The impact of paid work on the academic performance of students: a case study from the University of Canberra.

Daly, Anne

This article uses data collected from a survey of students at the University of Canberra to test the effects of paid employment on the average grade obtained in second semester, 2002. The results show that students who do well at school also tend to do well at university and that private study improves grades. Missing classes had a negative effect on grades. Paid employment did not have a large effect on grades. Our results show that some paid employment improves grades slightly, but working more than twenty-two hours per week has a negative effect.

Key Words

grades (scholastic)

university attendance

employment

part time employment

study

academic achievement

**********

Growing numbers of university teachers have expressed concern that students are becoming disengaged from their university experience because of time commitments in non-academic activities. More students are engaged in paid employment and the increasing proportion of mature-age students are likely to have family commitments. McInnis, James and Hartley (2000), in their surveys of first-year students in seven Australian universities, found that between 1994 and 1999 the percentage of full-time students in paid employment had grown from forty-two per cent to fifty-one per cent. By 2001, this share had risen to seventy-three per cent (McInnis & Hartley, 2002). Among those working in 1999, over half worked for eleven or more hours per week, compared with forty per cent in 1994.

The issue of the effect of paid work on university performance is not restricted to Australia. It has received attention in the UK literature (Metcalf, 2003; Winn, 2002; Hunt, Lincoln & Walker 2004). The British literature has focused on the equity implications of paid employment, arguing that those students from low income backgrounds are likely to be disadvantaged educationally by their need to engage in paid employment (see, for example, Metcalf, 2003: Hunt, Lincoln & Walker, 2004).

This article draws upon student data from the University of Canberra to address the issue of the extent to which engagement by full-time students in paid employment during the semester has an adverse impact on academic performance. An innovation of the study has been to combine the results of a survey with student administrative data. The university had 9,271 students enrolled in Semester 2, 2002, 6970 of these were undergraduates. More than half the students were female and two-thirds were under twenty-five years of age.

Various detailed studies of undergraduate students have suggested negative implications of non-academic activities on the university experience but this does not seem to be translated into lower grades. McInnis (2001) summarised the results of the first-year experience surveys conducted by him and his colleagues:

 Our findings suggest that compared with those who do not work,
 younger first year students who work part-time are more likely to
 spend fewer days on campus, to not work with other students on areas
 of their course, and to have studied inconsistently through the
 semester. They also tend to anticipate getting lower marks, and are
 more likely to seriously consider deferring at an early point of
 their student experience ... We "also know that these negative
 factors are amplified the more hours students work, and they feel
 seriously burdened by overcommitment. (p. 5).

He emphasised, however, that these negative implications of paid employment must be set against the positive benefits of part-time work in terms of promoting organisational skills and exposing students to new situations.

This issue was further explored by McInnis and Hartley (2002) in a survey of 1,563 working students who were enrolled full-time in nine Australian universities in 2001. This survey was directed at post-first-year students. In seven of the universities it was conducted by campus interviews and in two by a mailed questionnaire. They also conducted thirty phone interviews.

The survey found that seventy-eight per cent of students had worked in the past year and seventy-three per cent during semester time. On average, students worked fifteen hours per week, with more than forty per cent of them working more than sixteen hours per week (see also Long & Hayden, 2001).This compared with an average of five hours per week in 1984. McInnis and Hartley (2002) used regression techniques to estimate the impact of employment on average grades. They found that students' entry score, being a delayed or mature-age student, study motivation and academic commitment all had a positive effect on grades. Negative impacts were found for study and work conflict and having more than twenty-one hours of class contact per week. (1) No significant effect was found for hours in paid employment. However, in a separate regression, a significant negative effect of hours of employment on average grades was identified for students entering university directly from school.

In another Australian study, conducted in 1998 at the University of Sydney, Jarkey, Noble and Dalziel (2000) surveyed 300 Asian language students. They considered the influence of non-academic commitments (paid employment, family responsibilities and voluntary work) on academic results and found no statistically significant correlation.

There have been a number of studies in the UK investigating the implications of the increasing rate of employment among undergraduates for their experiences at university. In a study of 2,054 full-time and 747 part-time students in 1998-99, Callender and Kemp (2000) found that sixty per cent of full-time students were working during the academic year. Although the study did not directly examine the impacts on academic performance, the results indicate that financial difficulties hindered some students from fully participating in university life.

Metcalf's (2003) study of 782 students in four universities showed similar results, but she did not test for the impact on final grades. She found that those in paid employment during term-time came disproportionately from disadvantaged backgrounds. Other results for the UK draw a similar conclusion to those reported for Australia. Paid employment may detract from the university experience but it does not seem to have a substantial negative effect on results (Winn, 2002).

A recent article by Hunt, Lincoln and Walker (2004), however, reported a negative effect on academic attainment. They surveyed full-time undergraduates at Northumbria University over a three-year period, 1999 to 2001. On the basis of a total of 2,737 responses, they were able to divide their sample into seven subject groups and compared results for those in paid employment with those who were not. They found that paid employment had a significant negative effect on academic performance of between 4.9 and 0.7 percentage points. However, they did not attempt to control for other factors likely to influence performance, so this result does not hold other things equal. On the basis of this result, it cannot be concluded that paid employment was the cause of the poorer academic performance of the students in this sample.

In summary, the literature in general has not found a significant negative effect of paid employment on students' results. There are some groups--those entering university directly from school and those working long hours as well as studying full-time--whose academic grades might be negatively affected by paid employment. This may appear a surprising finding to many readers, but it probably reflects the success with which students are able to balance the competing demands on their time.

In the economic models of time allocation developed by Becker (1974) and others, students can be thought of as making rational optimising decisions about the allocation of their time subject to the constraints they are facing. Achieving a particular grade in a subject can be thought of as a production decision. Students bring to their studies a certain level of motivation and natural ability. This can be used in combination with time spent in classes and individual study to produce certain grade outcomes. Other factors such as the need to work and generate

income, time spent caring for dependents or training for an elite sport will also affect the level of grades students aim for or can achieve.

One of the largest costs involved in undertaking higher education is the opportunity cost of foregone earnings. By engaging in employment while studying, students can reduce these costs substantially but they will only be willing to engage in paid employment up to the point where the benefits of a higher income are equal to the costs of acquiring that income. The costs will include any detrimental effect on grades.

Paid employment while studying has a significant effect on the calculated private rate of return to higher education, as up-front costs incurred while studying, given appropriate discounting for the future benefits, have a substantial effect on the calculated rate of return. For example, Daly, Fleming and Lewis (2003) estimated the private rate of return to a university degree compared to completion of high school in Australia for males in 2001 to be around fifteen per cent. However, if students earned $A11,466 per year (the average hourly wage reported for students in McInnis and Hartley (2002), multiplied by the average weekly hours of paid employment, multiplied by fifty-two), the private rate of return rose to twenty-three per cent (see also Lewis, Daly & Fleming, 2004). If students can work part-time without affecting grades and presumably future employment and income opportunities, working is a rational choice.

Methodology

The results reported in this article are based on a combination of two data sets; students' answers to a questionnaire relating to their experiences in second semester 2002 and student records held by Student Administration at the University of Canberra for the same semester. We chose a completed semester so student responses would not be affected by answering questions relating to a semester still in progress and the administrative data on grades were finalised. The focus of the survey was on student time-use, their paid employment record and its impact on their studies. We were also interested in suggestions of how the university could better accommodate the needs of working students. The survey was approved by the University of Canberra Ethics Committee on Research into Human Subjects. We piloted the questionnaire with some students and marketing research staff.

The use of student administrative data had two advantages: it reduced our dependence on subject recall for some of the key variables such as university entry score (UAI) and grades and it reduced the time required to fill in the questionnaire to about ten minutes.

In order to access the students' records we were required to have informed consent from each student. Forty-six students (ten per cent of respondents) completed the questionnaire but were unwilling to give us the consent to access their files. We have been able to use their responses for some of the descriptive results, but not in the regression analysis.

We administered the surveys on campus to large second- and third-year classes across the three divisions at the university in April, 2003, because we decided that this was a better option than mailing questionnaires for achieving a higher response rate. (2) The disadvantage of this methodology is that students involved in long hours of non-academic activity may be less likely to be attending lectures and therefore to be in our sample. However, these students may well have been too busy to complete a mail questionnaire.

We supplemented our lecture-based administration of the questionnaire with a mailed questionnaire to students who had received a faculty warning, that is, they had failed half of the subjects they were enrolled in for Semester 2, 2002. The response rate from this mail-out was a very low seven per cent. We also had questionnaires available for completion at the Student Association office. We provided Student Administration with a list of names and identification numbers for students who had agreed to grant us access to their student records and we combined the two data sets for the analysis presented below.

In order to quantify the recorded grades, marks were allocated at the midpoint of the range according to the following scale:

NN--a fail without submitting any assessment items--0

NX, NC, NS--a fail after completing some or all of the assessment items--35

PX--provisional pass--45

P--pass--55

CR--credit--70

DI--distinction--80

HD--higher distinction--92.5

The results

Table 1 presents some summary statistics for our sample. Column 1 presents the results for the whole sample for whom we have data and column 2 presents results for those full-time students who are included in the regression results discussed below. Our sample contained a higher proportion of females than the undergraduate population of the university as a whole. The age distribution of respondents, however, was close to that of the university population with some over-representation of the twenty to twenty-four age group (60.5 per cent of the sample compared with fifty-four per cent of the population) and under-representation of the twenty-five to thirty-four age group (eighteen per cent of the sample compared with twenty-three per cent of the population).

The concentration of students in the under-twenty-five age group was associated with a minority having responsibility for dependents. Over half of the dependents were children under the age of twelve, and there were only a few respondents who were responsible for the care of aged people. Three per cent of respondents were registered with the university Disability Office, meaning that they had supplied the necessary medical or professional documentation of their condition required to be eligible for special assistance. The University of Canberra is located very close to the Australian Institute of Sport (AIS), which has been established for the development of elite athletes. There are close links between the university and the AIS, and this is reflected in the significant group of students who were training for state or national teams while studying.

The average student in our sample spent 12.7 hours per week in class contact, 11.5 hours per week in extra study outside the direct class contact hours and 20.9 hours in paid employment .This adds up to 45.1 hours per week in these activities, in excess of a standard full-time worker's load. The share of the full sample in paid employment was slightly higher than the national average from the earlier study by McInnis and Hartley (seventy-seven per cent compared with seventy-three per cent). Ninety-five per cent of respondents who worked said the primary motivation for working was to earn an income, with only a small group working to gain work experience or as part of their university study. Employment was an important source of income for these students, accounting for about sixty per cent of total income. (3)

Paid employment was a regular feature of these students' lives. Ninety-two per cent of those working worked either all or most weeks of the semester. During the mid-semester teaching break, almost half (forty-nine per cent) increased the number of hours they worked. A similar share of students worked during the long summer break as during term, and on average they increased the number of hours they worked to 30.8 hours per week.

Although not enshrined in university policy, there is an expectation that students will spend about twelve hours per week on each subject they study, including class contact time and private study. (4) The results of this survey suggest that these expectations are not being met. The average for the students in our sample who had a full-time load was 8.2 hours per week per subject, falling well short of the expected twelve hours. On average, respondents to the survey reported they missed about nine per cent of classes which is just over one full week of classes for the semester. Casual empiricism and discussions with the lecturers suggests that this is probably an underestimate of classes missed.

All University of Canberra students have access to computing facilities and the Internet on campus and eighty-two per cent of our respondents had access at home. Access to computing facilities does not appear to be a major problem for our students. However, access to academic staff and to the library was more difficult for the students, given their other commitments (see Table 2).

When asked what the university could do to most help students combine their studies with other commitments, almost half the responses asked for more material available on the Internet (see Table 3). Another significant group wanted more access to academic staff. This suggests that the availability of material on the Internet was not seen as a substitute for face-to-face contact but rather as a complement. Among this group of undergraduates, there was less enthusiasm for evening and weekend classes.

Table 4 reports the simple correlation coefficients between hours of paid employment and selected variables. The results show that for the full sample, more hours in paid employment were associated with fewer contact hours, more classes missed and a perception of a more negative effect of employment on grades. On the whole, respondents did not think that their other commitments had a substantial negative impact on their semester grades, but twelve per cent of them felt they had. For the full-time sample, there was no strong correlation between hours of employment and contact hours, but there was a positive correlation with the percentage of classes missed and a perception of a more negative effect of employment on grades.

The determinants of grades

If students make a simultaneous choice of their preferred average grade and hours of work, then the hours of work become an endogenous variable and Ordinary Least Squares regression is no longer the appropriate estimating technique.

We have tested for endogeneity using the Hausmann test and our results accept the null hypothesis that hours of paid employment can be thought of as exogenous. We have estimated regression results for those students who were studying a full-time load, that is, more than two subjects in Semester 2, 2002. (5) Our results are reported in Table 5. We have regressed the average mark per subject on the following explanatory variables; university entrance score (UAI), hours per week of paid employment, the percentage of classes missed, hours per week of private study per subject, whether the person was a primary carer of dependents, sex and registration with the Disabilities Office. The other variables are entered as controls to isolate the effect of paid employment from other important determinants of grades.

Our results show that the average mark per subject is positively related to UAI; the higher a student's university admission score, the higher was their average grade per subject (see also, McInnis & Hartley, 2002). Students also benefited from additional private study; the more hours of study per subject, holding everything else constant, the higher the average grade. However, the higher the proportion of classes missed, the lower the average grade.

The variable of particular interest in this study is the effect of hours of paid employment on the average grade. An initial plotting of the data showed the relationship between hours of paid work and grades was non-linear. We tried several functional forms and found the quadratic form (the addition of a squared term) as illustrated in Figure 1, appeared to most accurately describe the data. The first of the paid employment terms is not quite statistically significant at the five per cent level, but the squared term is.

Figure 1 shows the combined effect of these two variables on average marks out of 100. For these full-time students, the effect of hours of work varies with the numbers of hours involved, holding all other variables constant. Working up to about eleven hours per week improves marks marginally, an estimated average of 1.75 marks. However, the beneficial effects of paid employment appear to decline after eleven hours of work and the effect is estimated to become negative after twenty-two hours of paid employment. According to these results, students working thirty hours per week could expect a reduction in their average grade by 3.5 marks. We did not find a statistically significant effect on grades of being a primary carer with dependents, being female or being registered with the Disability Office.

[FIGURE 1 OMITTED]

Conclusion

This article reports some results of the effects of paid employment on average grades from a survey of students at the University of Canberra. Our results show that for students in this cross-section, those who do well at school also do well at university, and that additional study time also contributes to higher grades. The results do not show a large negative effect of paid employment on average grades. Doing some paid employment actually helps grades, perhaps by encouraging good time-management skills, but paid employment for long hours per week has a small but negative effect on average marks for a full-time student. These results are therefore in accordance with other results in the literature discussed earlier, showing that paid employment does not have a substantial effect on academic grades except for particular groups, including those who work long hours or who have come to university directly from school (McInnis & Hartley, 2002).

Evidence from other studies suggests that students are accommodating the extra time spent in paid employment by reducing the amount of leisure time. The survey shows that other commitments have some negative impact on the ability of students to access the library and academic staff members. While students may be able to manage their time to ensure there is no large negative effect of paid employment on grades, there may be other significant implications. Students may now gain less from their university experience than students in the past, and the stress involved in time-management may reduce the general levels of satisfaction with their lives.

We have focused this research on the effect of paid employment on grades, but there are a number of other important questions that require further research, particularly arising from recent changes in higher education policy favouring a stronger user pays principle. These include the effect of paid employment on the whole university experience for students, the issue of student poverty, the implications of rising costs for the type of students who are able to attend university, and the long-term effects of higher university charges for numbers and the remuneration of graduates. We have not considered any long-term effects of paid employment on the quality of the educational experience for university students. Our results are consistent with other studies of the relationship between paid employment and university grades, and suggest that a negative effect is not evident unless students are working more than twenty hours per week during term time.

Acknowledgements

The research reported in this article was approved by the Human Ethics Committee at the University of Canberra. We also discussed the project with the Student Association and Student Administration at the university. We would like to thank Tim Bradley, Mandy Yap and especially Rebecca Cassells for their excellent research assistance. We would also like to thank Diane Adams, Paula Higgins, Coralie McCormack, David Sneddon, Gerald Tart-ant, Adam Verwey, Margaret Wallace and an anonymous referee for their comments and assistance on the project. The article has benefited from comments following presentations at the University of Canberra, a conference on teaching economics in Auckland, New Zealand, the annual Conference of Economists held in Canberra and the National Institute of Economic and Social Research in London.

References

Becker, G. (1974). Human capital: A theoretical and empirical analysis (2nd ed.). New York: NBER.

Callendar, C., & Kemp, M. (2000). Changing student finances: Income, expenditure and the take-up of student loans among full- and part-time higher education students in 1998/99, (Research report RR213). London: Department of Education and Employment.

Daly, A., Fleming D., & Lewis, P. (2003). Investing in a legal education: The private rate of return to a law degree (CLMR Discussion Paper Series 03/1). Crawley, Western Australia: The Centre for Labour Market Research and Division of Business, Law and Information Sciences, University of Canberra.

Hunt, A., Lincoln, I., & Walker, A. (2004). 'Term-time employment and academic attainment: Evidence from a large-scale survey of undergraduates at Northumbria University', Journal of Further and Higher Education, 28(1), 3-18.

Jarkey, N., & Noble, C., & Dalziel, J. (2000). 'Too busy to learn?' In R. Pesman (November), Synergy, Sydney: Institute for Teaching and Learning, University of Sydney (No. 14, pp. 2-4).

Lewis, P., Daly, A., & Fleming, D. (2004). 'Why study economics? The private rate of return to an economics degree', Economic Papers, 23(3), 234-43.

Long, M., & Hayden, M (2001). Paying their way: A survey of Australian undergraduate university student finances, 2000, Australia: Australian Vice-Chancellors Committee.

McInnis, C., James, R., & Hartley, R. (2000). Trends in the first year experience. Canberra: Department of Education, Training and Youth Affairs.

McInnis, C. (2001). Signs of disengagement? The changing undergraduate experience in Australian universities. Inaugural professorial lecture presented at the Centre for the Study of Higher Education, Faculty of Education, University of Melbourne, Melbourne, Vic.

McInnis, C. & Hartley, R. (2002). Managing study and work (Commonwealth Department of Education, Science and Training Report 02/6). Canberra: Commonwealth of Australia.

Metcalf, H. (2003). 'Increasing inequality in higher education: the role of term-time working'. Oxford Review of Education, 29(3), 315-329.

Winn, S. (2002). 'Student motivation: a socio-economic perspective'. Studies in Higher Education, 27(4), 445-457.

Craig Applegate

Anne Daly

University of Canberra

Notes

(1) Delayed entry students were those who entered university between the ages of twenty and twenty-four years, mature-age students entered aged twenty-five years and over. 'Academic commitment' was calculated on the basis of responses to four statements about desire to succeed and relationships with fellow students and staff. 'Study motivation' was calculated on the basis of responses to two statements about time management skills and motivation to study. The variable 'study and work conflict' was constructed on the basis of answers to eight items designed to assess possible conflict.

(2) We received valuable advice from a number of colleagues who are experts in the design of questionnaires. The survey was piloted among a group of volunteer students at the Student Association. A copy of the survey is available from the authors on request.

(3) The students were asked to estimate the share of income from various sources but were not given explicit instructions about how to account for income in kind. This is expected to be an important source of income for those students living with their parents.

(4) The University Handbook for 2003 states 'One credit point represents an average workload of four hours per week during the semester including class contact and time spent on other study associated with the subject' (p. 91). For a full-time student doing twelve credit points that implies a weekly workload of forty-eight hours. The students in our survey were doing substantially less than that.

(5) Mature-age students and overseas students are not admitted to the university on the basis of a UAI. They are therefore excluded from the sample. We experimented with predicting UAIs for this group on the basis of other characteristics but the estimates were unreliable so we have omitted them from the sample.

Craig Applegate lectures in Economics at the School of Business and Government in the Division of Business, Law & Information Sciences at the University of Canberra, University Drive, Bruce ACT 2617. Email Craig.Applegate@canberra.edu.au

Anne Daly lectures in Economics at the School of Business and Government in the Division of Business, Law & Information Sciences at the University of Canberra, University Drive, Bruce ACT 2617. Email Anne.Daly@canberra.edu.au

Table 1 Characteristics of sample, Semester 2, 2002

 Students with
 Full sample full-time load
 N=477 N=389
 (1) (2)

Variable Mean Mean

Age (years) (a) 24.7 24.6
Males (%) (a) 28 28
Females (%) (a) 72 72
Financial dependents 16.4 5.9
Primary carer 11.7 0.5
Registered with the Disability Office (%) 3 3.1
Training for state or national sports team 5.7 5.4
Contact hours/week 12.7 13.1
Number hours study/week 11.5 11.7
Access to internet at home (%) 82 83
Av. Paid hours workweek 20.9 20.7
Paid work in summer 73.2 75.3
No. hours workweek in summer job 30.8 30.5
% classes missed 9.2 9.2
Sources of income (% total)
 Employment 59.6 60.4
 Youth Allowance 14.1 13.8
 Partners 6.5 5.2
 Parents 14.9 15.6
 Student loans 0.9 1.0
 Other 3.9 3.8

Notes:

(a) These results are derived from student records and omit
those students who answered the questionnaire but did not
provide us with authority to access their student records.

Table 2 Do other commitments make it difficult for you to
access university services when you need them for your studies?
(Number of responses = 2862)

 Academic
 Computer Student staff
 labs Library Association members
 % % % %

Frequently 2.5 3.1 1.8 3.2
Sometimes 6.2 7.4 4.2 7.4
Never 6.5 5.1 9.0 4.6
Not Stated 1.5 1.1 1.8 1.4
Total 16.7 16.7 16.7 16.7

 Other
 university
 Student support
 administration services Total
 % % %

Frequently 2.5 0.5 13.5
Sometimes 5.4 0.6 31.2
Never 6.9 4.0 36.0
Not Stated 2.0 11.6 19.3
Total 16.7 16.7 100.0

Table 3 What measure could the university take that would
be most helpful in allowing you to combine your study with
other commitments? (Number of responses = 764)

 Per cent of responses (a)

More evening classes 15.7
More weekend classes 5.2
More material available on the internet 45.1
Academic staff more accessible 26.3
Administrators more accessible 7.6

Table 4 Correlation between hours of
paid employment and selected variables

 Full sample Full-time students

Contact hours -0.43 0.09
Percentage of classes missed 0.12 0.16
Effect on studies (a) -0.21 -0.38

Notes:

(a) Students were asked 'How do you think that your work has affected
your semester grades?' They were given the options of 'substantially
negative/ a little negatively/ not at all/ improved a little/
substantially improved'. These responses were coded from 1-5 so the
negative correlation suggests the more hours worked the larger the
perception of a negative effect.

Table 5 Dependent variable = average mark per subject

 Sample of students with
 a full-time load

Constant 31.4
 (4.44) ***
UAI (a) 0.41
 (5.02) **
Hours/week in paid employment 0.32
 (1.77)
Hours/week in paid employment (2) -0.015
 (-2.47) **
% Classes missed -0.21
 (4.25) **
Hours/week study/subject 1.13
 (4.98) **
Carer (b) 0.0042
 (0.02)
Sex (c) -0.73
 (-0.53)
Registered with Disability Office (d) -2.20
 (-0.65)
[R.sup.2] 0.31

Note that t statistics are in brackets. *** indicates
the coefficient was significant at the 1 per cent level.

(a) UAI--University Admissions Index

(b) Carer is a dummy that takes the value of one
for those who were primary carers of dependents.

(c) Sex is a dummy taking the value of one for females.

(d) Registered with the Disability Office is a dummy
taking the value of one for those who were registered.