984), and hence, arguably, attendance rates. A typology is offered by Honey and Mumford (1982) who argue there are four different learning styles: activists, reflectors, theorists and pragmatists. Involving themselves fully and without bias in new experiences, activists are happy to be dominated by immediate experiences. On the other hand, reflectors like to stand back to consider experiences with a tendency to observe them from different perspectives. The third type of student style is those that theorise, i.e. those that adapt and integrate observations into complex but logically sound theories. The final group are pragmatists, who are keen on trying out new ideas, theories and techniques to see if they work in practice. In terms of relating these styles to attendance, Sharif et al. (2010) found evidence that students with a predisposition towards activist-style learning tend to have lower attendance rates, relative to their peers. In contrast, King (1995) did not find a significant relationship between learning style and attendance per se, but did find a positive correlation between attendance and kinaesthetic learners – who are described as students that have to be engaged in order to learn. Finally, and employing a different typology of learning styles, Kember et al. (1995) find that engineering students geared towards surface-learning attend more classes than their deep-learning counterparts; presumably with little intrinsic interest in the subject matter, the surface-learners’ only guidance as to what they should be studying must come from the lecturer, thus they attend class more regularly. Other individual attributes worthy of empirical investigation are gender and personality. For instance, Woodfield et al. (2006) document clear differences in attendance rates by gender at the University of Sussex, their results highlighting far higher absence rates for male undergraduates relative to their female counterparts (see also Paisey and Paisey, 2004). They further find support for the hypothesis that greater levels of extraversion in males leads to worse study habits (in part arising due to the tendency to be distracted), while females score higher on agreeableness – traits which predispose them to a greater commitment to fulfilling course requirements, which include attending class. The literature is also not consistent with regard to the role of gender in attendance habits. Wyatt (1992) find that female first year students at a mid-west U.S. university skip more classes than their

3

See also Mattick et al. (2007) who find that classes starting at mid-day are better attended than early morning ones. Opposing evidence is put forward by Dolnicar et al. (2009) who find that both very easy or very hard papers are negatively associated with attendance. 4

124

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

male counterparts, while other literature finds no relationship between gender and attendance (e.g. Cohn and Johnson, 2006; Friedman et al., 2001; Kelly, 2012; van Blerkom, 1992). The final strand of the literature relates to student motivation. Students are motivated to attend lectures for a variety of reasons depending on the value they perceive lectures to have. Early work by Romer (1993) suggested that high rates of absenteeism reflect students’ perceptions that teaching quality is poor and thus the belief that attendance would lead to little ‘academic gain.’ More recently, Gysbers et al. (2011) argue that students can be viewed as strategic consumers who will optimise their time use in order to gain an advantage from their educational experience and weigh the educational, efficiency and social benefits of attendance against time and opportunity costs. Cohn and Johnson (2006) highlight that students gauge the importance of attendance based on the perceived benefits from interaction with the lecturer and with their classmates. Gysbers et al. (2011) also reveal that students who attended lectures stated that they enjoy the personal style of a lecturer, social interaction and the opportunity for peer assisted learning. On the other hand, Billings-Gagliardi and Mazor (2007) found that decisions to attend lectures were influenced by previous experiences with lecturers, predictions of what would occur during the class, personal learning preferences, and timespecific learning needs, with the overall goal being the maximisation of learning. From the discussion above we can see that we currently have limited and at times an inconsistent understanding of what drives student attendance. This is problematic for university staff seeking to increase attendance of their student populations as there is no clear direction for their efforts. Should educators focus on class characteristics, such as timetabling or course delivery structure, or would effort be better served providing resources, such as budgeting advice, to ease the conflicting time pressures on students? If attendance is driven primarily by student characteristics such as gender and personality, such efforts might, in fact, have very little impact. There also appears to be a gap in the literature relating student motivation to personal values and attitudes. Park and Guay (2009) relate these two concepts via the motivational processes of goal content and goal striving. The authors describe ‘values as preferences’ and ‘values as principles’ and in general ascribe to the view that the latter tend to be guiding principles that determine behaviour, and directly impact motivation. The following exploratory empirical analysis will make use of a survey instrument that covers many of the expected determinants (such as availability of material online, teacher quality, class characteristics, and socio-demographic information about the student), and also covers areas that have previously received scant attention (such as student attitudes towards the learning process, their values and motivations for attendance). The analysis will also be conducted for the full sample of respondents, as well as separately by gender, to add to the limited evidence available on the possible reasons for differences in attendance rates for males versus females. 3. Data Data were collected from a Business School located in the UK (University of the West of England, Bristol, UK) using a hardcopy questionnaire. The survey was distributed in four lectures that spanned all level 15 students within the faculty in the spring term; this allowed for the collection of data from students undertaking study in all disciplines taught in the Business School (Accounting, Business, Economics, Enterprise, Finance, HRM, Marketing, Operations and Strategy). The questionnaire was distributed in the week following the deadline for submission of module choices for level 2. All questionnaires were collected within the lecture time, and students were assured that their responses would be kept confidential and anonymous. We hoped that this would raise the response rate (as compared with allowing students to take away and return their questionnaires or by using online survey tools).6 Nevertheless, because the survey was conducted relatively late in the academic year, and because attendance tends to fall as the academic year progresses (Newman-Ford et al., 2008), only 286 responses were received out of a potential population of 987 students. We believe that the vast 5 Technically, in UK terms these students are at Level 4; however the modules are all at university level 1. We shall use ‘levels’ in this sense throughout. 6 As an extra incentive to complete the questionnaire, a prize of cinema vouchers was offered to a winner chosen randomly at a later date.

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

125

majority if not all students who attended class in the survey week completed the survey, and that the low numbers of survey responses is a direct illustration of attendance rates, rather than an indication of response rate. While these low numbers may be a source of concern, it also adds to the motivation behind this analysis: there is a growing need to understand the reasons for poor attendance. Additionally, this is an important caveat of this study, in that the sample may be biased towards the harder-working, higher-achieving students with relatively good attendance rates. Another related and potential limitation to acknowledge at this juncture is that the attendance rates in this study are selfreported, rather than independently measured. However, it is not clear whether we should expect any resulting measurement error to biased upward or downward. For example, while some students might regard it as socially desirable to report high attendance, others might avoid reporting high attendance for fear of appearing too keen. The questionnaire was geared towards mainly quantitative analysis, deploying predominantly closed questions which were pre-coded. All survey questions are provided in Appendix 1, and the choice of questions was motivated by the studies outlined in the literature review. The questionnaire was laid out into separate sections. The first concerns module choice (see questions 1 and 2) and the factors that affect it. Questions 3–5 of the survey focused on attitudes and aptitudes (see Gump (2006), who argues for the importance of controlling student attitudes when investigating determinants of class attendance). While the latter of these were self-reported, the questionnaire included a question to cross-check students’ perception of ability against reality. Specifically, question 4 assesses students’ perceived ability in a range of areas (such as verbal, organisational and problem solving), while question 27 asks for the actual test marks received in three core classes – Economic Principles, Global Business Context, and Economics for Business and Accounting. For the purpose of this study, questions 6–8 are most relevant as they cover attendance patterns and motivations. For example, possible responses to question 8, which sought to identify why the respondent does not have full attendance, include clashes with social activities, being able to pass without full attendance, conflicts with work schedule, etc. The remaining questions in the survey seek to gather biographical information, which may be relevant in identifying underlying factors that affect both attendance and module choice. 4. Analysis Table 1 provides descriptive statistics for question 7, which elicited self-perceived attendance rates; and question 8; the reasons for less than full attendance rates. Attendance rates are divided into quintiles (0–19%, 20–39%, 40–59%, 60–79% and 80–100%). While such categorisation can be viewed as an unnecessary collapse of continuous information into specific brackets, it may aid in reducing the influence of measurement error on our analysis. For instance, given that one of the usual limitations of self-reported attendance rates are that they are subject to measurement error, as a result of flawed memory and/or social desirability bias, the use of categories of attendance rates means that such cases of measurement error should be reduced. Such error will only affect our analysis if they have a substantial impact on the boundaries between categories. As explained earlier, this sample is potentially biased towards the relatively harder-working students who were more likely to attend when the questionnaire was distributed. As a result of this, just over half the respondents indicated their attendance rate was in the top category (80–100% of classes); and just over 11% of the sample responded that they attend less than 60% of classes. In terms of reasons for less than full attendance, reasons that did not feature heavily on students’ minds were ‘Length of class is too long’; ‘I don’t feel part of the class‘; and ‘I have constraints due to dependents’: each reason receiving at a maximum 3% of the blame from survey respondents. At the other end of the spectrum, more than half the students (54%) blamed illness and/or tiredness. The other big drivers of poor attendance were low stimulation (52%), poor teaching quality (51%), and the timing of the class (52%). Interestingly, 41% of students indicated they did not need to attend class fully due to the substitutability of materials posted on the University’s Virtual Learning Environment (VLE) (in this case, Blackboard). In question 6, we also asked students when they downloaded material from the VLE. As shown in Table 2, approximately one-third of our sample downloaded material prior to the lecture.

126

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136 Table 1 Attendance rates and reasons. Attendance rates (q7) – proportion of classes attended

Percent of sample

0–19% 20–39% 40–59% 60–79% 80–100%

0 3.35 7.76 33.49 55.50

Reasons for less than full attendance (q8)

Mean (std. dev.)

q8_1: Completing assignments at the last minute q8_2: Length of class is too long q8_3: Classes are not stimulating q8_4: I can pass modules without attending all classes q8_5: Material is available on Blackboard q8_6: My friends don’t attend q8_7: Clashes with social life q8_8: I don’t feel relaxed in class q8_9: I need to work to earn money now q8_10: Class attendance is not compulsory q8_11: I take material and information from friends q8_12: I cannot understand lessons q8_13: Illness or too tired q8_14: The teacher is uninspiring q8_15: Time of day of class q8_16: Travel/commuting problems q8_17: I don’t feel part of the class q8_18: I have constraints due to dependents Number of observations

0.38 0.04 0.53 0.24 0.41 0.17 0.16 0.07 0.10 0.27 0.13 0.10 0.58 0.51 0.53 0.25 0.04 0.01 209

(0.49) (0.19) (0.50) (0.43) (0.49) (0.37) (0.37) (0.25) (0.30) (0.44) (0.34) (0.29) (0.49) (0.50) (0.50) (0.44) (0.19) (0.10)

A similar proportion download during the week after the lecture, and just over 6% never download the material. The Pearson chi2 value is highly statistically significant (at the 1% level) suggesting that we can reject the null hypothesis that attendance is independent of time of download and we also find a negative relationship between attendance rates and use of VLE, with a correlation ratio of 0.31 between q6 and q7. This suggests that downloading from the VLE may be a substitute for attendance in general. However, closer inspection of Table 2 shows that although this negative relationship is particularly present for poor attenders, the opposite is likely to be the case for high attenders who seem to download earlier, and hence may view VLE material as a complement to attendance that enables more effective learning. Consequently, the relationship between attendance and downloading material from the VLE is far from simple.The next step in our exploratory analysis was to compare the descriptive portrait of students, by attendance rates. Table 3 provides descriptives of the most relevant variables from the survey instrument for the full sample versus ‘good attenders’, which we define as those students that report at least 80% attendance (n = 116). Relative to the full sample, and in terms of socio-demographic characteristics good attenders were less likely to be male, have had a parent study Table 2 Timing of VLE downloads by attendance rates.

Poor attendance 0% < attendance < 60% Average attendance 60% < attendance < 80% Good attendance 80% < Attendance < 100% Total Pearson chi2 (8) = 29.58 (p = 0.000).

Prior to lecture

During the week after lecture

End of term

Before the exam

Never

N

17.39%

13.04%

13.04%

39.13%

17.39%

23

21.43%

41.43%

8.57%

21.43%

7.14%

70

43.97%

32.76%

9.48%

9.48%

4.31%

116

33.49%

33.49%

9.57%

16.75%

6.70%

209

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

127

Table 3 Descriptive statistics. Variable

Socio demographic characteristics (q17–20) Male Ethnicity not White Age Father studied at university Mother studied at university Sibling studied at university

Mean (std. dev.) Good attenders

Full sample

0.55** (0.50) 0.21 (0.41) 19.31 (1.25) 0.28 (0.45) 0.21 (0.41) 0.42 (0.50)

0.61 0.19 19.29 0.31 0.23 0.42

(0.49) (0.39) (1.20) (0.46) (0.42) (0.49)

Values/attitudes of respondent (q5) Do you agree: (1 = Strongly Disagree, . . ., 5 = Strongly Agree) q5_1: I am ambitious q5_2: Do not expect my job to be fulfilling q5_3: Expect to change career several times q5_4: Annoyed that the programme is so hard q5_5: Immediately before exams get nervous q5_6: Learn more if tutorial full of capable students q5_7: Some classes interesting, others boring q5_8: The idea of giving up studies is appealing q5_9: Important to perform well at university q5_10: Put a lot of effort to understand everything q5_11: Degree will be beneficial to future job q5_12: Important to parents I perform well q5_13: Degree is interesting q5_14: Want good grades, so I work hard q5_15: If others in tutorial work hard, it makes me work hard too q5_16: Can keep up with requirements of course q5_17: Am very bored during classes q5_18: Degree is fun q5_19: The smarter the other students in seminar, the harder I work

4.43*** (0.64) 2.24 (0.90) 2.84 (0.87) 2.31 (0.90) 3.75 (1.07) 3.79 (1.04) 4.11 (0.79) 2.35* (1.18) 4.78*** (0.46) 4.19** (0.62) 4.45*** (0.65) 4.18 (0.84) 4.16*** (0.62) 4.58*** (0.50) 4.09 (0.82) 4.15*** (0.66) 2.97 (0.83) 3.16* (0.87) 3.59 (0.85)

4.32 2.29 2.92 2.39 3.66 3.72 4.14 2.47 4.64 3.89 4.24 4.15 4.01 4.35 4.02 4.04 3.04 3.06 3.52

(0.68) (0.99) (0.88) (0.87) (1.12) (0.98) (0.74) (1.16) (0.55) (0.78) (0.77) (0.85) (0.72) (0.67) (0.80) (0.64) (0.83) (0.86) (0.89)

Reasons for less than full attendance (q8) q8_1: Completing assignments at the last minute q8_2: Length of class is too long q8_3: Classes are not stimulating q8_4: I can pass modules without attending all classes q8_5: Material is available on Blackboard q8_6: My friends don’t attend q8_7: Clashes with social life q8_8: I don’t feel relaxed in class q8_9: I need to work to earn money now q8_10: Class attendance is not compulsory q8_11: I take material and information from friends q8_12: I cannot understand lessons q8_13: Illness or too tired q8_14: The teacher is uninspiring q8_15: Time of day of class q8_16: Travel/commuting problems q8_17: I don’t feel part of the class q8_18: I have constraints due to dependents

0.28*** (0.45) 0.03 (0.18) 0.47* (0.50) 0.15*** (0.36) 0.31*** (0.46) 0.08*** (0.27) 0.06*** (0.24) 0.05 (0.22) 0.10 (0.31) 0.15*** (0.36) 0.09* (0.29) 0.08 (0.27) 0.54 (0.50) 0.48 (0.50) 0.38*** (0.49) 0.24 (0.43) 0.03 (0.16) 0.01 (0.09)

0.38 0.04 0.53 0.24 0.41 0.17 0.16 0.07 0.10 0.27 0.13 0.10 0.58 0.51 0.53 0.25 0.04 0.01

(0.49) (0.19) (0.50) (0.43) (0.49) (0.37) (0.37) (0.25) (0.30) (0.44) (0.34) (0.29) (0.49) (0.50) (0.50) (0.44) (0.19) (0.10)

Note: Good attenders reported attendance at 80% or more of class time, n = 116. Full sample n = 209. * 10% significant difference in means between the good attenders sample (n=116) and the remainder of the sample (n=93). ** 5% significant difference in means between the good attenders sample (n=116) and the remainder of the sample (n=93). *** 1% significant difference in means between the good attenders sample (n=116) and the remainder of the sample (n=93).

at university; and more likely to be non-white. The remainder of Table 3 is devoted to the variables expected to influence attendance rates directly – question 5 (attitudes), and question 8 (reasons for lack of full attendance). Attitudinal descriptives in Table 3 provide a clear picture of some of the key differences between good attenders and the full sample. The former of which are much more likely to be ambitious and

128

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

have the ethic of ‘Want good grades, so I work hard’. These good attenders are less likely to express feelings of boredom with classes, or think about giving up. They also place greater importance on parental feelings and peer effects (‘learn more if tutorial full of capable students’; and ‘the smarter the other students in seminar, the harder I think’, etc.). The final section in Table 3 compares the descriptives for reasons behind lack of full attendance for the good attenders versus the full sample. It appears clear that good attenders are less likely to blame most of the reasons available, and in particular ‘Completing assignments at the last minute’, and ‘Material available on Blackboard’. This is indicative of good attenders being harder working students, who are probably more organised, and not often forced to complete assignments in the last minute; as well as using Blackboard as more of a complement, rather than a substitute for attendance. 4.1. Factor analysis An intuitive next step in our exploratory journey is to conduct factor analyses with respect to the core sets of independent variables within question 5 and 8 in an attempt to reveal the individual variables that cluster into components which may be useful in predicting attendance. As Table 4 shows, there are nineteen variables from question 5 that encompass a range of values/ attitudes on behalf of the survey respondent. The Kaiser–Meyer–Olkin Measure of Sampling Adequacy statistic indicates a large proportion of the variance of our variables is caused by five underlying factors (KMO = 0.736) and each of the variables had high individual measures of sampling adequacy (KMO no less than 0.5). Selecting only those factors with eigenvalues greater than one achieved a total explained variance of 53.5%. Application of this factor analysis reveals the existence of four coherent factors, which we name ‘Achievement’, ‘Interest’, ‘Peer’ and ‘Quit’. The first factor brings together variables which explicitly discuss effort and success. The ‘Interest’ factor coheres around contradictory statements (the degree is fun vs. boring) which can feed into initiatives that can emphasise relevance or change teaching techniques. The ‘Peer’ factor concerns the effects of social dynamics on learning and effort (and, by implication, attendance). For instance, this factor contains high positive loadings for q5_19 (‘The smarter the other students in seminar, the harder I work’) and q5_15 (‘If other students in tutorial work hard, it makes me work hard too’). The factor denoted ‘Quit’ comprises parts of q5 concerned with low motivation, low expected satisfaction, giving up (both professionally and academically) and annoyance that the programme is hard (q5_4). This variable suggests that lower confidence may act to Table 4 Factor analysis on question 5 (values/attitudes). Achievement Interest Peer q5_10: Put a lot of effort to understand everything q5_14: Want good grades, so I work hard q5_11: Degree will be beneficial to future job q5_9: Important to perform well at university q5_1: I am ambitious q5_13: Degree is interesting q5_5: Immediately before exams get nervous q5_17: Am very bored during classes q5_7: Some classes interesting, others boring q5_18: Degree is fun q5_19: The smarter the other students in seminar, the harder I work q5_15: If other students in tutorial work hard, it makes me work hard too. q5_6: Learn more if tutorial full of capable people q5_16: Can keep up with requirements of course q5_3: Expect to change career several times q5_4: Annoyed that the programme is so hard q5_2: Do not expect my job to be fulfilling q5_8: The idea of giving up studies is appealing q5_12: Important to parents I perform well

0.855 0.767 0.705 0.608 0.453 0.383 0.382

Quit

Else

0.344 0.647 0.674 0.634 0.468 0.721 0.863 0.740 0.619 0.328

0.303 0.643 0.631 0.623 0.453 0.826

Note: Extraction Method: Principal Component Factor Analysis. Rotation Method: Oblique Promax with Kaiser Normalisation. KMO = 0.736.

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

129

reduce apparent interest and motivation, and thence attendance. The fifth and final factor in Table 4 is labelled ‘Else’ and encompasses just one option of q5 – ‘Important to parents I perform well.’ The eighteen variables obtained from q8 (reasons for attendance) were included in another factor analysis. The variables that had a low KMO (below 0.5) were systematically dropped from the analysis. This resulted in twelve remaining variables, and an overall KMO of 0.703. Selecting only those factors with eigenvalues greater than one achieved a total explained variance of 50.5%, and four factors. These results are shown in Table 5. The ‘Alternative sources of info’ factor corresponds to an ease of gathering information from friends and via Blackboard. The factor denoted ‘Low priority’ corresponds to variables indicating the student may be effort minimising, e.g. ‘I can pass modules without attending all classes’, and ‘Class attendance is not compulsory’. This may correspond to a ‘surface’ learning approach where achievement of a pass is the basic motivation driving the individual. The third factor ‘Schedule’ signals time conflicts, and general issues around time management such as ‘Completing assignments at the last minute’ and ‘Time of day of class’. The fourth and final factor in Table 5 is ‘Not belonging/exclusion’ and is made up of just two factors ‘I don’t feel part of the class’ and ‘I cannot understand lessons’. These drivers may also relate to the student’s (lack of) confidence around whether they should be part of the class. 4.2. Regression analysis The statistical analysis thus far has painted a comprehensive portrait of level 1 students in our sample and an exploratory investigation into motivations behind lack of attendance. Additionally, factor analyses on q5 and q8 yielded some informative factors that may explain patterns in attendance rates. Our next and final step in this exploratory study is to estimate an ordered logistic regression with attendance outcomes as the explanandum and the regressors being a combination of the factors just generated along with other personal/demographic characteristics. The regression is applied to the whole sample and each gender separately. The regression results are shown in Table 6. It is important to note that no respondents self-reported that they attend 0–19% and given the small numbers in the categories of 20–39% and 40–59%, these two categories are aggregated. Consequently, there are only three operative categories in this Likert scale response variable (20–59%; 60–79%; and 80–100%), and two cuts. Table 6 reveals several key findings. ‘Achievement’ has a strong and positive influence on attendance. This factor encompassed extrinsic values of students, where they strongly agreed with statements such as ‘Want good grades, so I work hard’; ‘Degree will be beneficial to future job’; and ‘I am ambitious’. These are attitudes that show interest in the end goal of university education, rather than the process of learning per se. The impact of this factor is the greatest in Table 6, and similar for both males and females. Students with higher levels of this factor are two-three times more likely to move up a category of attendance rates.

Table 5 Factor analysis on question 8 (reasons for not full attendance). Alternative sources of info q8_5: Material is available on Blackboard q8_11: I take material and information from friends q8_6: My friends don’t attend q8_1: Completing assignments at the last minute q8_4: I can pass modules without attending all classes q8_7: Clashes with social life q8_3: Classes are not stimulating q8_10: Class attendance is not compulsory q8_15: Time of day of class q8_13: Illness or too tired q8_17: I don’t feel part of the class q8_12: I cannot understand lessons

0.689 0.665 0.608 0.350 0.330

Low priority

Schedule

Not belonging/ exclusion

0.549 0.590 0.787 0.435 0.424 0.321

0.524 0.609 0.680 0.731 0.723

Note: Extraction Method: Principal Component Factor Analysis. Rotation Method: Oblique Promax with Kaiser Normalisation. KMO = 0.703.

130

Table 6 Ordered logistic regression: explaining the proportion of classes that the respondent attends (q7). Males

Odds ratio N Achievement Interest Peer Quit Alternative sources of information Low priority Schedule Not belonging/exclusion Access Blackboard: Prior to lecture During week of lecture End of term Before exam Never

(Std. error) 209 (0.604)*** (0.235) (0.167) (0.160) (0.121)** (0.126)** (0.095)*** (0.188) (0.624)

3.370 1.038 1.154 1.038 0.507 0.769 0.433 0.998 0.781

0.863 0.458 0.530

(0.497) (0.208)* (0.346)

0.840 0.721 0.372

Non-white Male Parent studied at university Either brother or sister studied at university Age Grade

1.146 1.193 0.870 0.995 0.877 0.997

(0.545) (0.439) (0.870) (0.995) (0.877) (0.997)

2.088 – 0.981 1.357 0.703 1.013

Cut1 Cut2 Pseudo R2

5.866 2.971

(3.215)

9.767 5.849

Log likelihood *

Signifies statistical significance at the 10% level. ** Signifies statistical significance at the 5% level. *** Signifies statistical significance at the 1% level.

142.925

(Std. error)

Odds ratio

128

2.760 1.196 0.988 0.905 0.612 0.616 0.553 0.986 1.340

0.269

Females

Odds ratio

81

(1.038)*** (0.242) (0.261) (0.249) (0.507)*** (0.769) (0.433)*** (0.269) (0.480) (Base category) (0.631) (0.407) (0.327) (1.119) (0.497) (0.620) (0.134)* (0.019) (4.284) 0.333 79.77

(Std. error)

3.378 2.187 0.942 0.631 0.963 0.384 0.660 0.871 1.568

(1.377)*** (0.946)* (0.334) (0.260) (0.305) (0.141)*** (0.213) (0.271) (1.313)

0.218 0.122 0.954

(0.397) (0.142)* (1.193)

0.900 – 0.610 0.518 1.230 0.975

(1.129) (0.550) (0.402) (0.363) (0.022)

1.043 1.309

(6.013) 0.355 142.925

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

Whole sample

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

131

Variables that appear to be exerting a downward influence on attendance rates are ‘Alternative sources of information’, ‘Schedule’ and ‘Low priority’. The first of these relate to accessing information from friends or the VLE. This factor also appears to be more significant for males, relative to females. Perhaps this is a signal that males in this business programme are more likely to view online resources as a substitute to going to class, relative to their female counterparts. Additionally, time management issues that surround the factor ‘Schedule’ are also more of an influence for males, relative to females. The opposite is true for the factor ‘Low Priority’. This factor includes effort minimising variables such as the belief that the respondent can pass the module without attending all classes, and that class attendance is not compulsory. This factor is more dominant for females compared to males. This result may be indicative of a cluster of female students who may be over confident and believe they can pass modules without attending classes. Finally, none of the socio-demographic characteristics were statistically significant in explaining variation in attendance rates, except a marginal result showing that older males are less likely to increase attendance levels. This may complement the finding that scheduling issues and concerns around time management are more prevalent for males, and you would imagine there may be more of these worries if older and more likely balancing study with work or other commitments. 5. Conclusions This paper employed a survey of level 1 students in a British Business School to explore reasons for poor attendance rates. There was initial descriptive evidence that students who viewed class materials online as a substitute were poor attenders, while good attenders tended to download these materials prior to the lecture and thus likely viewed them as a complement. Furthermore, application of regression analysis revealed that ‘alternative sources of information’ was one of three key factors that significantly acted to hinder attendance rates. This corroborates findings by Friedman et al. (2001) who found putting class content up on-line negatively affected attendance rates. In this study, the factor ‘alternative sources of information’ encompasses sourcing class material from either online or from friends, and appeared to be more significant for males, indicating that this group may be placing equal or similar value on information from friends or Blackboard, relative to that gained in class. The other two factors that appear to significantly dampen attendance levels are ‘Low priority’ and ‘Schedule’. The first of these factors is based on intentions of minimising effort, e.g. can pass the module without attending all classes. This was more prevalent for females, while scheduling issues seemed to be more of a concern for males. Perhaps educators could focus efforts/resources on initiatives to increase the time management skills or other steps aimed at resolving scheduling conflicts for the male sub-group within the student population. At the other end of the spectrum, the key factor that greatly increased the likelihood of increasing attendance was achievement values/ attitudes. These included mostly extrinsic motivations where the student was ambitious and wanted positive outcomes at the end of their education, such as good grades, or a good job. Given the potential for sample selection bias, that many of the students surveyed are more likely to be harder-working and more likely to attend, further investigation of attendance is clearly necessary. One possible direction for such research is a qualitative study, as we believe this would allow us to explore individual cases (and possible connections between them) most effectively. Focus groups and individual interviews of students might be deployed to explore different student types and different combinations of factors affecting attendance among these types. In particular, researchers must target ‘poor’ attenders who are under-represented in our sample. Clearly also, as this study – in line with most in the extant literature – examines only one university, another future direction for this strand of research is to better understand the generalizability of these results across different areas of the university system, relative to the business school; as well as exploring the specificity of results by country. Finally, we should note that for the educator concerned with attendance, there are alternatives. This paper has focused on encouraging attendance by understanding student motivations. Another approach is to be more coercive. The use of explicit attendance policies was highlighted by Self (2012), who finds strong support for using such policies to reduce absenteeism. In particular, Self (2012) found that policies that punish students for missing class rather than ones which reward students for good attendance, were the most effective.

132

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

Acknowledgements The authors would like to thank conference participants for their helpful feedback at the Developments in Economics Education Conference and Jessica Dye for invaluable research assistance. Appendix 1. Survey questions

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

133

134

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

135

References Andrietti, V., 2014. Does lecture attendance affect academic performance? Panel data evidence for introductory macroeconomics. Int. Rev. Econ. Educ. 15, 1–16. Barlow, J., Fleischer, S., 2011. Student absenteeism: whose responsible? Innov. Educ. Teach. Int. 48 (3) 227–237. Bean, J.P., 1985. Interaction effects based on class level in an explanatory model of college student dropout syndrome. Am. Educ. Res. J. 22 (1) 35–64. Bennett, R., 2003. Determinants of undergraduate student drop out rates in a university business studies department. J. Furth. High. Educ. 27 (2) 123–141. Billings-Gagliardi, S., Mazor, K.M., 2007. Student decisions about lecture attendance: do electronic course materials matter? Acad. Med. 82 (10.) S73–S76. Brauer, J., 1994. Should class attendance be compulsory? J. Econ. Perspect. 8 (3) 205–215. Caldas, S., 1993. Re-examination of input and process factor effects in public school achievement. J. Educ. Res. 86 (4) 206–214. Clifford, M., Willis, L., Eastwick, C., 2011. Angels and Ghosts: The Demographics of Lecture (Non-) Attendance. Available at: www.sefi.be/wp-content/papers2011/T7/23.pdf. Cohn, E., Johnson, E., 2006. Class attendance and performance in principles of economics. Educ. Econ. 14 (2) 211–233. Copley, J., 2007. Audio and video podcasts of lectures for campus-based students: production and evaluation of student use. Innov. Educ. Teach. Int. 44, 387–399. Crede, M., Roch, S.G., Kieszczynka, U.M., 2010. Class attendance in college: a meta-analytic review of the relationship of class attendance with grades and student characteristics. Rev. Educ. Res. 80 (2) 272–295. Devadoss, S., Foltz, J., 1996. Evaluation of factors influencing student class attendance and performance. Am. J. Agric. Econ. 78, 499–507. Dolnicar, S., 2004. What makes students attend lectures? The shift towards pragmatism in undergraduate lecture attendance. In: Conference Proceedings of the Australian and New Zealand Marketing Academy, 29 November–1 December 2, Wellington, New Zealand. Dolnicar, S., 2005. Should we still lecture or just post examination questions on the web? The nature of the shift towards pragmatism in undergraduate lecture attendance. Qual. High. Educ. 11 (2) 103–115. Dolnicar, S., Kaiser, S., Matus, K., Vialle, W., 2009. Can Australian universities take measures to increase the lecture attendance of marketing students? J. Mark. Educ. 31 (3) 203–211. Friedman, P., Rodriguez, F., McComb, J., 2001. Why students do and do not attend classes: myths and realities. Coll. Teach. 49 (4) 124–133. Gatherer, D., Manning, F.C.R., 1998. Correlation of examination performance with lecture attendance: a comparative study of first-year biological sciences undergraduates. Biochem. Educ. 26, 121–213. Gump, S.E., 2006. Guess who’s (not) coming to class: student attitudes as indicators of attendance. Educ. Stud. 32 (1) 39–46. Gysbers, V., Johnston, J., Hancock, D., Denyer, G., 2011. Why do students still bother coming to lectures when everything is available online? Int. J. Innov. Sci. Math. Educ. 19 (2) 20–36. Honey, P., Mumford, A., 1982. Manual of Learning Styles London. P Honey. Kelly, G.E., 2012. Lecture attendance rates at university and related factors. J. Furth. High. Educ. 36 (1) 17–40. Kember, D., Jamieson, Q.W., Pomfret, M., Wong, E., 1995. Learning approaches, study time and academic performance. High. Educ. 29 (3) 329–343. King, J., 1995. Learning styles and absenteeism: is there a connection? J. Freshm. Year Exp. 7 (1) 67–82. Kirby, A., McElroy, B., 2003. The effect of attendance on grade for first year economics students in University College Cork. Econ. Soc. Rev. 34 (3) 311–326. Kolb, D.A., 1984. Experiential Learning Experience as a Source of Learning and Development. Prentice Hall, New Jersey.

136

A. Mearman et al. / International Review of Economics Education 17 (2014) 120–136

Kottasz, R., 2005. Reasons for student non-attendance at lectures and tutorials: an analysis. Investig. Univ. Teach. Learn. 2 (2) 5–16. Lamdin, D., 1996. Evidence of students’ attendance as an independent variable in education production functions. J. Educ. Res. 89, 155–162. Laurillard, D., 1979. The processes of student learning. High. Educ. 8, 395–410. Longhurst, R., 1999. ‘Why aren’t they here?’ Student absenteeism in a further education college. J. Furth. High. Educ. 30 (6) 61–80. Marburger, D.R., 2001. Absenteeism and undergraduate exam performance. J. Econ. Educ. 32 (2) 99–109. Mattick, K., Crocker, G., Bligh, J., 2007. Medical student attendance at non-compulsory lectures. Adv. Health Sci. Educ. 12, 201– 210. Newman-Ford, L., Fitzgibbon, K., Lloyd, S., Thomas, S., 2008. A large-scale investigation into the relationship between attendance and attainment: a study using an innovative, electronic attendance monitoring system. Stud. High. Educ. 33 (6) 699–717. Paisey, C., Paisey, N.J., 2004. Student attendance in an accounting module – reasons for non-attendance and the effect on academic performance at a Scottish University. Account. Educ. 13 (Suppl. 1) 39–53. Park, K.H., Kerr, P.M., 1990. Determinants of academic performance: a multinomial logit approach. J. Econ. Educ. 21 (2) 101–111. Park, L., Guay, R.P., 2009. Personality, values, and motivation. Personal. Individ. Diff. 47, 675–684. Rau, W., Durand, A., 2000. The academic ethic and college grades: does hard work help students to ‘make the grade’? Sociol. Educ. 73 (1) 19–38. Rodgers, J.R., 2001. A panel-data study of the effect of student attendance on university performance. Aust. J. Educ. 45, 284–295. Romer, D., 1993. Do students go to class? Should they?. J. Econ. Perspect. 7 (3) 167–174. Sharif, S., Barber, J., Morris, G., Gifford, L., 2010. The relationship between learning styles, attendance and academic performance of pharmacy undergraduates. Pharm. Educ. 10 (2) 138–143. Schmulian, A., Coetzee, S., 2011. Class absenteeism: reasons for non-attendance and the effect on academic performance. Account. Res. J. 24 (2) 178–194. Self, S., 2012. Studying absenteeism in principles of macroeconomics: do attendance policies make a difference? J. Econ. Educ. 43 (3) 223–234. van Blerkom, M.L., 1992. Class attendance in undergraduate courses. J. Psychol. 126 (5) 487–494. Woodfield, R., Jessop, D., McMillan, L., 2006. Gender differences in undergraduate attendance rates. Stud. High. Educ. 31 (1) 1–22. Wyatt, G., 1992. Skipping class: an analysis of absenteeism among first-year college students. Teach. Sociol. 20 (3) 201–207.