Journal of Mental Health

ISSN: 0963-8237 (Print) 1360-0567 (Online) Journal homepage: http://www.tandfonline.com/loi/ijmh20

Academic and non-academic predictors of student

psychological distress: the role of social identity
and loneliness

Jason C. McIntyre, Joanne Worsley, Rhiannon Corcoran, Paula Harrison

Woods & Richard P. Bentall

To cite this article: Jason C. McIntyre, Joanne Worsley, Rhiannon Corcoran, Paula Harrison
Woods & Richard P. Bentall (2018): Academic and non-academic predictors of student
psychological distress: the role of social identity and loneliness, Journal of Mental Health, DOI:
10.1080/09638237.2018.1437608

To link to this article: https://doi.org/10.1080/09638237.2018.1437608

View supplementary material

Published online: 13 Feb 2018.

Submit your article to this journal

Article views: 5

View related articles

View Crossmark data

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

http://www.tandfonline.com/action/journalInformation?journalCode=ijmh20
 http://tandfonline.com/ijmh
ISSN: 0963-8237 (print), 1360-0567 (electronic)

J Ment Health, Early Online: 1–10

ß 2018 Informa UK Limited, trading as Taylor & Francis Group. DOI: 10.1080/09638237.2018.1437608

ORIGINAL ARTICLE

Academic and non-academic predictors of student psychological

distress: the role of social identity and loneliness
Jason C. McIntyre1 , Joanne Worsley1, Rhiannon Corcoran1 , Paula Harrison Woods2, and Richard P. Bentall3
1
Institute of Psychology, Health and Society, University of Liverpool, Liverpool, UK, 2Student Administration and Support Division, University of
Liverpool, Liverpool, UK, and 33Department of Psychology, Clinical Psychology Unit, The University of Sheffield, Sheffield, UK

Abstract Keywords
Background: University students experience high rates of stress and mental illness; however, Academic stress, social identity, loneliness,
few studies have comprehensively examined the impact of academic and non-academic mental health, stress, student mental
stressors on student mental health. Similarly, there has been little focus on the role of social health
groups in protecting against mental distress in this young adult group.
Aim: To identify the key social determinants of mental health symptoms in a student History
population.
Methods: Using an online survey, we administered measures of social connectedness and Received 30 June 2017
mental health symptoms alongside academic and non-academic stressors to a large sample of Revised 22 November 2017
UK university students. Accepted 20 December 2017
Results: Loneliness was the strongest overall predictor of mental distress, while assessment Published online 13 February 2018
stress was the most important academic predictor. Strong identification with university
friendship groups was most protective against distress relative to other social identities, and the
beneficial impact of identification on symptoms was mediated by reduced loneliness.
Conclusions: The study highlights the benefits of establishing strong social connections at
university and the importance of minimising stress associated with assessment tasks.

Introduction determinant of depressive symptoms (Lester, 2014). Indeed,

in a review of 40 qualitative studies, it was found that
Adolescence and early adulthood are known to be periods of
relationship stressors (i.e. family, romantic, peer and faculty
peak risk for the onset of mental disorders (Kessler et al.,
relationships) were the most commonly reported source of
2007). Following the recent expansion in access to higher
stress among university students. Other commonly reported
education in the UK, with a cohort entry rate for English
stressors included high expectations from oneself and others,
school leavers reaching 42.1% in 2014 (University Central
and a lack of tangible coping resources such as time, sleep,
Admission Service, 2015), a substantial section of young
support and money (Hurst et al., 2013).
people in this risk period are now studying at university.
Stressors encountered by students are similar to those in
Indeed, a 2016 poll of over one thousand UK students found
other professions, hence researchers have applied organisa-
that more than a quarter of respondents suffered from a
mental health problem (YouGov, 2016). Here, we aim to tional psychology models, such as the Job Demands-
detail the prevalence of mental distress in the student Resources Model (JD-R; Bakker & Demerouti, 2007), to
population, to provide a comprehensive examination of the student mental health (e.g. Pluut et al., 2015). The JD-R
impact of academic and non-academic stressors on student model asserts that high-pressure jobs place heavy demands on
mental health and to identify social factors that might mitigate the mental and physical resources of employees (or students),
these issues. which in turn lead to health problems and poor performance.
Students experience psychological demands both inside Job resources, conversely, help people to achieve work goals
and outside of the classroom. Financial, academic and social- and reduce psychological distress. Like people in the
related stressors are some of the most common that university workforce, students have resources upon which they can
students encounter (Ross et al., 1999). University-related draw to help them cope with university life, including social
stressors are a moderate predictor of depression; however, (e.g. fellow students), psychological (e.g. self-esteem) and
general life stressors have been found to be a more important practical (e.g. student support services) resources. These
resources may help to alleviate some of the demands outlined
Correspondence: Jason C. McIntyre, Institute of Psychology, Health and earlier, such as financial strain and expectation stress. The JD-
Society, University of Liverpool, Waterhouse Building, Block B, 1-5
Brownlow St, Liverpool L69 3GL, United Kingdom. E-mail: R model therefore seems appropriate to apply to student
j.mcintyre@liverpool.ac.uk populations.
2 J. C. McIntyre et al. J Ment Health, Early Online: 1–10

A study by Pluut and colleagues (2015) that applied the Higher education presents students with challenges and
JD-R model found that academic stressors were associated stressful circumstances, but also offers opportunities to derive
with reduced well-being and poorer performance among meaning, purpose and belonging through mastery and social
Dutch university students. Moreover, conflict between leisure connection. In the present research, we aim to detail the
and study activities, along with social support, were moderate prevalence of mental distress in a student population and
predictors of both academic satisfaction and academic assess a diverse set of social determinants, which include
performance, which suggests that psychosocial factors are academic stressors such as performance and assessment
important determinants of well-being and success in academic stress, as well as social connection (e.g. loneliness and
settings. Longitudinal evidence indicates that entering higher social identification) and background variables (e.g. child-
education has both positive and negative effects on mental hood deprivation and maltreatment) known to be associated
health. A 2004 study found that among UK students who had with poor mental health (Cruwys et al., 2014; Hill, 2003;
no psychological symptoms at course entry, 9% became McIntyre et al., 2017; Varese et al., 2012). We also look at
clinically depressed and 20% clinically anxious by the mid- additional stressors likely to be relevant to young people
point of their degrees. Financial stress and relationship attending university, such as cyberbullying, financial stress
difficulties were identified as the main predictors of depres- and poor living conditions. We examine these stressors in the
sion and anxiety, respectively. However, during the course of context of three distinct mental health symptoms: depression,
the study, 36% of students with prior conditions showed some anxiety and paranoia, which are common among young adults
recovery, suggesting that universities may also afford positive and have been associated with social determinants (Cruwys
effects on mental health (Andrews & Wilding, 2004). One et al., 2014; Lee & Robbins, 1998; McIntyre et al., 2017). A
plausible explanation for the improvements in mental health final aim was to test whether identification with university-
observed for these students is that universities provide relevant groups can be considered a psychosocial resource
opportunities for meaningful social connections. Indeed, that reduces symptom risk, and to identify the mechanisms by
social support has been shown consistently to reduce stress which social groups might improve mental health.
in the workplace (see meta-analysis by Viswesvaran et al.,
1999) and to protect people from developing mental health Method
symptoms following exposure to stressors (Hagerty & Participants and design
Williams, 1999), particularly when those stressors are
severe (Smith et al., 2013). The online survey was conducted in October 2016 across the
According to the Social Cure Model of health (Jetten et al., faculties of a large university in northern England. A total of
2012), when people feel bonded to a social group and the 1545 students attempted the survey. Surveys missing more
group is incorporated into their sense of self through the than 25% of responses were considered incomplete, leaving a
process of social identification (Tajfel & Turner, 1979), the final sample of 1135. Students from all three faculties
group becomes a psychological resource that improves health completed the survey: Health and Life Sciences (30%),
both directly and indirectly. When the groups to which we Humanities and Social sciences (42%) and Science and
belong are positive and successful, they foster positive Engineering (18%). First-year students comprised 46% of the
emotions and enhance our sense of self-worth. Because of sample, while second- and third-year students made up 35%
this, social identities are central to psychological health and and 21%, respectively. The majority of participants were from
well-being. white ethnic backgrounds (82%). Seventy-one percent identi-
Social identity is distinct from social support, but tends to fied as female and 26% identified as male. The average age of
change the way social support is given and received. For participants was 20.78 years, SD ¼ 4.35.
example, the positive effects of workplace social support on
employee training outcomes are most pronounced when Measures
people identify with their workplace (Pidd, 2004). Further,
Academic stress
social support is more likely to be given, received and
effective when the support is built on a foundation of shared Academic Stress Scale. We administered an adapted 15-item
social identity (Haslam et al., 2009). version of the Academic Stress Scale (Kohn & Frazer, 1986).
Social identity has also been shown to have a direct effect Three sub-scales consisting of five items tapped how stressed
on symptoms of mental illness, such as depression (Cruwys students were about performance (e.g. ‘‘Final grades’’;
et al., 2013, 2014, 2015), paranoid ideation (McIntyre et al.,  ¼ 0.75), teaching (e.g. ‘‘Fast paced lectures’’;  ¼ 0.72),
2017; Sani et al., 2017; Thomas et al., 2017), anxiety and facilities (e.g. ‘‘Poor access to computing facilities’’;
(Wakefield et al., 2013), well-being and post-traumatic stress  ¼ 0.79). Response options ranged from 1 ¼ not at all
(Swartzman et al., 2017). Moreover, in line with the Social stressed to 5 ¼ extremely stressed. Some of the original items
Cure Model and Tajfel and Turner’s (1979) original concep- that were irrelevant (e.g. forgetting pencil/pen) were either
tualization of identity, part of the relationship between social dropped or modified to reflect modern learning environments.
identification and better mental health can be explained by the See supplementary materials (Appendix A) for the complete
notion that belonging to social groups promotes more positive modified scale.
self-attributions (Cruwys et al., 2015; McIntyre et al., 2017).
Social groups can therefore be conceptualised as a psycho- We administered an abbreviated version
Expectations stress.
logical resource that provides people with fortification against of the Academic Expectations Stress Inventory (Ang & Huan,
distress by increasing belonging and self-worth. 2006), which tapped stress related to students’ own
DOI: 10.1080/09638237.2018.1437608 Social determinants of student mental health 3

expectations and those of their teachers and parents (‘‘I feel ‘‘To what extent does your debt worry you?’’. Participants
stressed when I know I/my parents/my university teachers are responded on a five-point scale ranging from 1¼ not at all to
disappointed in my grades’’; Anchors: 1 ¼ never true, 5 ¼ a lot. Factor summary scores were derived via PCA for
5 ¼ always true) A principle components analysis (PCA) of the two items, and these scores constituted the financial worry
the three items yielded a single factor (eigenvalue ¼ 2.03; variable, eigenvalue ¼ 1.75; 88% of variance accounted for.
68% of variance accounted for). Factor scores were derived
using the regression method and used as summary scores in Perceived discrimination
all subsequent analyses. Participants completed the two-item Perceptions of
Relative performance. Two novel items assessed how well Discrimination scale (Major et al., 2007) for five different
students were performing relative to their peers and relative to types of potential discrimination (10 items in total). Items
their own expectations (‘‘In relation to your grades, how well included ‘‘My [group] is discriminated against’’ and ‘‘Other
are you doing relative to your own expectations/peers at members of [my group] experience discrimination’’.
university’’). A PCA of the two items yielded a single factor Participants responded to both items for each of the following
(eigenvalue ¼ 1.45; 72% of variance accounted for). Factor groups: ethnicity, gender, sexuality, disability and religion/
scores were derived using the regression method and used as beliefs. Pearson’s rs ranged from 0.82 to 0.90. Response
summary scores in subsequent analyses. options ranged from 1 ¼ strongly disagree to 7 ¼ strongly
agree.
Loneliness
Cybervictimisation
The eight-item UCLA Loneliness Scale (ULS-8) assesses
how withdrawn people are from social relationships and Participants answered two items adapted from Hinduja &
companionship (e.g. ‘‘I feel lonely from others’’ and ‘‘people Patchin (2010) related to the frequency with which they had
are around me but not with me’’). The scale showed good experienced ‘‘cyberbullying’’ and ‘‘cyberstalking’’. Response
internal consistency  ¼ 0.87. options ranged from 1 ¼ never to 6 ¼ very often,
r(1133) ¼ 0.45.
Social identity
Childhood disadvantage
We included a three-item scale adapted from Doosje et al.
(1995) measure of in-group identification. Participants Participants completed the 12-item Perceived Inequality in
responded to each of the three items for six different social Childhood Scale (Wickham et al., 2013). For example, ‘‘In
groups (18 items in total). Specifically, they indicated the comparison to other children in your school and neighbour-
extent to which they felt ‘‘strong group ties’’, ‘‘belonging’’ hood, your parents’ involvement in your education was:’’
and ‘‘identification’’ with each group. Groups were selected 1 ¼ far less to 5 ¼ significantly more,  ¼ 0.83.
that were (a) likely to be important to university students, (b)
likely to be relevant to the majority of participants and (c) had Childhood maltreatment
the potential to be incorporated into university social The 10-item ACEs scale (Felitti et al., 1998) assesses
connection programs. These identities included: country of childhood abuse (e.g. ‘‘Did a parent or other adult in the
birth ( ¼ 0.90), England ( ¼ 0.92), university city household often push, grab, slap or throw something at you?
( ¼ 0.91), university ( ¼ 0.87), primary online community Or ever hit you so hard that you had marks or were injured?’’)
( ¼ 0.91) and university friends,  ¼ 0.93. and maladaptive family environments (e.g. ‘‘Did you live
with anyone who was a problem drinker or alcoholic or who
Living conditions used street drugs?’’). The total number of ‘‘yes’’ responses
Living conditions were assessed with three items taken from was tallied.
the English Housing Survey (Department for Communities
and Local Government, 2015). The items assessed accom- Paranoia
modation-related health issues (Does the condition of your
accommodation affect your health in any way? Anchors: Paranoia was assessed with an abbreviated five-item perse-
1 ¼ all of the time, 4 ¼ never), house maintenance (Overall, cution subscale of the persecution and deservedness scale
how satisfied are you with the way your landlord repairs and (PaDS; Melo et al., 2009). For example, ‘‘You should only
maintains your home? Anchors: 1 ¼ very satisfied, 5 ¼ very trust yourself’’. Response options ranged from 1 ¼ strongly
dissatisfied) and mould (During the winter months, are there disagree to 5 ¼ strongly agree,  ¼ 0.84.
patches of mould or fungus in any room in your home, apart
from bathrooms or toilets? Options: 1 ¼ no, 2 ¼ yes). A PCA Depression
yielded a single factor with an eigenvalue of 1.31, which
Depression was measured with the nine-item Patient Health
accounted for 43.65% of the variance in the data. The derived
Questionnaire (PHQ-9; Kroenke & Spitzer, 2002).
factor scores constituted the living conditions variable.
Participants indicated how often they had been bothered by
problems such as ‘‘Feeling down, depressed or hopeless’’ and
Financial worry
‘‘Thoughts that you would be better off dead’’ over the last
The Debt Worry Scale (Cooke et al., 2004) consists of two two weeks. Response options ranged from 1 ¼ not at all to
items: ‘‘Are financial concerns a current issue?’’ and 4 ¼ nearly every day,  ¼ 0.89.
4 J. C. McIntyre et al. J Ment Health, Early Online: 1–10

Anxiety as not life-threatening, whilst 2.5% of the entire sample

reported a determined suicide attempt.
The Generalized Anxiety Disorder-7 (GAD-7; Spitzer et al.,
2006) is a seven-item scale that assesses frequency of anxious
symptoms over the past two weeks, for example, ‘‘worrying Academic and non-academic stressors
too much about different things’’. Response options ranged To understand how an accumulation of stress factors impact
from 1 ¼ not all to 4 ¼ nearly every day,  ¼ 0.92. on mental health, a series of hierarchical regressions were
conducted to determine which stressors were the most
Self-harm
important determinants of symptoms. We included depres-
Four self-harm items were taken from separate sources. Item sion, anxiety and paranoia as dependent variables. Predictor
1 (‘‘During the past 12 months, did you ever seriously variables were entered into the model at different blocks, with
consider attempting suicide?’’) was adapted from the Youth each block representing a distinct cluster of associated
Risk Behaviour Surveillance System survey (Youth Risk stressors. Entering the variables in this manner allowed us
Behavior Surveilance System, 2015). Item 2 (Have you to examine the role of categories of stressor and also
deliberately hurt yourself without trying to kill yourself individual stressors. Listwise deletion was used to account
anytime in the last year?) was taken from the revised for missing values. Variables were entered into the model
Diagnostic Interview for Borderlines (Zanarini et al., 1989). predicting each symptom as follows: Block 1: demographic
Items 3 and 4 (adapted from Plöderl et al., 2011) asked variables (age, gender, ethnicity); Block 2: childhood adver-
participants about suicidality and non-suicidal self-injury, for sity (childhood trauma, childhood deprivation); Block 3:
example, ‘‘I hurt/harmed myself but I knew that I would not economic adversity variables (living conditions, financial
have died from this’’. stress); Block 4: discrimination variables (ethnicity, gender,
sexuality, religion/beliefs, disability, cybervictimisation);
Results Block 5: social identity variables; Block 6: loneliness;
Block 7: academic stressors (relative performance, expect-
Extent of mental health issues
ations, assessment, teaching, facilities).
All analyses were conducted using SPSS version 22 software
(Nie et al., 1970). As shown in Figure 1, using the published
criteria for moderate anxiety (10–14; Spitzer et al., 2006) and Depression
depression (10–14; Kroenke & Spitzer, 2002), the proportion As shown in Table 2, at Block 1 the demographic variables
of students above these cuts off was 42.2% for anxiety, and explained a significant portion of variance in depression.
25.0% for depression and 22.2% for comorbid depression and While age and ethnicity were unrelated to depression, women
anxiety. Using the more stringent criteria (GAD-7: 15–21, reported significantly higher levels of depression than men,
PHQ-9: 15–27), 20.9% met criteria for severe anxiety, 11.3%  ¼ 0.09, p ¼ 0.014. At Block 2, the childhood adversity
met criteria for severe depression and 9.0% met the severe variables contributed significantly to the model. Childhood
criteria for both. No cut-offs are available for the paranoia deprivation was unrelated to depression, while having more
scale. experiences of childhood trauma was significantly associated
In total, 18.9% of students reported suicidal thoughts with higher depression scores,  ¼ 0.27, p50.001. At Block
during the last 12 months and 20.1% reported that, at some 3, the inclusion of the economic adversity measures
time in their lives, they had self-harmed. Of the entire sample, contributed significantly to the model. Both higher levels of
12.1% reported a lifetime suicide attempt. From Table 1, it financial stress ( ¼ 0.25, p50.001) and poor living condi-
can be seen that the majority of these incidents were regarded tions ( ¼ 0.09, p ¼ 0.012) were associated with higher
levels of depression. At Block 4, the discrimination variables
100 explained a unique and significant portion of variance in
90 depression. Experiencing higher levels of discrimination
% of students meeting threshold

based on one’s disabilities ( ¼ 0.10, p ¼ 0.010) or sexuality

80 Anxiety ( ¼ 0.14, p50.001) predicted higher levels of depression, as
70 Depression did more experiences of cybervictimisation,  ¼ 0.09,
60 Both p ¼ 0.013. All other discrimination variables were unrelated
to depression. At Block 5, the social capital variables
50 contributed significantly to the model. This was primarily
40 due to the inclusion of university friendship group identity,
which was the only significant predictor of depression at this
30
block,  ¼ 0.26, p50.001. At Block 6, loneliness also
20 contributed significantly to the model. Students who reported
10 feeling more lonely also reported higher levels of depression
( ¼ 0.52, p50.001).
0 Finally, at Block 7, academic stressors also contributed
Moderate symptoms Severe symptoms significantly to the model. Both relative performance
Figure 1. Proportion of students in the sample meeting the criteria for ( ¼ 0.14, p50.001) and assessment stress ( ¼ 0.17,
moderate and severe mental health symptoms. p50.001) were associated with higher levels of depression.
DOI: 10.1080/09638237.2018.1437608 Social determinants of student mental health 5

No other academic stressors significantly predicted financial stress ( ¼ 0.25, p50.001) and poor quality accom-
depression. modation ( ¼ 0.10, p ¼ 0.004) were associated with higher
levels of anxiety. At Block 4, the discrimination variables
Anxiety explained a significant portion of variance in anxiety.
As reported in Table 3, at Block 1 the demographic variables Experiencing higher levels of cybervictimisation ( ¼ 0.11,
explained a significant portion of variance in anxiety. Age was p ¼ 0.003) and discrimination based on one’s disabilities
unrelated to anxiety; however, women reported significantly ( ¼ 0.11, p ¼ 0.005) was associated with higher levels of
higher levels of anxiety compared to men ( ¼ 0.11, anxiety. All other discrimination variables were not significant
p ¼ 0.002), and white students reported higher levels of predictors of anxiety. At Block 5, the social capital variables
anxiety compared to Black and Minority Ethnic students, contributed significantly to the model. Identifying more
 ¼ 0.10, p ¼ 0.009. At Block 2, childhood adversity strongly with a friendship group was associated with lower
contributed significantly to the model predicting anxiety. levels of anxiety ( ¼ 0.25, p50.001) and identifying more
Childhood deprivation was unrelated to anxiety; however, strongly with England was also associated with lower levels of
having more experiences of childhood trauma was significantly anxiety ( ¼ 0.10, p ¼ 0.017). At Block 6, loneliness
associated with higher anxiety scores,  ¼ 0.27, p50.001. At contributed significantly to the model. Feeling lonely was
Block 3, the inclusion of the economic adversity measures strongly associated with higher levels of anxiety ( ¼ 0.50,
contributed significantly to the model. Both higher levels of p50.001). At Block 7, academic stressors explained a

Table 1. Types of self-harm among students reporting suicidal thoughts.

Cumulative %
Frequency % of total sample % of valid reports of valid reports
I only thought seriously about hurting/harming myself 16 1.5 12.2 12.2
I had everything prepared but did not hurt/harm myself 22 2.1 16.8 29.0
I stopped hurting/harming myself in the last second. I knew 15 1.4 11.5 40.5
that it would not have been lethal
I hurt/harmed myself, but I knew that I would not have died 51 4.8 38.9 79.4
from this
I hurt/harmed myself and I knew that I would die from this 27 2.5 20.6 100.0
Total 131 12.3 100.0

Table 2. Hierarchical regression analysis of stressors predicting depression.

Block 1 Block 2 Block 3 Block 4 Block 5 Block 6 Block 7

Gender (F) 0.09* 0.07 0.03 0.03 0.05 0.07* 0.04
Age 0.05 0.09* 0.09* 0.08* 0.10** 0.06* 0.05
Ethnicity (BME) 0.07 0.05 0.01 0.02 0.01 0.01 0.01
Childhood trauma – 0.27*** 0.23*** 0.21*** 0.21*** 0.18*** 0.18***
Childhood deprivation – 0.03 0.00 0.02 0.08* 0.11** 0.10**
Living conditions – – 0.09* 0.08* 0.07* 0.03 0.02
Financial stress – – 0.25*** 0.23*** 0.22*** 0.16*** 0.10**
Ethnicity discrimination – – – 0.00 0.04 0.04 0.05
Gender discrimination – – – 0.00 0.01 0.00 0.00
Sexuality discrimination – – – 0.14*** 0.12** 0.09** 0.10**
Religion discrimination – – – 0.05 0.06 0.06 0.06
Disability discrimination – – – 0.10* 0.11** 0.07* 0.05
Cyberbullying – – – 0.09* 0.08* 0.06* 0.05
University identity – – – – 0.04 0.05 0.01
Country identity – – – – 0.01 0.00 0.02
Online identity – – – – 0.00 0.02 0.02
Liverpool identity – – – – 0.02 0.04 0.00
English identity – – – – 0.05 0.04 0.05
University friends identity – – – – 0.26*** 0.00 0.00
Loneliness – – – – – 0.52*** 0.42***
Academic expectations – – – – – – 0.01
Relative performance – – – – – – 0.14***
Assessment stress – – – – – – 0.17***
Teaching-related stress – – – – – – 0.04
Facilities stress – – – – – – 0.01
Fchange 4.02** 30.29*** 29.40*** 6.44*** 12.46*** 199.68*** 14.14***
R2change 0.02 0.08 0.07 0.04 0.08 0.16 0.05
R2adjusted 0.01 0.09 0.16 0.20 0.27 0.43 0.48

*p50.05.
**p50.01.
***p50.001.
6 J. C. McIntyre et al. J Ment Health, Early Online: 1–10

Table 3. Hierarchical regression analysis of stressors predicting anxiety.

Block 1 Block 2 Block 3 Block 4 Block 5 Block 6 Block 7

Gender (F) 0.11** 0.09* 0.06 0.04 0.06 0.08** 0.05
Age 0.01 0.05 0.05 0.04 0.06 0.03 0.01
Ethnicity (BME) 0.10** 0.08* 0.05 0.04 0.04 0.01 0.01
Childhood trauma – 0.27*** 0.23*** 0.20*** 0.20*** 0.18*** 0.17***
Childhood deprivation – 0.02 0.01 0.03 0.08* 0.11** 0.09**
Living conditions – – 0.10** 0.09** 0.08* 0.04 0.04
Financial stress – – 0.25*** 0.24*** 0.23*** 0.17*** 0.10**
Ethnicity discrimination – – – 0.00 0.03 0.03 0.05
Gender discrimination – – – 0.03 0.03 0.02 0.01
Sexuality discrimination – – – 0.06 0.04 0.00 0.03
Religion discrimination – – – 0.06 0.07 0.08* 0.07*
Disability discrimination – – – 0.11** 0.11** 0.08* 0.06
Cyberbullying – – – 0.11** 0.10** 0.07* 0.05
University identity – – – – 0.02 0.02 0.01
Country identity – – – – 0.03 0.03 0.02
Online identity – – – – 0.03 0.05 0.03
Liverpool identity – – – – 0.05 0.07 0.04
English identity – – – – 0.10* 0.09* 0.09*
University friends identity – – – – 0.25*** 0.01 0.01
Loneliness – – – – – 0.50*** 0.39***
Academic expectations – – – – – – 0.03
Relative performance – – – – – – 0.03
Assessment stress – – – – – – 0.23***
Teaching-related stress – – – – – – 0.08
Facilities stress – – – – – – 0.04
Fchange 5.66** 28.52*** 32.50*** 4.82*** 11.91*** 185.91*** 20.53***
R2change 0.02 0.07 0.08 0.03 0.08 0.15 0.07
R2adjusted 0.02 0.09 0.17 0.19 0.26 0.42 0.49

*p50.05.
**p50.01.
***p50.001.

significant amount of variance in anxiety. Assessment stress lonely was associated with higher levels of paranoia ( ¼ 0.46,
was the only significant predictor of anxiety ( ¼ 0.23, p50.001). At Block 7, academic stressors also contributed
p50.001). Students who felt more stressed about their significantly to the model. However, no individual academic
assessment tasks reported higher levels of anxiety. stressor significantly predicted paranoia.

Paranoia Relationships between social identification and

As shown in Table 4, at Block 1 the demographic variables as a mental health
set explained a significant portion of the variance in paranoia. Descriptive statistics and bivariate correlations between
However, no individual demographic predictors reached identities and symptoms are reported in Table 5. To test the
significance. At Block 2, the childhood adversity variables relationship between identification and mental health we ran a
contributed significantly to the model. While there was no series of multiple regressions. The five different types of
effect of childhood deprivation, having more experiences of social identification (identification with: England, university
childhood trauma was significantly associated with higher city, university, online community, university friends) were
paranoia scores,  ¼ 0.28, p50.001. At Block 3, the inclusion entered into the models as predictors of each mental health
of the economic adversity measures contributed significantly symptom.
to the model. Both lower quality accommodation ( ¼ 0.09,
p ¼ 0.013) and higher levels of financial stress ( ¼ 0.16,
Social identity and depression
p50.001) were associated with higher levels of paranoia. At
Block 4, the discrimination variables explained a significant Together, the five social identity variables accounted for 10%
portion of variance in paranoia. Reporting the experience of of the variance in depression scores. Results of the multiple
higher levels of cybervictimisation ( ¼ 0.22, p50.001) and regression predicting depression revealed that stronger iden-
discrimination based on one’s sexuality ( ¼ 0.11, p ¼ 0.004) tification with England ( ¼ 0.09, p ¼ 0.006) and university
were each associated with higher levels of paranoia. All other friends ( ¼ 0.28, p50.001) predicted lower levels of
discrimination variables were not significant predictors of depression. University identification also marginally pre-
paranoia in this sample. At Block 5, the social capital variables dicted lower depression ( ¼ 0.07, p ¼ 0.050).
contributed significantly to the model. Identifying more Unexpectedly, identification with university city was asso-
strongly with a friendship group was associated with lower ciated with higher levels of depression in the model,  ¼ 0.07,
levels of paranoia ( ¼ 0.21, p50.001). At Block 6, lone- p ¼ 0.049. Identification with university friends was clearly
liness contributed significantly to the model. Feeling socially the most important predictor of lower depression, explaining
DOI: 10.1080/09638237.2018.1437608 Social determinants of student mental health 7
Table 4. Hierarchical regression analysis of stressors predicting paranoia.

Block 1 Block 2 Block 3 Block 4 Block 5 Block 6 Block 7

Gender (F) 0.07 0.04 0.02 0.01 0.00 0.02 0.00
Age 0.07 0.11** 0.12** 0.09* 0.11** 0.07* 0.06
Ethnicity (BME) 0.03 0.01 0.01 0.03 0.02 0.00 0.00
Childhood trauma – 0.28*** 0.25*** 0.20*** 0.20*** 0.18*** 0.18***
Childhood deprivation – 0.05 0.03 0.02 0.01 0.04 0.03
Living conditions – – 0.09* 0.08* 0.07* 0.04 0.03
Financial stress – – 0.16*** 0.13*** 0.12** 0.07* 0.03
Ethnicity discrimination – – – 0.06 0.04 0.04 0.02
Gender discrimination – – – 0.03 0.05 0.04 0.03
Sexuality discrimination – – – 0.11** 0.10** 0.07* 0.08*
Religion discrimination – – – .01 0.00 0.01 0.01
Disability discrimination – – – 0.01 0.01 0.04 0.06
Cyberbullying – – – 0.22*** 0.22*** 0.20*** 0.18***
University identity – – – – 0.05 0.06 0.04
Country identity – – – – 0.02 0.01 0.01
Online identity – – – – 0.02 0.00 0.01
Liverpool identity – – – – 0.08 0.10* 0.08*
English identity – – – – 0.02 0.02 0.01
University friends identity – – – – 0.21*** 0.03 0.02
Loneliness – – – – – 0.46*** 0.40***
Academic expectations – – – – – – 0.05
Relative performance – – – – – – 0.01
Assessment stress – – – – – – 0.06
Teaching-related stress – – – – – – 0.08
Facilities stress – – – – – – 0.04
Fchange 2.73* 34.18*** 13.77*** 10.98*** 6.45*** 145.15*** 5.19***
R2change 0.01 0.09 0.03 0.08 0.04 0.13 0.02
R2adjusted 0.01 0.09 0.13 0.19 0.23 0.36 0.38

*p50.05.
**p50.01.
***p50.001.

Table 5. Bivariate correlations between social identity and mental health variables.

Variable M SD Depression (PHQ-9) Anxiety (GAD-7) Paranoia (PaDS-5)

Country of birth ID 15.88 4.18 0.19*** 0.16*** 0.16***
English ID 14.48 4.43 0.17*** 0.17*** 0.13***
University city ID 15.11 3.86 0.12*** 0.09** 0.11***
University ID 15.10 3.86 0.19*** 0.15*** 0.15***
University friends ID 15.37 4.17 0.31*** 0.27*** 0.25***
Online ID 16.02 4.08 0.09** 0.06* 0.08**
Loneliness 19.49 5.25 0.58*** 0.54*** 0.53***

*p50.05.
**p50.01.
***p50.001.

5% of the variance in depression after taking into account the Social identity and paranoia
effects of all other social identities, sr2 ¼ 0.05. The five measures of social identity, together, accounted for
7% of the variance in paranoia scores. University friends
Social identity and anxiety
identification emerged as the only significant predictor of
Combined, the five social identity variables explained 9% paranoid symptoms,  ¼ 0.23, p50.001. Stronger identifi-
of the variance in anxiety scores. When examining the cation with university friends was associated with lower
individual predictors, stronger identification with England paranoia. University friends identification explained 4% of
( ¼ 0.11, p ¼ 0.001) and university friends ( ¼ 0.26, the variance in paranoia scores, controlling for all other social
p50.001) predicted lower levels of anxiety, while stronger identities, sr2 ¼ 0.04.
identification with university city predicted higher levels of
anxiety ( ¼ 0.09, p ¼ 0.016). Identification with university
friends was the strongest independent predictor of lower Mediation models
anxiety, explaining 5% of the variance in anxiety after Given that the previous analyses suggested that identifying
taking into account the effects of all other social identities, with university friends was more protective than the other
sr2 ¼ 0.05. measured identities, we investigated whether this effect is best
8 J. C. McIntyre et al. J Ment Health, Early Online: 1–10

understood by the notion that when people identify with a causal pathway, as does recent evidence that interventions
university friendship group they feel less lonely. The medi- designed to foster group memberships improve both social
ation models controlled for the effects of age, ethnicity and connectedness and mental health (Haslam et al., 2016).
gender. We first tested whether friendship group identification Despite this, the proposed causal relationships suggested here
and loneliness were correlated. If the correlation was too high should be interpreted with some caution given the cross-
(i.e.40.70), then it would not be possible to continue with the sectional nature of the data. Indeed, it is plausible that the
mediation because it would be likely that the scales were observed relationships are cyclical in that academic stressors
measuring the same construct. The two variables correlated at lead to poor mental health, which in turn impacts on a
0.52 so we proceeded with the mediation analyses. student’s ability to study, leading to stress. Moreover, mental
Mediations were conducted in SPSS using the PROCESS health symptoms can make it more difficult for people to join
extension (Hayes, 2012). Indirect effects and associated and identify with social groups. This suggestion is supported
confidence intervals were calculated via bootstrapping with by recent longitudinal work with high-school students that
1000 resamples. found belonging to fewer social groups leads to worse mental
Loneliness mediated the relationship between friendship health over time, and that worse mental health in turn predicts
group identification and mental health for depression joining fewer groups (Miller et al., 2017). A further limitation
(IE ¼ 0.30, CI[0.34, 0.27]), anxiety (IE ¼ 0.42, of the study is that our sampled population attended a single
CI[0.48, 0.36]) and paranoia (IE ¼ 0.36, CI[0.41, British university. Thus, it is not clear whether the observed
0.31]). Effect size estimates indicated that the completely effects would permeate geographic and cultural boundaries.
standardized indirect effects were 0.30 for depression, Finally, the anxiety and depression measures specifically
0.28 for anxiety and 0.28 for paranoia. Of note, we also assessed distress over a two-week period, which meant we
ran mediation analyses with identification and loneliness were unable to examine the impact of stressors on chronic
switched in the model to test whether feelings of loneliness mental health conditions.
may make it more difficult for people to form friendships and The findings suggest that academic stressors place heavy
friendship group identities. None of these models were demands on psychological resources, and that students are at
significant, supporting the hypothesis that identity affects high risk of mental health issues. However, campus environ-
feelings of loneliness rather than vice versa. ments also represent opportunities for students to develop
meaningful social connections that are beneficial to their
mental health. By increasing awareness of the importance of
Discussion
social groups and by facilitating the organisation of social
We conducted a comprehensive mental health survey of events and communities of interest, universities may be able
students attending a large university in northern England. We to enhance the efficacy of existing social connection programs
aimed to assess the prevalence of severe mental distress in the to achieve improvements in student mental health. Given our
student population and to identify key stressors and protective findings, it will be important to focus these interventions on
factors. Overall, there were high rates of mental distress in our smaller group connections (e.g. seminar groups or degree
sample, with particularly elevated rates of clinically severe programs) as opposed to superordinate groups such as
anxiety (21%) and depression (11%), in addition to high faculties or universities. Universities and students would
comorbidity (9%). One-fifth of students reported suicidal also benefit by reducing stress related to assessment and
thoughts and 2% reported a determined suicide attempt. performance. One practical measure might involve facilitating
Contextual factors including childhood adversity, economic the formation of study groups that will provide opportunities
deprivation, discrimination and loneliness all contributed to for students to connect with people facing similar academic
poor mental health. Of note, our complete set of social demands, and simultaneously foster group memberships.
determinants were able to explain nearly half of the variance The findings are consistent with the JD-R model of mental
in both depression and anxiety, and over a third of the health insofar as we found that students are subjected to a
variance in paranoia. Of the academic variables, assessment range of academic and non-academic demands that contribute
stress and relative performance were most predictive of to symptomology. However, universities also provide
mental distress. resources such as opportunities to form social connections
Feelings of loneliness consistently emerged as the and new identities, and also to reduce stressors related to
strongest predictor of poor mental health. Planned analyses concerns over performance and assessment tasks. Our
of the social identity variables suggested that university findings are consistent with work by Pluut et al. (2015) who
friends are the most important social group with whom to also applied the JD-R model to students, finding that
identify in order to protect against depression, anxiety and academic stressors contributed to low well-being and per-
paranoia. Follow-up mediation analyses confirmed that iden- formance, while social support enhanced satisfaction and
tifying with university friends alleviated these symptoms by performance.
decreasing feelings of loneliness. The findings are consistent Consistent with previous work (Hill, 2003; Morrison et al.,
with a growing body of literature on social cure models of 2003; Shevlin et al., 2007; Varese et al., 2012), the findings
mental health (Cruwys et al., 2013, 2014, 2015; McIntyre suggest that the negative psychological effects of childhood
et al., 2016; Sani et al., 2017), but additionally demonstrate trauma flow through to young adulthood. Ensuring univer-
that bonds formed with fellow students at university are sities are aware of the potential impact of background
particularly important for maintaining mental health. stressors, which, when combined with academic stressors,
The reverse mediations provided evidence for the proposed may place students at particularly high risk of severe
DOI: 10.1080/09638237.2018.1437608 Social determinants of student mental health 9

symptoms is particularly important according to the present Haslam C, Cruwys T, Haslam SA, et al. (2016). Groups 4 health:
Evidence that a social-identity intervention that builds and strengthens
findings. Indeed, it may be that measures to proactively social group membership improves mental health. J Affect Disord,
identify those most at risk of distress by making confidential 194, 188–95.
enquiries into the area of previous trauma is something that Haslam S, Jetten J, Postmes T, Haslam C. (2009). Social identity, health
universities should consider as a matter of routine. This and well-being: An emerging agenda for applied psychology. Appl
Psychol, 58, 1–23.
information could then be provided, with the consent of the Hayes AF. (2012). PROCESS: A versatile computational tool for
student, to counseling and disability services for information observed variable mediation, moderation, and conditional process
if a student presents with difficulties. Of course, the modeling. University of Kansas, KS.
acquisition of this information would need to be handled in Hill J. (2003). Childhood trauma and depression. Curr Opin Psychiatry,
16, 3–6.
a sensitive manner and with full disclosure to students Hinduja S, Patchin JW. (2010). Bullying, cyberbullying, and suicide.
regarding the use of information. Arch Suicide Res, 14, 206–21.
Our work provides a comprehensive overview of the types Hurst CS, Baranik LE, Daniel F. (2013). College student stressors: A
of stressors most likely to impact student mental health. We review of the qualitative research. Stress Health, 29, 275–85.
Jetten J, Haslam C, Alexander SH. (2012). The social cure: Identity,
also identify ways that universities can furnish students with health and well-being. New York: Psychology Press.
social and psychological resources to minimise distress. As Kessler RC, Amminger GP, Aguilar-Gaxiola S, et al. (2007). Age of
student numbers continue to rise, there is likely to be onset of mental disorders: A review of recent literature. Curr Opin
increasing pressure on institutions to ensure students’ mental Psychiatry, 20, 359.
Kohn JP, Frazer GH. (1986). An academic stress scale: Identification and
health and well-being. We hope that this work will usefully rated importance of academic stressors. Psychol Rep, 59, 415–26.
inform future policies and interventions designed to achieve Kroenke K, Spitzer RL. (2002). The PHQ-9: A new depression
this aim. diagnostic and severity measure. Psychiatr Ann, 32, 509–15.
Lee RM, Robbins SB. (1998). The relationship between social connect-
edness and anxiety, self-esteem, and social identity. J Couns Psychol,
Declaration of interest
45, 338–45.
No potential conflict of interest was reported by the authors. Lester D. (2014). College student stressors, depression, and suicidal
ideation. Psychol Rep, 114, 293–6.
JCM and RC are supported by funding from the National Major B, Kaiser CR, O’brien LT, McCoy SK. (2007). Perceived
Institute for Health Research Collaboration for Leadership in discrimination as worldview threat or worldview confirmation:
Applied Health Research and Care – North West Coast. The Implications for self-esteem. J Pers Soc Psychol, 92, 1068.
views expressed are those of the authors and not necessarily McIntyre JC, Elahi A, Bentall RP. (2016). Social identity and psychosis:
Explaining elevated rates of psychosis in migrant populations. Soc
those of the NHS, the NIHR or the Department of Health. Personal Psychol Compass, 10, 619–33.
McIntyre JC, Wickham S, Barr B, Bentall RP. (2017). Social identity and
ORCID psychosis: Associations and psychological mechanisms. Schizophr
Bull, sbx110. Available from: https://doi.org/10.1093/schbul/sbx110
Jason C. McIntyre http://orcid.org/0000-0002-5601-524X Melo S, Corcoran R, Shryane N, Bentall RP. (2009). The persecution
Rhiannon Corcoran http://orcid.org/0000-0001-8900-9199 and deservedness scale. Psychol Psychother: Theory, Res Pract, 82,
247–60.
Miller K, Wakefield JRH, Sani F. (2017). On the recirprocal effects between
multiple group identifications and mental health: A longitudinal study of
References Scottish adolescents. Br J Clin Psychol, 56, 357–71.
Andrews B, Wilding JM. (2004). The relation of depression and anxiety Morrison AP, Frame L, Larkin W. (2003). Relationships between
to life-stress and achievement in students . Br J Psychol, 95, 509–21. trauma and psychosis: A review and integration. Br J Clin Psychol, 42,
Ang RP, Huan VS. (2006). Academic expectations stress inventory: 331–53.
Development, factor analysis, reliability, and validity. Educ Psychol Nie NH, Bent DH, Hull CH. (1970). SPSS: Statistical package for the
Meas, 66, 522–39. social sciences. New York: McGraw-Hill.
Bakker AB, Demerouti E. (2007). The job demands-resources model: Pidd K. (2004). The impact of workplace support and identity on training
State of the art. J Manag Psychol, 22, 309–28. transfer: A case study of drug and alcohol safety training in Australia.
Cooke R, Barkham M, Audin K, et al. (2004). Student debt and its Int J Train Dev, 8, 274–88.
relation to student mental health. J Further High Educ, 28, 53–66. Plöderl M, Kralovec K, Yazdi K, Fartacek R. (2011). A closer look at
Cruwys T, Dingle GA, Haslam C, et al. (2013). Social group self-reported suicide attempts: false positives and false negatives.
memberships protect against future depression, alleviate depression Suicide Life-Threat Behav, 41, 1–5.
symptoms and prevent depression relapse. Soc Sci Med, 98, 179–86. Pluut H, Curşeu PL, Ilies R. (2015). Social and study related stressors
Cruwys T, Haslam SA, Dingle GA, et al. (2014). Depression and social and resources among university entrants: Effects on well-being and
identity: An integrative review. Pers Soc Psychol Rev, 18, 215–38. academic performance. Learn Individ Differ, 37, 262–8.
Cruwys T, South EI, Greenaway KH, Haslam SA. (2015). Social identity Ross SE, Niebling BC, Heckert TM. (1999). Sources of stress among
reduces depression by fostering positive attributions. Soc Psychol college students. Soc Psychol, 61, 841–6.
Person Sci, 6, 65–74. Sani F, Wakefield JR, Herrera M, Zeybek A. (2017). On the association
Department for Communities and Local Government. (2015). English between greater family identification and lower paranoid ideation
Housing Survey. Available from: https://www.gov.uk/government/ among non-clinical individuals: Evidence from cypriot and spanish
collections/english-housing-survey. students. J Soc Clin Psychol, 36, 396–418.
Doosje B, Ellemers N, Spears R. (1995). Perceived intragroup variability Shevlin M, Dorahy D, Martin J, Adamson G. (2007). Trauma and
as a function of group status and identification. J Exp Soc Psychol, 31, psychosis: an analysis of the national comorbidity survey. Am J
410–36. Psychiatry, 164, 166–9.
Felitti VJ, Anda RF, Nordenberg D, et al. (1998). Relationship of Smith BN, Vaughn RA, Vogt D, et al. (2013). Main and interactive
childhood abuse and household dysfunction to many of the leading effects of social support in predicting mental health symptoms in men
causes of death in adults: The Adverse Childhood Experiences (ACE) and women following military stressor exposure. Anxiety Stress
study. Am J Prevent Med, 14, 245–58. Coping, 26, 52–69.
Hagerty BM, Williams A. (1999). The effects of sense of belonging, Spitzer RL, Kroenke K, Williams JB, Löwe B. (2006). A brief measure
social support, conflict, and loneliness on depression. Nurs Res, 48, for assessing generalized anxiety disorder: The GAD-7. Arch Intern
215–9. Med, 166, 1092–7.
10 J. C. McIntyre et al. J Ment Health, Early Online: 1–10

Swartzman S, Booth JN, Munro A, Sani F. (2017). Posttraumatic stress Wakefield JR, Bickley S, Sani F. (2013). The effects of identification
disorder after cancer diagnosis in adults: A meta-analysis. Depress with a support group on the mental health of people with multiple
Anxiety, 34, 327–39. sclerosis. J Psychosom Res, 74, 420–6.
Tajfel H, Turner JC. (1979). An integrative theory of intergroup conflict. Wickham S, Shevlin M, Bentall RP. (2013). Development and validation
Soc Psychol Intergr Relat, 33, 74. of a measure of perceived relative deprivation in childhood. Person
Thomas J, Bentall RP, Hadden L, O’Hara L. (2017). Ethnic identity and Individ Differ, 55, 399–405.
paranoid thinking: Implicit out-group preference and language YouGov. (2016). Available from: https://d25d2506sfb94s.cloudfront.net/
dominance predict paranoia in Emirati women. J Behav Ther Exp cumulus_uploads/document/obtomdatp4/Survey_Results.pdf [last
Psychiatry, 56, 122–8. . accessed 15 Jun 2017].
University Central Admission Service. (2015). End of cycle report 2015: Youth Risk Behavior Surveilance System (YRBSS). (2015).
UCASE analysis and research. Available from: https://www.cdc.gov/healthyyouth/data/yrbs/index.
Varese F, Smeets F, Drukker M, et al. (2012). Childhood adversities increase htm.
the risk of psychosis: A meta-analysis of patient-control, prospective-and Zanarini MC, Gunderson JG, Frankenburg FR, Chauncey DL.
cross-sectional cohort studies. Schizophr Bull, 38, 661–71. (1989). The revised diagnostic interview for borderlines: dis-
Viswesvaran C, Sanchez JI, Fisher J. (1999). The role of social support in the criminating BPD from other axis II disorders. J Personality
process of work stress: A meta-analysis. J Vocat Behav, 54, 314–34. Disorders, 3, 10–8.

Supplementary material available online