A Longitudinal Study of Financial Difficulties and Mental Health in a National Sample of British Undergraduate Students.

TLDR: This paper examined longitudinal relationships over time between financial variables and mental health in students and found that greater financial difficulties predicted greater depression and stress cross-sectionally, and also predicted poorer anxiety, global mental health and alcohol dependence over time.

Abstract: Previous research has shown a relationship between financial difficulties and poor mental health in students, but most research is cross-sectional. To examine longitudinal relationships over time between financial variables and mental health in students. A national sample of 454 first year British undergraduate students completed measures of mental health and financial variables at up to four time points across a year. Cross-sectional relationships were found between poorer mental health and female gender, having a disability and non-white ethnicity. Greater financial difficulties predicted greater depression and stress cross-sectionally, and also predicted poorer anxiety, global mental health and alcohol dependence over time. Depression worsened over time for those who had considered abandoning studies or not coming to university for financial reasons, and there were effects for how students viewed their student loan. Anxiety and alcohol dependence also predicted worsening financial situation suggesting a bi-directional relationship. Financial difficulties appear to lead to poor mental health in students with the possibility of a vicious cycle occurring.

Full text

ORIGINAL PAPER
 Thomas Richardson
thr1g10@soton.ac.uk
1
Mental Health Recovery Team North, Solent NHS Trust,
St. Mary’s Community Health Campus, Milton Road,
Portsmouth PO3 6AD, UK
2
School of Psychology, University of Southampton,
Southampton SO17 1BJ, UK
3
Psychology, Kingston University, London, UK
4
Department of Psychology, Kingston University,
Surrey KT1 2EE, UK
Received: 8 December 2015 / Accepted: 22 July 2016
© The Author(s) 2016. This article is published with open access at Springerlink.com
A Longitudinal Study of Financial Difculties and Mental Health
in a National Sample of British Undergraduate Students
Thomas Richardson
1,2,3
· Peter Elliott
2,3
· Ron Roberts
4
· Megan Jansen
1,2,3
Keywords Debt · Financial · Mental Health ·
Undergraduate · Student
Introduction
University represents a high risk time for mental health
problems, with the start of university coinciding with the
mean age of onset for many psychiatric disorders (Reav-
ley et al. 2012). A United States (US) nationwide survey
reported that almost half of all university-aged students
have a psychiatric disorder which has functionally impaired
them during the last academic year; however similar rates
were reported for similar aged peers who did not attend
university (Blanco et al. 2008). Similarly, Eisenberg et al.
(2007) found that 15.6 % of US university students met cri-
teria for a depressive or anxiety disorder. In Turkey, Bayram
and Bilgel (2008) found moderately severe depression in
27 % of students and moderately severe anxiety in 47 %.
Research also suggests that mental health may worsen over
the course of university: Andrews and Wilding (2004) found
that 9 % of United Kingdom (UK) students without a history
of mental health problems at the start of university went on
to develop clinical depression halfway through their degree.
They also found that 20 % became clinically anxious over
this time period.
The potential impact of poor mental health amongst stu-
dents has raised growing concerns with studies reporting it
to interfere with university attendance, as well as reducing
the likelihood of completing university (Blanco et al. 2008).
High rates of substance use and alcohol use disorders are
reported in students (Dawson et al. 2004; Slutske 2005),
though rates may be similar to non-student populations
(Blanco et al. 2008). Research in the US has also shown a
Abstract Previous research has shown a relationship
between nancial difculties and poor mental health in
students, but most research is cross-sectional. To exam-
ine longitudinal relationships over time between nancial
variables and mental health in students. A national sample
of 454 rst year British undergraduate students completed
measures of mental health and nancial variables at up to
four time points across a year. Cross-sectional relationships
were found between poorer mental health and female gen-
der, having a disability and non-white ethnicity. Greater
nancial difculties predicted greater depression and
stress cross-sectionally, and also predicted poorer anxiety,
global mental health and alcohol dependence over time.
Depression worsened over time for those who had consid-
ered abandoning studies or not coming to university for
nancial reasons, and there were effects for how students
viewed their student loan. Anxiety and alcohol dependence
also predicted worsening nancial situation suggesting a
bi-directional relationship. Financial difculties appear to
lead to poor mental health in students with the possibility
of a vicious cycle occurring.
Community Ment Health J (2017) 53:344–352
DOI 10.1007/s10597-016-0052-0
/ Published online: 7 July 2016
123

(Richardson et al. 2015a). This same data set has also been
examined in relation to eating disorder risk and nancial
difculties (Richardson et al. 2015b). In the current study
a longitudinal design was used to assess whether nancial
variables inuence changes in mental health overtime in
undergraduate students.
Measures
The following standardised measures were used to assess
mental health. For all the measures higher scores represent
more severe symptoms/worse mental health.
● Clinical Outcomes Routine Evaluation-General Popula-
tion Version (CORE-GP) (Sinclair et al. 2005): This is
a 14 item measure of global mental health with ques-
tions such as ‘I have felt optimistic about my future’ and
‘I have felt tense, nervous or unhappy’. In the current
sample α at time 1
=
.90.
● 7 Item Generalized Anxiety Disorder Questionnaire
(GAD-7) (Spitzer et al. 2006): A seven item measure
designed to screen for generalised anxiety disorder asking
frequency of symptoms in past 2 weeks such as ‘trouble
relaxing’ and ‘not being able to stop or control worrying’.
Scores above 10 are suggestive of generalized anxiety
disorder (Spitzer et al. 2006). This measure has also been
used to measure anxiety the general population (Löwe et
al. 2008). In the current sample α at time 1
=
.91.
● Centre for Epidemiological Studies Depression Scale
(CES-D) (Radloff 1977): This 20 item measure is
designed to measure symptoms of depression in the past
2 weeks, and is designed specically for epidemiologi-
cal research with general population samples. Questions
ask about frequency of symptoms such as ‘I was happy’
and ‘I felt that people disliked me’. Scores above 15 are
suggestive of depression (Radloff 1977). In the current
sample α at time 1
=
.95.
● Perceived Stress Scale (PSS) (Cohen et al. 1983): This
10 questionnaire assess global perceived stress in the
last month, using items such as how often individuals
have felt ‘unable to control the important things in your
life’ or ‘felt that things were going your way?’ In the
current sample α at time 1
=
.90.
● Alcohol Use Disorder Identication Test (AUDIT)
(Saunders et al. 1993): This 10 item scale was devel-
oped to assess for alcohol problems via questions such
as ‘How often do you have a drink containing alcohol?’
and ‘Have you or someone else been injured because of
your drinking?’ Scores above 7 are suggestive of pos-
sible alcohol abuse or dependence (Babor et al. 2001).
The AUDIT has been shown to be accurate in detecting
alcohol problems in US university students (Kokotailo
et al. 2004). In the current sample α at time 1
=
.89.
high prevalence of suicidal ideation in students (Garlow et
al. 2008).
One factor which has consistently been shown to predict
poor mental health in students is nancial difculties. A
number of studies examining UK based students have shown
that mental health problems are linked to nancial problems
(Andrews and Wilding 2004; Roberts et al. 2000, 1999),
level of debt (Carney et al. 2005) and concern about nances
(Cooke et al. 2004; Jessop et al. 2005). The pooled ndings
from a meta-analysis by Richardson et al. (2013) found that
41.7 % of those with a mental health disorder report being in
debt, in comparison to 17.5 % who report having no debt. For
those who were in debt, 15.5 % had a mental health disorder
in comparison to 8.9 % of those not in debt. A statistically
signicant relationship was also found between debt and
depression, suicide completion or attempt, problem drink-
ing, drug dependence, neurotic disorders and psychotic dis-
orders (Richardson et al. 2013). The chief methodological
issues identied with research in the area concern the use of
non-validated measures of mental health problems, as well
as the noted paucity of longitudinal studies.
Three longitudinal studies conducted with students are of
particular relevance. Cooke et al. (2004) followed students
for 3 years and found those with a high level of concern
about their nances had a greater deterioration in mental
health over time. Richardson et al. (2015a) examined the
impact of the recent rise in tuition fees for UK students on
mental health, nding no signicant impact with those pay-
ing more having poorer mental health at only one out of
four time points. However, using the same data Richardson
et al. (2015b) found that nancial difculties in students
increased eating disorder risk in students up to a year later.
The relationship was partly bi-directional with eating dis-
order risk also increasing the risk of nancial difculties
3 months later. It may therefore be that it is nancial dif-
culties such as ability to pay the bills which is more impor-
tant than size of student loan.
At present therefore there has been little longitudinal
research on the relationship between nances and mental
health in students, and in particular only one previous study
has examined whether nances predict poor mental health
or vice versa in students. The present study therefore aimed
to address this gaps in the literature by measuring a range of
mental health symptoms and nancial variables over time in
a UK student sample.
Methods
Design
This study uses data from a prospective cohort study
on tuition fees amount and mental health in students
Community Ment Health J (2017) 53:344–352
345
123

points, 27.8 % (n
=
155) completing three time points and
34.1 % (n
=
155) completing two time points. The sample
was 77.9 % (n
=
352) female, 89.6 % (n
=
405) white ethnic-
ity, 5.7 % (n
=
28) mixed ethnicity, 1.5 % (n
=
7) Black, 1.5 %
(n
=
7) Asian, 1.1 % (n
=
5) ‘Other’ and 0.8 % (n
=
2) did not
state. Ages ranged from 17 to 57 with a mean of 19.9 years.
Eleven per cent (n
=
50) reported being mature students (age
21 or over at start of university) and 8.8 % (n
=
40) reported
that they had a disability.
Statistical Analyses
Missing data were lled in with the mode. All measures
were normally distributed. Hierarchical, linear hierarchical
multiple regression was used to see whether the nancial
variables (FAS, IFS, considering dropping out or not com-
ing to university due to nances, whether got rst choice,
how stressed about debt) predicted scores at each time point.
Demographic variables were also included in the model
(age, gender, disability, mature student, ethnicity). For times
2 to 4 the baseline scores for that mental health measure
were also included, for example to see if anxiety at time 3
was affected by demographics after accounting for anxiety
score at baseline. The dummy variable was the most com-
mon variable, and listwise deletion was used for missing
data in the regression. A linear regression was also used to
see whether baseline mental health impacted follow-up IFS.
Results
Baseline Finances and Follow-Up Mental Health
The nal linear regression models examining the impact on
nancial variables at baseline on follow-up mental health
are shown in Tables 1 and 2. Female gender predicted higher
anxiety and stress but lower alcohol dependence at baseline.
Having a disability predicted poorer global mental health,
higher depression, anxiety and stress at baseline and greater
anxiety and stress at T2. Having a disability also predicted
lower alcohol dependence at T4. Mature students had sig-
nicantly lower alcohol dependence at baseline and time 2.
There was no effect of age on any of the variables.
Other ethnicity (compared to white) was associated with
poorer global mental health at time 2, and white ethnicity
was associated with greater alcohol dependence at base-
line when compared to those of black or Asian ethnicity.
Family afuence was not related to any variables. Greater
nancial stress predicted greater anxiety, depression, stress,
alcohol dependence and poorer global mental health at
baseline. Greater nancial stress at baseline also predicted
greater anxiety at T2 and greater alcohol dependence at
T3. Greater subjective stress about debt predicted greater
The following measures of nances were used:
● Family Afuence Scale (FAS) (Currie et al. 1997): This
four item measure is designed to measure the socio-eco-
nomic status of adolescents via four questions such as
‘Does your family own a car, van or truck?’ This was
used to measure the socio-economic status of student’s
families. (α cannot be calculated for this measure as
item responses differ between questions).
● Index of Financial Stress (IFS) (Siahpush and Carlin
2006): This measures nancial difculties/stress over
the past 6 months via questions such as ‘Could not pay
the mortgage or rent on time’. In the current sample α at
time 1
=
.71.
● Author constructed questions were developed on other
nancial variables. Participants were asked ‘How
stressed do you feel about your level of debt?’ with
response options ‘Not stressed’, ‘A Little stressed’,
‘Quite stressed’ or ‘Very stressed’. They were asked
‘Was this your rst University choice?’ with response
options ‘Yes: was my rst choice’, ‘No: Was an insur-
ance or back-up choice’ or ‘No: I got the offer through
clearing’. Participants were asked: ‘How do you see
your student loan?’ with response options ‘Debt I will
have to pay back’, ‘Debt I might have to pay back’ or
‘An extra tax (rather than debt)’. Finally, participants
were asked ‘Have you seriously considered abandoning
your course because of any nancial difculties?’ (For
example talking to your tutor about doing so, looking
into career options etc.), with a Yes/No response. They
were also asked ‘Did you seriously consider not coming
to University due to nancial concerns?’ (For example
did you look into other career options, apply for jobs
etc.), with a Yes/No response.
Participants and Procedure
Recruitment is described in detail in the original paper
(Richardson et al. 2015a, b). British rst year undergradu-
ate students were eligible to take part. Students were con-
tacted through their university students union who were
contacted by researchers and invited to advertise a survey
examining factors which effect mental health in students.
The universities included a wide range in terms of ranking
and geographic location, and students were from a range of
disciplines.
The measures were completed online at four time points
over the participants rst 2 years at university. Each time
point was 3–4 months apart, with the overall length from
time 1 to time 4 being just over 1 year. Participants were
only included in current analysis if they completed baseline
and at least one other time point. A total of 454 participants
were included with 38.1 % (n
=
173) completing all four time
Community Ment Health J (2017) 53:344–352
346
123

GAD-7 (anxiety) CES-D (depression) PSS (stress)
Baseline T2 T3 T4 Baseline T2 T3 T4 Baseline T2 T3 T4
Overall model
n 427 371 242 215 435 377 248 219 423 362 241 213
F (df) 5.42***
(20,
406)
18.25***
(21,
349)
10.42***
(21,
220)
5.84***
(21,
193)
6.40***
(20,
414)
20.94***
(21,
355)
12.91***
(21,
226)
6.19***
(21,
197)
5.95***
(20,
402)
17.14***
(21,
340)
9.44***
(21,
219)
7.71***
(21,
191)
R
2
.21 .52 .50 .39 .24 .55 .55 .40 .23 .51 .48 .46
Individual predictors (β)
a
Gender (female)
−
.09*
−
.07
−
.07
−
0.4
−
.03
−
.01
−
.07 .02
−
.11*
−
.06
−
.10 .02
(No disability) vs, disability .14** .08*
−
.01
−
.02 .14** .05
−
.03
−
.03 .11* .09* .05 .02
(Not mature student) vs. mature student
−
.08
−
.01 .03
−
.06
−
.12 .03
−
.06
−
.01
−
.12 .07
−
.01
−
.03
Age
(17–19) vs. 20–29 .03
−
.03
−
.04 .06 .04
−
.03
−
.01 .05 .01
−
.07
−
.04 .05
(17–19) vs. 30
+ −
.02 .02
−
.02 .02
−
.04
−
.04 .04 .05 .00
−
.04
−
.01
−
.02
Ethnicity
(White) vs. other
−
.04 .06
−
.02 .05
−
.06 .05 .02 .03
−
.08 .04 .01 .10
(White) vs. mixed .02 .01
−
.04
−
.06 .05 .04
−
.01
−
.02 .02 .02 .03 .00
(White) vs. Asian .03
−
.01
−
.05
−
.06
−
.03 .00
−
.03
−
.05
−
.01
−
.02
−
.06
−
.08
(White) vs. Black
−
.03 .01
−
.04
−
.05
−
.00 .02
−
.05
−
.03 .03 .03
−
.04
−
.06
Family afuence scale
−
.03
−
.03 .02 .00
−
.02
−
.00
−
.02
−
.02
−
.03
−
.07
−
.05
−
.05
Baseline index nancial stress .15* .13* .01 .05 .19** .08 .00 .01 .17** .06 .08 .14
How stressed about debt
(Not at all) vs. a little .06 .04 .05 .15* .14** .02 .04 .09 .13* .10* .09 .15*
(Not at all) vs. quite .14** .05 .04 .16* .16** .04
−
.02 .12 .16** .09 .03 .17*
(Not at all) vs. very .09
−
.04
−
.02 .03 .11
−
.03
−
.07 .02 .17** .00
−
.06 .02
Considered abandoning studies due to nances
(No) vs. yes .11 .04 .10 .16 .12*
.09 .13* .18* .11 .01 .12 .13
Considered not coming to University due to nances
(No) vs. yes .12*
−
.03 .01
−
.06 .12* .02 .13*
−
.11 .06 .06 .03
−
.03
How see student loan
(Debt have to pay back) vs. an extra tax
−
.09 .05 .03 .03*
−
.04* .06 .03
−
.01*
−
.04 .06 .06 .05
(Debt have to pay back) vs. debt might have to pay back .01 .07 .04 .08 .07* .05 .06
−
.00 .10* .03 .02 .03
Table 1 Final regression models of nancial variables and follow-up anxiety, depression and stress
Community Ment Health J (2017) 53:344–352
347
123

anxiety at baseline and T4, greater depression at baseline,
greater stress at baseline T2 and T4, and poorer global men-
tal health at baseline and T4. However, those who were less
stressed about their nances had greater alcohol dependence
at T4.
Considering abandoning studies due to nancial reasons
predicted higher depression at T3 and T4. Considering not
coming to university for nancial reasons predicted greater
anxiety and poorer global mental health at baseline and
greater depression at T3. Those who saw their student loan
as debt they have to pay back had lower scores on anxiety
at T4 than those who saw it as an extra tax. However, those
who saw it as debt they had to pay back had more severe
depression than those who saw it as an extra tax at base-
line and T4. Those who saw it as debt they might have to
pay back had more severe depression and stress at baseline
than those who saw it as debt they have to pay back. Finally
those who got their rst choice university had more severe
depression than those who got their back-up choice at T3,
and more severe alcohol dependence at T2.
Baseline Mental Health and Follow-Up Finances
A regression model with baseline IFS, FAS demographics
and all of the mental health measures signicantly predicted
IFS at T2: F(14,357)
=
52.5, p
<
.001, R
2
=
.67. Individual
signicant predictors of higher IFS at T2 were baseline
IFS (β
=
.71, p
<
.001) being age 30
+
(compared to 17–19):
β
=
.09, p
<
.01, Other ethnicity (compared to white): β
=
.08,
p
<
.01 and low Family Afuence: β
=
−
.08, p
<
.05. None
of the mental health measures at baseline were signicant
predictors.
The model signicantly predicted IFS at T3
F(14,233)
=
21.7, p
<
.001, R
2
=
.57. Individual signicant
predictors of higher IFS at T2 were baseline IFS (β
=
.76
p
<
.001), being age 30 (compared to 17–19): β
=
.15, p
<
.01,
Other ethnicity (compared to white): β
=
.15, p
<
.001 and
baseline CORE-GP score (β
=
.26, p
<
.05).
Finally, the model signicantly predicted IFS at T4
F(14,204)
=
11.7, p
<
.001, R
2
=
.45. Individual signicant
predictors of higher IFS at T2 were baseline IFS (β
=
.57
p
<
.001) being Other ethnicity (compared to white): β
=
.16,
p
<
.01 and baseline AUDIT score (β
=
.13, p
<
.05).
Discussion
The present study examined the longitudinal relationship
between nancial variables and mental health in a UK
student population using standardised measures. Greater
nancial stress such as being unable to pay the bills pre-
dicted poorer global mental health and higher anxiety,
depression, stress and alcohol dependence when examined
GAD-7 (anxiety) CES-D (depression) PSS (stress)
Baseline T2 T3 T4 Baseline T2 T3 T4 Baseline T2 T3 T4
University choice
(First choice) vs. clearing .05 .06 .03 .02 .07 .05
−
.02 .08 .07 .03
−
.02 .04
(First choice) vs. back-up
−
.03 0.2
−
.06
−
.01 .00
−
.03
−
.10*
−
.01
−
.02
−
.03
−
.06
−
.08
Baseline mental health measure score .62*** .62*** .48*** .64*** .63*** .52*** .59*** .55*** .46***
*p
<
.05, **p
<
.01, ***p
<
.001
a
Dummy/reference variables are in brackets. If β values are
+
then the comparison variable is associated with a higher score, if β values are
−
, then the dummy variable is associated with a
higher score
Table 1 (continued)
Community Ment Health J (2017) 53:344–352
348
123