Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Journal of Education and Work

ISSN: 1363-9080 (Print) 1469-9435 (Online) Journal homepage: http://www.tandfonline.com/loi/cjew20

Working while studying: the impact of term-time


employment on dropout rates

Elisabeth Hovdhaugen

To cite this article: Elisabeth Hovdhaugen (2015) Working while studying: the impact of term-
time employment on dropout rates, Journal of Education and Work, 28:6, 631-651, DOI:
10.1080/13639080.2013.869311

To link to this article: http://dx.doi.org/10.1080/13639080.2013.869311

Published online: 18 Dec 2013.

Submit your article to this journal

Article views: 320

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=cjew20

Download by: [Florida Atlantic University] Date: 07 March 2016, At: 12:30
Journal of Education and Work, 2015
Vol. 28, No. 6, 631–651, http://dx.doi.org/10.1080/13639080.2013.869311

Working while studying: the impact of term-time employment on


dropout rates
Elisabeth Hovdhaugen*

Research and Education, NIFU – Nordic Institute for Studies in Innovation, Oslo,
Norway
(Received 20 January 2012; final version received 12 November 2013)
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

There are many possible reasons why students leave university prior to
degree completion, and one of the more commonly cited is being
employed while studying. This paper analyses the impact of employ-
ment status on dropout rates using survival analysis. It finds that
employment status does have an impact on dropout rates; students who
work full time alongside studying full time are less likely to complete
their programme than students working short part-time or not working
at all. However, it seems as if there is a threshold to how much students
can work, as working more than 20 h a week (long part-time work)
increase the risk of dropout as much as full-time work. Integrating
employment status into the analysis does not change the effect of vari-
ables known to have an influence on dropout, such as grades, gender
and social background, but it contributes to further explain who are at
risk of dropout. This implies that models for dropout and retention must
also take such external factors into account, not just consider what
happens at university, as in model of student departure.
Keywords: student dropout; term-time employment; survival analysis

Introduction
There has been a considerable amount of research on the issue of term-time
employment during the last decade, covering a broad range of issues.
Several studies have looked at the extent of paid work among students,
some have looked at the reasons for working, while a few others have
investigated the relationship between working while studying and comple-
tion or dropout rates. When students are asked about reasons for leaving
higher education before completing a degree, ‘being employed’ is one of
the most commonly mentioned issues (Yorke 1999; Hovdhaugen and
Aamodt 2009). Given that the number of students working while studying
is high in most countries, it is a relevant research question to find out how

*Email: elisabeth.hovdhaugen@nifu.no
© 2013 Taylor & Francis
632 E. Hovdhaugen

much greater the risk of dropout is for working students, compared to


students who do not work.
This paper is focused on students studying full time, and their employment
patterns. Employed students are divided into three groups: those who work up
to 1–19 h a week, labelled ‘short part-time’, those who work 20–30 h a week,
labelled ‘long part-time’ and those who work more than 30 h a week, labelled
‘full-time’. Unlike other studies that assume that students who work full time
while studying are actually part-time students (Berg and Kyvik 1992; Darmody
and Smythe 2008; Beerkens, Mägi, and Lill 2011) I assume it is possible for
some students to handle both full-time studies and more than 30 h of work a
week. Earlier studies suggest that working only slightly reduces time spent
studying for Norwegian students: study time is reduced by 5–10 min for every
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

hour a student spends working (Wiers-Jenssen and Aamodt 2002;


Hovdhaugen 2004). It therefore seems likely that a moderate degree of
employment does not have a strong impact on studies, and in most cases work
interferes with students’ leisure-time, not their study-time.
Few studies have focused explicitly on the relationship between student
employment and dropout. However, a study by Roksa (2011) investigates the
relationship between employment, parental education and degree completion
at American community colleges and four year institutions. She finds that
employment up to 20 h a week facilitates completion, while students who
work more than 35 h a week are less likely to complete a degree, when con-
trolling for students’ family background and type of institution. Community
college students are more likely to work while studying, which probably
explains at least part of the difference in degree attainment between four-year
institutions and community colleges in America. Moulin et al. (2012) also
finds that the threshold for dropout due to working too many hours is
relatively high, and only has a significant effect on students who drop out,
not students who transfer to a different programme or institution.
The question addressed in this paper involves what impact university
students’ employment status has on their likelihood of dropping out. Are
there differences between groups of students with regards to employment
patterns? Is the risk of dropping out lower for students who do not work
than for students who do? While several other studies have looked at stu-
dent employment and dropout (see for e.g. Roksa 2011) these have been
undertaken in countries with tuition fees. This paper offers a new perspec-
tive on these issues by looking at working while studying in a context with-
out student tuition fees and where higher education is a low-cost
investment.

The Norwegian context


The higher education system in Norway consists of two main types of
institutions: universities and university colleges. These provide different
Journal of Education and Work 633

types of education. Universities offer bachelor’s and master’s programmes


in a range of fields and longer professional degrees such as medicine and
law. University colleges mainly provide three to four year professional
degrees, such as nursing and teaching. This study is confined to undergradu-
ate university education in humanities, social science and science pro-
grammes. These types of university programmes are known to have had
somewhat higher dropout rates than corresponding programmes at university
colleges (Aamodt 2001; Aamodt and Hovdhaugen 2011). The estimated
dropout rate in undergraduate programmes in the humanities, social sciences
and sciences is about 15% after the first year, rising to 30% after four years
(Hovdhaugen 2011).
Public higher education institutions in Norway do not charge tuition fees.
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Students in higher education get loans and grants, provided by a govern-


mental agency, Lanekassen. This form of student support is available to
most students, and about 75–80% of all students have some sort of support
from Lanekassen. However, students who earn more than a certain amount
in a year might lose their right to a grant (Opheim 2008). Hence, students
are quite well supported in Norway, and the support is intended to be suffi-
cient to make working while studying unnecessary. Despite this, a majority
of students work while studying (Lyngstad and Øyangen 1999; Ugreninov
and Vaage 2006; Arnesen et al. 2011).
In 2003, Norway implemented a higher education reform, making
degrees more in line with the Bologna system, by shortening undergraduate
university degrees from four to three years. In addition, the structure of the
university undergraduate programmes was changed, from loosely structured
programmes to more strictly structured programmes (Michelsen and Aamodt
2006). These changes, in addition to the changes in the funding structure
were intended to reduce the number of students working while studying,
but it does not seem to have done so (Opheim 2011).

How many work while studying?


It seems the extent to which students work varies between countries and
has shifted over time. A study conducted in Norway in the late 1990s esti-
mate that 56% of students worked during term-time (Lyngstad and Øyangen
1999) while studies conducted in 2005 and 2010 found that 62% of
Norwegian students worked during term-time (Ugreninov and Vaage 2006;
Arnesen et al. 2011). Norwegian students are not unique in this sense: a sur-
vey of Estonian students found that over 60% of full-time students work
(Beerkens, Mägi, and Lill 2011); and, in Ireland the level of employment
among full-time students is 61% (Darmody and Smythe 2008). A review of
American research states that 80% of students work, although there are
institutional differences in student employment (Pascarella and Terenzini
2005). Another American study found that 44% of students at four-year
634 E. Hovdhaugen

institutions work, while the corresponding figure at community colleges is


72% (Roksa 2011, 299). In the 1990s in England, only a minority of
students were working while studying, but this has increased in recent years
to about half of all students (Broadbridge and Swanson 2005; Richardson,
Evans, and Gbadamosi 2009).
There are, of course, variations in the number of hours students work on
average. Ugreninov and Vaage (2006, 67) found that Norwegian students
worked nine hours a week on average in 2005, while a 2010 study found
that full-time students work eight hours a week (Arnesen et al. 2011). Previ-
ous studies of student working hours in Norway have suggested full-time
students work approximately 10–12 h a week (Berg and Aamodt 1987;
Wiers-Jenssen and Aamodt 2002; Hovdhaugen 2004). International studies
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

often show slightly higher average weekly working hours; an Australian


study found that students work 14–15 h a week (Bradley 2006), while two
British studies on student employment reported average working hours as
between 12 and 14 h per week (Lucas and Lammont 1998, 50; Barke et al.
2000, 9). In summary, it seems that in most countries it is common for stu-
dents to work while studying, and for full-time students to work 10–15 h a
week on average.

Reasons for working


Several studies have investigated why students work while studying. In
many British and American studies ‘financial reasons’ are a common reason
to work. However, the term ‘financial reasons’ might cover a broad range
of motivations: in countries with tuition fees this can include both students
who work to sustain themselves while studying and who work to fund their
fees. Two recent British studies found that around 80% of students work to
cover living expenses (Richardson, Evans, and Gbadamosi 2009; Hall 2010)
but one of these studies also found a substantial amount of students also
work to fund their fees (Richardson, Evans, and Gbadamosi 2009). An
Australian study suggests that most students work in order to provide for
everyday needs (Devlin, James, and Grigg 2008). Most American studies
on student employment show that having to work to fund fees is more
common, due to the cost of attending higher education in the USA (Schuh
2005).
However, as the studies described illustrate, much of the research investi-
gating the role financial factors and work play in students leaving higher
education have been undertaken in countries with tuition fees. As Norway
offers higher education at public institutions free of charge, raising money
for tuition fees cannot be one of the reasons why students work. On the
other hand, the cost of living in Norway is high and this might force some
students to work, despite student loans and grants which are intended to
provide for living costs while studying. Financial motivations might also go
Journal of Education and Work 635

beyond living and studying costs, where students need or wish to increase
their standard of living while studying and so choose to work to maintain
or increase their income.
While these financial rationales are clearly central, there are other likely
reasons for students working while studying. Students may wish to get work
experience and be better prepared for working life after graduation. Wang
et al. (2010) argue that there is a positive effect of working on grade point
average for those students who say they are working to enhance their future
employability. Arnesen and Try (2001) found that Norwegian students who
worked while studying had a significantly higher chance of being employed
six months after graduation than those who did not. Students also describe
work as a social experience and a way to increase their skills (Lucas and
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Lammont 1998). Others argue that working while studying helps them build
their self-confidence, gain independence and develop their CV (Hodgson
and Spours 2001, 379). Hence, working while studying can become a way
for students to ‘differentiate themselves from other students’ that have not
had work experience (Broadbridge and Swanson 2005, 237) and thereby
enhance their employability.
Students who work to improve their future employability or to maintain
a certain standard of living can be seen as having chosen to work, while
those that work to fund tuition fees or living expenses can be seen as being
forced to work. This is similar to the dichotomy Richardson, Evans, and
Gbadamosi (2009) identified in a factor analysis that divided part-time
working students into two groups: those focusing on the benefits of working
part-time and who stated a desire to enhance their employability (who the
researchers call ‘aspirational’) and those students focusing on the direct sur-
vival aspects of working part-time while studying (who were described as
‘functional’).

The impact of working on study progression


Previous research on the impact of work on study progression is inconclu-
sive. There is research indicating that employment can have negative conse-
quences for study progression, but also studies showing no significant effect.
The negative effects are related to grades, study delays and increase dropout
risk. Students who work might get lower grades than they would have if they
had concentrated solely on their studies; Humphrey (2006) found that stu-
dents with part-time jobs receive lower grades, on average, than students who
do not work. Other studies have found that the concern that work might inter-
fere with studies or reduce grades is the most commonly given reason for not
working (Lucas and Lammont 1998, 50; Barke et al. 2000, 16). In addition,
it seems that working can delay studies, either because students cannot study
full-time and thus complete fewer courses, or because too much time working
causes them to fail an exam or course. Hence, students who are employed
636 E. Hovdhaugen

while studying full-time might well be at greater risk of dropping out of their
studies. Yorke (1999) found that financial reasons and constraints due to
combining studying with part-time work are common denominators
among students who leave higher education prior to degree completion. In a
re-analysis of several studies on non-completion in England, Yorke and
Longden (2004) found financial reasons to be among the most common given
by students for drop out, along with choosing the wrong programme or insti-
tution, and academic difficulties.
Several studies have investigated the effect of working on academic
performance. Humphrey (2006) found that term-time working does reduce
end-of-year grade averages. A Chinese study found no general effect of
part-time work on grade point average, but did find that extensive hours of
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

working reduced students’ grade point average (Wang et al. 2010). Bradley
(2006) found no significant relationship between term-time employment and
grade point average over the course of a year, but his study involved quite
few students, and very few of them (15%) did not work at all.

Theoretical approaches to the effect of employment on dropout


Even though employment is known to be one of the external obligations
that can be a barrier to persistence in higher education, it has not been at
the forefront of the debate in the retention literature. In the initial version of
Tinto’s (1975) longitudinal model of institutional departure and in Bean’s
(1980) model of attrition, employment is not included at all as a relevant
factor, in models of retention. Later versions of Tinto’s model indirectly
include working while studying as one of several external commitments that
students might have (Tinto 1993, 114), but it is still not at the core of the
analysis. Another model, building on a similar framework does spell out
‘work responsibilities’ as a factor to consider but only as one of over 30
variables in the model, so it is not a major focus of the theoretical frame-
work (Nora, Barlow, and Crisp 2005, 131). However, as term-time employ-
ment takes place while students are at a higher education institution, it
might well influence persistence, and might be more important for decisions
to leave than American models of retention have considered.
Since there is no established framework for interpreting the effects of
employment on dropout, other perspectives have to be used. Gambetta
(1987) uses two different perspectives on educational attainment: ‘pushed-
from-behind’ and ‘pulled-from-the-front’. These might be used to explain
why students who work a lot are more at risk of dropout than those
working more moderately. These perspectives encourage us to consider how
patterns of dropout might be related to the reasons for working: whether
students are forced to work to make ends meet or whether they choose to
work. These two types of reasons trigger different interpretations of why
students might drop out as a result of employment.
Journal of Education and Work 637

Students who work because they have to in order to provide for them-
selves might well experience a sense of being ‘pushed-from-behind’, away
from studies, where their working commitment interferes with their studies
or forces them to leave before degree completion. Students from less well-
educated families are more likely to work than students from better edu-
cated families, and this group is also more at risk of dropping out of higher
education. Devlin, James, and Grigg (2008, 120) conclude in their study on
student employment that students ‘trade off the time available for study to
meet their living costs – many find this to be an insidious equation in which
long-term benefit is sacrificed to short-term necessity’. This suggests that
students work because they need the money now, even though they are
aware this might have consequences for their grade point average or
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

chances of completion. As Gambetta (1987, 61) argues these students


choices are constrained by external factors.
The alternative perspective on work and drop out is that students are
being ‘pulled-from-the-front’, implying that students choose to work, despite
being aware of the risks for their academic performance and progression.
Students who work while studying may evaluate these costs against per-
ceived benefits of working while studying and decide it is in their interest
to work. If and when students experience larger or additional costs, for
example having to miss classes or getting a lower grade than they expect,
they can choose to reduce their working hours and increase the time they
spend studying. Signalling theory (Arrow 1973; Spence 1973) might be an
alternative way of interpreting students that choose to work while studying,
especially mature students who had a job before they started working. By
taking more education, these students signal effort and talent to potential
employers and expect to increase their employability by getting more
educational credentials.
Given that public higher education in Norway is tuition free and students
can get public support in the form of grants and loans we can assume that
most Norwegian students work because they choose to. However, a com-
mon sense assumption could be that few choose to study full-time alongside
working full-time, and that these students more often feel forced to work
while studying.

Data and methods


In this study, register data from National educational data base administered
by Statistics Norway have been used to track students’ movements in and
out of, and through the higher education system1. The data consist of one
cohort: all students who were enrolled in full-time undergraduate education
in humanities, social sciences and sciences at university, for the first time,
in the autumn semester of 2003. The data-set was in the form of a
638 E. Hovdhaugen

person-year file, making it possible to use survival analysis, also known as


event history analysis, and the analyses were done using SPSS.
The data contain information on whether students are still studying, have
graduated or if they have left higher education, either temporarily or perma-
nently. Students who are absent from higher education for a single year are
defined as being on a break, while absences of two consecutive years are
defined as dropping out. One reason for allowing a one year study break,
and not counting all students leaving higher education as ‘dropouts’, is that
we know it is quite common in Norway for students to take study breaks
(Berg 1997; Hovdhaugen and Aamodt 2005). In the USA, students who
take breaks from study are commonly described as ‘stop-outs’ (Tinto 1993).
Survival analysis can be used when one is concerned with patterns of
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

occurrence of specific events (Allison 1984; Yamaguchi 1991). One of the


method’s central ideas is to start by looking at all students, and for every
observed year students either keep studying, drop out of higher education or
complete their studies. Those who complete their studies are taken out of
the data, so the focus is on those at risk of dropout in any given year. In
addition to removing students who complete their studies from the data-set
the year they graduate, students who drop out are taken out of the data-set
the year after dropping out, since they are no longer included in the remain-
ing group at risk. The students are observed for three consecutive years,
and the number of students at risk of dropout in each year in the analysis is
presented in Table 1. The number at risk every year is decreasing over time,
due to individuals completing and dropping out who are successively taken
out of the data.
The method of analysis used is discrete-time survival analysis, a
common form of analysis for educational careers (Singer and Willett 2003).
This implies that the students’ status (still studying or having dropped out)
is measured once a year. The data were analysed using binary logistic
regression. The dependent variable in the multivariate analysis is dropout,
as opposed to remaining a student or having a study break.

Independent variables
In survival analysis, it is common to make a distinction between explana-
tory or independent variables that are fixed over time and those that are

Table 1. Number of cases for each years in the analysis.


N %
First year 5200 40.9
Second year 4235 33.3
Third year 3291 25.9
Total number of cases 12,726 100.0
Journal of Education and Work 639

time-varying (Jenkins 2004; Mills 2011). In this analysis all the independent
variables, apart from employment status, are constant over time. The back-
ground variables in the regression analysis are study year, employment sta-
tus, field of study, parents’ education, age, grades from upper secondary
school, gender, geographical background (if the student comes from the uni-
versity region) and if the student had a child when starting education. The
frequency of all the independent variables, apart from study year is pre-
sented in Table A1–A3 in the Appendix 1.
Employment status is a time-varying variable in the analysis, as students
can work part-time one year, not work the year after and then work part-
time again in their third study year. This variation is controlled for in the
multivariate analyses. The variable is based on information from the regis-
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

ter-based employment statistics (Statistics Norway) which contains informa-


tion on whether the individual has an employment contract or not. The
register is updated every time employment contracts change, and the data
used in this analysis have been retrieved in November each year. The num-
ber of hours worked per week by those with an employment contract is reg-
istered in three bands; 1–19 h per week, 20–30 h per week and more than
30 h per week. Hence, this is not a continuous variable, which might have
been preferable, but the only available data are on grouped categories for
those holding a job.
A full-time position in Norway is usually 37.5 h a week, but as Statistics
Norway only record if contracts exceed 30 h a week, people holding such
contracts are here considered as having a full-time job, while everybody on
a contract with 30 h or less is defined as working part-time. Persons with no
record in the employment statistics have been defined as not working, as
the statistics only register those that have a job.
In this study, employment status at a given point in time, in late autumn,
is registered, and the variable is therefore lagged in the analysis. This means
that information on the students’ employment status is taken from the previ-
ous year in the analysis, in order to deal with the causal ordering problem
(Box-Steffensmeier and Jones 2004). Lagging the variable employment
status in this way makes sure ‘that changes in the time-dependent variable
precede the actual event’ (Mills 2011, 110) taking account of the problem
of simultaneous cause and effect in the analysis. This is also based on the
assumption that the number of working hours registered in November gives
a reasonable indication of that students’ labour market engagement for the
entire year. This is of course a registration at one point in time, and
students’ working hours during a semester might vary. However, this regis-
tration of working hours might also be viewed as a random sampling of
varying employment contracts, giving a reasonable indication how much
time that student spend working while studying.
As noted above, the other independent variables do not vary over time.
Grades from upper secondary education are reported as admission points for
640 E. Hovdhaugen

higher education. Grades in Norwegian upper secondary school range from


1 (lowest) to 6 (highest) and the mean in this cohort is slightly above 4.1.
However, when used as admission points the mean is multiplied by 10. In
the analysis, this variable is centred on the mean and the effect for a one
point increase is reported on. The variable field of study has three different
values: Humanities, Social sciences and Sciences. Parents’ level of educa-
tion is used as an indicator of social background, and information about the
parent with the higher level of education is used in the analysis. This vari-
able has four values: compulsory education only; completed upper second-
ary education; higher education up to four years’ duration; and higher
education of more than four years’ duration.
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Students’ pattern of working


A good place to start to present the results from this data is a description of
what students’ pattern of working looks like in Norway. In this analysis, as
in the later analyses, it is important to study only those that are still active
students, with those that complete or drop out removed from the data-set
for the following years. Figure 1 shows the proportion of students not work-
ing, working short part-time, long part-time and full-time, during the first
four years of studies, among active students.
The majority of students work short part-time while studying, and the
proportion of students working increases over time in higher education,
from about 60% working in the starting semester up to about 70% in the
autumn of their fourth year. The increase in percentage working is due to a

Figure 1. Proportion of students not working, working part-time and working


full-time.
Journal of Education and Work 641

slight increase over time in all categories of employed students. The propor-
tion of students working full-time, approximately 1 out of 10, is relatively
stable over time. This group might include a small number of students who
are supported (paid) by an employer while taking a full-time study, but who
are not expected to work full-time.
The general level of employment among students found in this study is
comparable to that found previously in studies conducted in Norway and in
other countries (see for e.g. Ugreninov and Vaage 2006; Darmody and
Smythe 2008; Arnesen et al. 2011; Beerkens, Mägi, and Lill 2011; Roksa
2011). Analyses also indicate a relatively high level of stability in employ-
ment patterns year-to-year among students, apart for those few who are
working full-time while studying full-time. More than 60% of those not
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

working in the first year of study do not work the second year either, and
70–75% of those working short part-time keep working part-time the
following year. In contrast, only 40% of those working full time while
studying keep doing so from one year to the next.

Who works while studying?


Another important question is if there are differences between groups of stu-
dents in patterns of working. To investigate this, an analysis of employment
pattern in the first year was conducted for different groups, as presented in
Table 2. The table summarises the analysis for: gender, age (in groups),
grades (in groups), parents’ educational level and field of study.
There are significant differences between groups; men, students older
than 26 and science students show patterns of working that are distinctly
different from the average pattern observed in Figure 1. Female students are
overall more likely to be employed than male, a finding which corresponds
to international research findings (Humphrey 2006; Darmody and Smythe
2008). The majority of students aged 26 years or older when they start a
degree do not work while studying, but these older students do also work
full-time while studying more often than the average student. However,
students who are 21–25 years when starting a degree are the age group
most likely to work full-time. The overall working patterns of students show
only small differences according to grades and parents’ educational level.
However, a significantly higher percentage of students with below average
grades do not work while studying, and a significantly lower proportion of
the most able students work full-time while studying. Among students from
well-educated families (with at least one parent with a higher education
degree that exceeds four years) there is a small, but significantly higher,
proportion of non-working students. There is also a higher proportion of
science students who do not work while studying compared to students in
the humanities and social sciences.
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

642

Table 2. Bivariate analysis of employment status, by background variables in the first study year.
E. Hovdhaugen

Not working Short part-time Long part-time Full-time work N (= 100%) χ2-test
Women 34.3% 54.4% 4.6% 6.7% 3005 Sig 0.000
Men 46.1% 36.9% 4.5% 12.5% 1655
Age 20 or younger 38.4% 52.7% 3.6% 5.3% 3308 Sig 0.000
Age 21–25 years old 36.0% 39.6% 6.8% 17.6% 1151
Age 26 or older 54.7% 22.9% 7.5% 14.9% 201
Grades under average 41.8% 41.8% 5.5% 10.9% 1765 Sig 0.000
Average grades 36.2% 50.3% 4.4% 9.1% 1339
Grades over average 36.7% 53.6% 3.6% 6.1% 1556
Parent’s education comp/upper sec edu 37.3% 48.4% 5.0% 9.3% 1922 Sig 0.000
Parent’s education HE – 4 years 38.3% 48.4% 4.5% 8.8% 1776
Parent’s education HE 4 + years 41.3% 47.3% 3.7% 7.7% 962
Humanities 37.5% 47.3% 4.2% 10.9% 1370 Sig 0.000
Social sciences 37.7% 50.4% 4.5% 7.5% 2532
Sciences 43.0% 42.5% 5.3% 9.2% 758
Journal of Education and Work 643

Does employment status have an effect on dropout rates?


The first multivariate analysis only includes year of study and employment
status as independent variables. A version with an interaction between the
two variables was tested, in order to explore how the effect of employment
status on dropout varies over the years, but it proved insignificant and there-
fore is not presented. However, the explanatory power of this analysis is
quite low; the calculated explained variance (Nagelkerke) is only 3.3%. The
analysis is displayed in Table 3.
The logistic regression in Table 3 shows that the risk of dropping out is
lower in the second and third year of studies, compared to the first year.
Full-time work increases the risk of dropout the most, but there is also a
significant effect of working 20–30 h a week (long part-time) on risk of
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

dropout. Compared to students not working, there is no significant effect of


working short part-time on risk of dropout, when other background vari-
ables are not controlled for. Hence, students who work up to 19 h a week
(short part-time) do not appear to face a significantly greater risk of
dropping out than students who do not work. An estimation of the dropout
probability is displayed in Figure 2.
There is no significant difference in the risk of dropout between students
who do not work at all and students who work short part-time, while
students who work long part-time face a higher risk of dropping out, but
not as high as students who work full-time. The probability that a first-year
student who works full-time will drop out is one in five; the corresponding
risk for first year students who work short part-time or do not work is about
1 in 10. The figure also underscores the result that the risk of dropping out

Table 3. Binary logistic regression on probability for dropout by years of study


and employment status.
B S.E.
**
Employment status
Work part-time −0.029 0.071
Work long part-time 0.552** 0.125
Work full-time 0.766** 0.095
**
Years of studying
Two years −0.704** 0.076
Three years −0.542** 0.079
Constant −2.084** 0.060
N 12,726
2 Log likelihood 7577.118
Nagelkerke R2 0.033
Note: Variables in italics are entered as categorical, to test the whole variable. Stars indicate
if the variable as a whole is significant.
Constant: Student in the first year who does not work.
*
p < 0.05. **p < 0.01.
644 E. Hovdhaugen
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Figure 2. Estimated probability of dropout, by years of study and employment


status.

is higher during the first year than later on in a student’s studies; this
finding is generally well supported in previous research (cf. Egge 1992;
Aamodt 2001; Hovdhaugen 2009; for international studies, see Astin 1975;
Pascarella and Terenzini 1991, 2005; Tinto 1993; Yorke 1999; Braxton
2000).
The second analysis, which also includes student background characteris-
tics and field of study, in addition to year and employment status, is
presented below (Table 4). The number of cases included in the analysis is
slightly lower in this regression than in the previous ones, as we have
included grades, a variable with some missing values. The explanatory
power of this model is somewhat increased, to 10.1%, compared to the pre-
vious model2 (Table 3).
The effect of employment status on drop out only changes marginally,
when controlling for background characteristics and field of study. It still
seems as if there is a threshold, that working up to 19 h a week has no
significant effect on the risk of dropout, but working more than 20 h a week
significantly increase the risk of dropping out. There is no significant
difference in risk of dropout between students who work long part-time and
students who work full-time. The relationship is visualised in Figure 3.
Based on figure 3, students can be divided into two risk groups; the risk
of dropping out for those who do not work or work up to 19 h a week is
6–7%, while the risk of dropping out for students who are working more
than 20 h a week is 10–11%. Hence, when controlling for other variables,
working long part-time increases the risk of dropout as much as working
full-time while studying does.
Most of the other independent variables also prove significant for drop out
patterns. Men have a higher risk of dropping out than women, and there is a
Journal of Education and Work 645

Table 4. Binary logistic regression on probability for dropout by employment


status, years of study demographic variables and grades.
B S.E.
**
Employment status
Work short part-time 0.157 0.081
Work long part-time 0.617** 0.141
Work full-time 0.569** 0.113
**
Years of study
Two years −0.619** 0.084
Three years −0.418** 0.087
Gender (female = 0) 0.336** 0.075
Age (19 = 0) 0.062** 0.011
Recruited from university region 0.145* 0.071
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Had a child when starting studies 0.166 0.215


Grades (centred around mean) −0.070** 0.007
Grades*grades 0.002* 0.001
**
Parent’s education
Compulsory education 0.506** 0.115
Upper secondary education 0.284* 0.113
HE up to 4 years 0.056 0.108
**
Field of study
Social science −0.265** 0.077
Science −0.614** 0.117
Constant −2.754** 0.128
N 11,528
2 Log likelihood 6068.840
Nagelkerke R2 0.102
Note: Variables in italics are entered as categorical, to test the whole variable. Stars indicate
if the variable as a whole is significant.
Constant: Studepnt in the first year, not working, female, 19 years old, average grades, not
from the university region, parent’s highest education is HE more than four years, field of
study is Humanities.
*
p < 0.05. **p < 0.01.

linear relationship between parents’ educational level and dropout (the more
education parents have, the less risk there is of dropout). However, there is no
significant difference observed between students who have parents with long
or short higher education, there is only a difference between students with
parents that have higher education of some sort and those that do not (only
compulsory education or upper secondary education). Academic achievement
is also significant: for every grade point increase from the mean, the risk of
dropping out is slightly weakened, but the relationship is curve-linear.
Increasing age when starting education also increases the risk of dropout.
Overall, students in science programmes are least at risk of dropping out,
compared to students in the other two fields of study, although the risk of
dropping out is lower for social science students than humanities students.
There is no significant effect of having a child when starting education. The
646 E. Hovdhaugen
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Figure 3. Estimated probability of dropout by years of study and employment


status, controlled for background variables and discipline.

findings for age, gender, grades and parents’ education corresponds to


previous findings (Tinto 1993; Yorke 1999; Hovdhaugen 2009).

Discussion
This study reiterates that it is now common for students to be working
while studying, but finds that most students work part-time. In total,
60–70% of students hold a job while studying, but only 10% of all students
hold a full-time job. These findings correspond well to earlier findings in
Norway and elsewhere. Hence, even though Norway has no tuition fees in
public higher education, and generous governmental student support
systems, students still work while studying to a similar extent as students in
countries with tuition fees. We can assume that the majority of Norwegian
students work because they want to, and that only a minority feel that they
have to work in order to make ends meet, given the level of public student
support. However, the general pattern that the majority of students tend to
work also in Norway probably indicates that student’s eagerness to work
while studying is a way to get work experience, in order to enhance their
future employability (Cranmer 2006).
The analysis shows that there is a threshold for how much a student can
work in order not to increase the risk of dropout. Students who work
full-time or long part-time are less likely to complete their undergraduate uni-
versity programme than students who work short part-time or do not work at
all. Hence, working more than 20 h a week might make Norwegian students
more vulnerable to the risk of dropout. It is interesting to note that even in a
country that has generous student loans and grants students still work and the
Journal of Education and Work 647

effect of working on risk of dropout is similar to the risks found in countries


that charge tuition fees (see Roksa 2011; Moulin et al. 2012).
However, employment did not change the effect of any of the other
variables known to have an impact on dropout. As found in many earlier
studies the more resources a student possesses, the less likely it is they will
drop out, regardless of whether the student works or not (Yorke 1999;
Yorke and Longden 2004; Mastekaasa and Hansen 2005; Hovdhaugen
2009). This might indicate that the availability of financial support and the
absence of tuition fees only have a minor influence on the likelihood of
working and dropping out, indicating that students’ in countries with no tui-
tion fees are similar to students’ in countries with tuition fees in this regard.
Overall, the findings in this paper supports the findings of Roksa (2011)
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

and Moulin et al. (2012) that employment does affect dropout to some
extent, but it has a limited effect. However, its effect should not be
neglected, as it has been in several models of student departure. The find-
ings in this study illustrate that dropout is not just a matter of what happens
at university, as Tinto (1993) has argued, but that there are many factors
outside of higher education institutions’ control that might increase the risks
of dropout. When analysing their own institutional dropout rate, universities
should be aware that some of the dropout they observe possibly is a result
of students’ personal choices, rather than matters of higher education
experience or engagement, that the institution can change or improve.

Notes
1. Information on database: http://www.ssb.no/a/mikrodata/datasamlinger/nudb_
11514.html.
2. A model with interactions between employment status and independent
variables was also tested but is not displayed since it proved insignificant.

Notes on contributor
Elisabeth Hovdhaugen is a senior researcher at NIFU and holds a PhD in Sociology
from the University of Oslo. Her research interests centres around students: comple-
tion and dropout, student satisfaction and learning outcomes.

References
Aamodt, P. O. 2001. Studiegjennomføring og studiefrafall En statistisk oversikt
[Completion and Dropout, A Statistical Overview]. NIFU skriftserie 14/2001.
Oslo: NIFU.
Aamodt, P. O., and E. Hovdhaugen. 2011. Frafall og gjennomføring i lavere grads
studier før og etter Kvalitetsreformen. En sammenlikning mellom begynnerkul-
lene fra 1999, 2003 og 2005. oversikt [Dropout and Completion in Undergradu-
ate Degrees Before and After the Quality Reform]. NIFU rapport 38/2011. Oslo:
NIFU.
648 E. Hovdhaugen

Allison, P. D. 1984. Event History Analysis: Regression for Longitudinal Event


Data. Quantitative Applications in the Social Sciences: A Sage University Paper
Series No. 46. Beverly Hills, CA: Sage.
Arnesen, C. Å., E. Hovdhaugen, J. Wiers-Jenssen, and Per Olaf Aamodt. 2011.
Studiesituasjon og studentøkonomi. Norske resultater fra den europeiske
studentundersøkelsen Eurostudent IV [Study Situation and Student Finances].
NIFU rapport 33/2011. Oslo: NIFU.
Arnesen, C. Å., and S. Try. 2001. Karakterers betydning for overgangen fra høyere
utdanning til arbeidsmarked [The Significance of Grades for the Transition from
Higher Education to the Labour Market]. NIFU rapport 6/2001. Oslo: NIFU.
Arrow, K. J. 1973. “Higher Education as a Filter.” Journal of Public Economics 2
(3): 193–216.
Astin, A. W. 1975. Preventing Students from Dropping Out. San Francisco, CA:
Jossey-Bass.
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Barke, M., P. Braidford, M. Houston, A. Hunt, I. Lincoln, C. Morphet, I. Stone,


and A. Walker. 2000. Students in the Labour Market. Nature, Extent and
Implications of Term-time Employment among Niversity of Northumbria Under-
graduates. Department of Education and Employment Research Report No 215.
London: Department of Education and Skills.
Bean, J. P. 1980. “Dropouts and Turnover: The Synthesis and Test of a Causal
Model of Student Attrition.” Research in Higher Education 12 (2): 155–187.
Beerkens, M., E. Mägi, and L. Lill. 2011. “University Studies as a Side Job:
Causes and Consequences of Massive Students Employment in Estonia.” Higher
Education 61 (6): 679–691.
Berg, L. 1997. Studieløpet. Om tidsbruksvalg, faglige valg og kunnskapsteoretiske
valg [In the Course of Study]. NIFU rapport 3/97. Oslo: NIFU.
Berg, L., and P. O. Aamodt. 1987. Tid til studier? [Time to Study?]. Delrapport 3
fra prosjektet Studentenes tidsbruk og studieforhold Notat 2/87. Oslo: NAVFs
utredningsinstitutt.
Berg, L., and S. Kyvik. 1992. “Part-time Studying at Norwegian Universities.”
Higher Education 24 (2): 213–223.
Box-Steffensmeier, J. M., and B. S. Jones. 2004. Event History Modeling. A Guide
for Social Scientists. New York: Cambridge University Press.
Bradley, G. 2006. “Work Participation and Academic Performance: A Test of
Alternative Propositions.” Journal of Education and Work 19 (5): 481–501.
Braxton, J. M., ed. 2000. Reworking the Student Departure Puzzle. Nashville, YN:
Vanderbilt University Press.
Broadbridge, A., and V. Swanson. 2005. “Earning and Learning: How Term-time
Employment Impacts on students’ Adjustment to University Life.” Journal of
Education and Work 18 (2): 235–249.
Cranmer, S. 2006. “Enhancing Graduate Employability: Best Intentions and Mixed
Outcomes.” Studies in Higher Education 31 (2): 169–184.
Darmody, M., and E. Smythe. 2008. “Full-time Students? Term-time Employment
among Higher Education Students in Ireland.” Journal of Education and Work
21 (4): 349–362.
Devlin, M., R. James, and G. Grigg. 2008. “Studying and Working: A National
Study of Student Finances and Student Engagement.” Tertiary Education and
Management 14 (2): 111–122.
Egge, M. 1992. “Frafall blant begynnerstudentene [Dropout among First-year
Students].” In Begynnerstudenten Rapport 8/92, edited by L. Berg(red.), 51–62.
Oslo: NAVFs utredningsinstitutt.
Journal of Education and Work 649

Gambetta, D. 1987. Were they Pushed or Did they Jump? Individual Decision
Mechanisms in Education. Cambridge: Cambridge University Press.
Hall, R. 2010. “The Work-study Relationship: Experiences of Full-time University
Students Undertaking Part-time Employment.” Journal of Education and Work
23 (5): 439–449.
Hodgson, A., and K. Spours. 2001. “Part-time Work and Full-time Education in the
UK: The Emergence of a Curriculum and Policy Issue.” Journal of Education
and Work 14 (3): 373–388.
Hovdhaugen, E. 2004. Tidsbruk og ambisjon. Resultater fra stud.mag.-undersøkel-
sene 2001, 2002 og 2003 [Student Ambition and Use of Time]. NIFU Skriftser-
ie 16/2004. Oslo: NIFU.
Hovdhaugen, E. 2009. “Transfer and Dropout: Different Forms of Student
Departure in Norway.” Studies in Higher Education 34 (1): 1–17.
Hovdhaugen, E. 2011. “Do More Structured Studies Lead to Lower Rates of Drop-
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

out and Transfer from University?” Irish Educational Studies 30 (2): 237–251.
Hovdhaugen, E., and P. O. Aamodt. 2005. Frafall fra universitetet. En undersøkelse
av frafall og fullføring blant førstegangsregistrerte studenterved Universitetet i
Bergen [University Dropout: A Survey of Dropout and Completion Among
Undergraduate Students at University of Bergen], Universitetet i Oslo og Norges
teknisk-naturvitenskapelige universitetet (NTNU) høsten 1999. [University of
Oslo and NTNU] NIFU STEP Arbeidsnotat 13/2005. Oslo: NIFU STEP.
Hovdhaugen, E., and P. O. Aamodt. 2009. “Learning Environment – Relevant or
Not to Students’ Decision to Leave University?” Quality in Higher Education
15 (2): 177–189.
Humphrey, R. 2006. “Pulling Structured Inequality into Higher Education: The
Impact of Part-time Working on English University Students.” Higher
Education Quarterly 60 (3): 270–286.
Jenkins, S. P. 2004. Survival Analysis [Lecture Notes (Draft Book)]. Accessed
http://www.iser.essex.ac.uk/teaching/degree/stephenj/ec968/.
Lucas, R., and N. Lammont. 1998. “Combining Work and Study: An Empirical
Study of Full-time Students in School, College and University.” Journal of
Education and Work 11 (1): 41–56.
Lyngstad, J., and I. Øyangen. 1999. Sjung om studentens lyckliga dar: studenters
levekår 1998 [Student Living Conditions 1998]. Statistisk sentralbyrå rapport
15/99. Oslo: Statistisk sentralbyrå.
Mastekaasa, A., and M. N. Hansen. 2005. “Frafall i høyere utdanning: Hvilken
betydning har sosial bakgrunn?” [Dropout in Higher Education: The Signifi-
cance of Social Background] In Statistisk sentralbyrå: Utdanning2005 – deltak-
else og kompetanse, [Statistics Norway: Education 2005 – Participation and
Competence] Statistical Analyses, 98–121. Oslo: Statistisk sentralbyrå.
Michelsen, S., and P. O. Aamodt. eds. 2006. Kvalitetsreformen møter virkeligheten
[The Quality Reform Meets Reality]. Evaluering av Kvalitetsreformen: Delrap-
port 1. Oslo: NIFU STEP/Rokkansenteret/Forskningsrådet.
Mills, M. 2011. Introducing Survival and Event History Analysis. London: Sage.
Moulin, S., P. Doray, B. Laplante, and M. C. Street. 2012. “Work Intensity and Non-
completion of University: Longitudinal Approach and Causal Inference.” Journal
of Education and Work 26 (3): 333–356.
Nora, A., E. Barlow, and G. Crisp. 2005. “Student Persistence and Degree
Attainment Beyond the First Year in College: The Need for Research.” In Col-
lege Student Retention: Formula for Student Success, edited by A. Seidman,
129–154. Westport, CT: Praeger, ACE/Praeger Series on Higher Education.
650 E. Hovdhaugen

Opheim, V. 2008. “Student Finance in a Welfare State Effects of Reducing


Economic Barriers to Higher Education in Norway.” Doctoral diss., Faculty of
Social Science, University of Oslo, Oslo.
Opheim, V. 2011. “Changing the System of Student Support in Norway: Intended
and Unintended Effects on Students.” Scandinavian Journal of Educational
Research 55 (1): 39–59.
Pascarella, E. T., and P. T. Terenzini. 1991. How College Affect Students: Findings
and Insights from Twenty Years of Research. San Francisco, CA: Jossey-Bass.
Pascarella, E. T., and P. T. Terenzini. 2005. How College Affects Students. A Third
Decade of Research. San Francisco, CA: Jossey-Bass.
Richardson, M., C. Evans, and G. Gbadamosi. 2009. “Funding Full-time Studies
through Part-time Work.” Journal of Education and Work 22 (4): 319–334.
Roksa, J. 2011. “Differentiation and Work: Inequality in Degree Attainment in U.S.
Higher Education.” Higher Education 61 (3): 293–308.
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Schuh, J. H. 2005. “Finances and Retention: Trends and Potential Implications.” In


College Student Retention Formula for Student Success, edited by A. Seidman,
89–106. Westport, CT: American Council on Education/Praeger.
Singer, J. D., and J. B. Willett. 2003. Applied Longitudinal Data Analysis.
Modeling Change and Event Occurrence. New York: Oxford University Press.
Spence, M. 1973. “Job Market Signaling.” Quarterly Journal of Economics 87 (3):
355–374.
Tinto, V. 1975. “Dropout from Higher Education: A Theoretical Synthesis of
Recent Research.” Review of Educational Research 45 (1): 89–125.
Tinto, V. 1993. Leaving College. Rethinking the Causes and Cures of Student
Attrition. 2nd ed. Chicago: University of Chicago Press.
Ugreninov, E., and O. F. Vaage. 2006. Studenters levekår 2005 [Student living
conditions 2005]. SSB Rapport 2006/22. Oslo: Statistisk sentralbyrå.
Wang, H., M. Kong, W. Shan, and S. K. Vong. 2010. “The Effect of doing
Part-time Jobs on College Student Academic Performance and Social Life in
Chinese Society.” Journal of Education and Work 23 (1): 79–94.
Wiers-Jenssen, J. and P. O. Aamodt. 2002. Trivsel og innsats. Studenters tilfredshet
med lærested og tid brukt til studier [Student Effort and Student Welfare]
Resultater fra Stud.mag-undersøkelsene. Rapport 1/2002. Oslo: NIFU.
Yamaguchi, K. 1991. Event History Analysis. Applied Social Science Research
Methods Series. Vol. 28. Newbury Park, CA: Sage.
Yorke, M. 1999. Leaving Early Undergraduate Non-completion in Higher
Education. London: Falmer Press.
Yorke, M., and B. Longden. 2004. Retention and Student Success in Higher
Education. Maidenhead: The Society for Research into Higher Education and
Open University Press.
Journal of Education and Work 651

Appendix 1. Independent variable frequency

Table A1. Categorical variables.


Frequency Percentage
Employment status
Not-working 4780 37.6
Short part-time 6036 47.4
Long part-time 708 5.6
Working full-time 1202 9.4
Total number of cases 12,726 100
Field of study
Downloaded by [Florida Atlantic University] at 12:30 07 March 2016

Humanities 3950 31.0


Social sciences 6874 54.0
Sciences 1902 14.9
Total number of cases 12,726 100
Parent’s education
Compulsory education 2336 18.4
Upper secondary education 2944 23.1
HE 4 years or less 4865 38.2
HE more than 4 years 2581 20.3
Total number of cases 12,726 100

Table A2. Dichotomous variables.


% N
Share of women 65.1 12,726
Share of student from university region 40.1 12,726
Share of students with children 3.6 12,726

Table A3. Continuous variables.


Age Grades from upper secondary
Mean 20.9 0
Median 20 0.1
Std. deviation 3.8 6
Total number of cases 12,726 11,528

You might also like