Suicide and Life-Threatening Behavior 1

© 2018 The American Association of Suicidology

DOI: 10.1111/sltb.12437

An Empirical Test of the Three-Step Theory

of Suicide in U.K. University Students
KATIE DHINGRA, PHD, E. DAVID KLONSKY, PHD, AND VOJNA TAPOLA, PHD

The purpose of this study was to further our understanding of how indi-
viduals move down the pathway from first thinking about suicide to ultimately
attempting to take their own lives by empirically testing the Three-Step Theory
(3ST) in a sample of university students (n = 665). Results largely support the
theory’s central propositions. First, an interactive model of pain and hopeless-
ness accounted for substantial variance in suicidal desire. This result replicated
in both men and women, and across age groups (i.e., below 35 and at or above
35). Also, as predicted, connectedness was protective against ideation in those
high on both pain and hopelessness. However, contrary to our prediction, con-
nectedness was similarly protective among everyone else. Finally, suicide capac-
ity predicted suicide attempt history over and above current and lifetime suicide
ideation. These findings provide further support to the 3ST.

Suicidal behavior is a significant behavioral worldwide. However, a recent Cochrane

health concern among university students review found little evidence that these pro-
and is the second leading cause of death grammes lead to sustained reductions in
among this demographic group (World suicidality (Harrod, Goss, Stallones, & Di
Health Organization, 2016). Five to 35 per- Guiseppi, 2014).
cent of undergraduate students report hav- To effectively intervene and ulti-
ing considered suicide in the past mately prevent suicidal behavior in univer-
12 months (Robins & Fiske, 2009; Wong, sity students, we need to better understand
Brower, & Zucker, 2011), while up to 11% suicide and those at risk. Unfortunately,
report having made a suicide attempt although our understanding of suicide risk
(Eisenberg, Hunt, & Speer, 2013). Recent factors has improved considerably over the
figures show that there were 130 suicide past few decades (O’Connor & Nock,
deaths in England and Wales among stu- 2014), there remain significant gaps in
dents aged 18 or above in 2014 (ONS, knowledge that hinder our ability to predict
2016). In response to these concerning with sensitivity and specificity which indi-
statistics, a diverse array of interventions viduals are most at risk. Especially lacking
(e.g., educational curricula, means restric- is an understanding of how and why people
tion, and safety planning) has been devel- move along the pathway to suicide: from
oped and implemented in universities onset of suicide ideation, to developing a

KATIE DHINGRA, Leeds Beckett Univer- No financial support was received for this
sity, Leeds, United Kingdom; E. DAVID KLON- research.
SKY, Department of Psychology, University of Address correspondence to Katie Dhingra,
British Columbia, Vancouver, BC, Canada; Leeds Beckett University, Calverley Building,
VOJNA TAPOLA, University of Eastern Finland, Portland Way, Leeds, LS1 3HE, United King-
Joensuu, Finland. dom; E-mail: [email protected]
2 3ST UK UNIVERSITY

plan and intention, to ultimately attempting what brings about suicide ideation. Second,
to take their own lives (Klonsky & May, connectedness prevents suicide ideation
2014). This knowledge gap is critical, as from escalating in intensity in those at risk
considerable evidence suggests that out of (i.e., those experiencing both pain and
the relatively high percentage of individuals hopelessness). Finally, strong suicide idea-
who contemplate suicide, only one-third tion leads to a suicide attempt if, and only
make a suicide attempt (Nock et al., 2008). if, one has the means and capacity (disposi-
Moreover, recent studies continue to find tional, acquired, and practical) to make an
that traditionally cited risk factors for sui- attempt.
cide—including most mental disorders and Different aspects of the 3ST have
hopelessness—predict suicide ideation, but been empirically tested. For instance, stud-
fail to distinguish suicide attempters from ies of suicide attempters have found that
ideators (Dhingra, Boduszek, & O’Connor, suicide attempts are most often motivated
2015; May & Klonsky, 2016). by pain and hopelessness (May & Klonsky,
To address this very issue, studies are 2013). In addition, a recent study of 910
increasingly employing an “ideation-to- U.S. adults (Klonsky & May, 2015) found
action” framework (Klonsky, Qui, & Saffer, that suicide ideation was negligible in those
2017). This framework views the develop- low on both pain and hopelessness, and
ment of suicide ideation and the progres- those either high on pain or high on hope-
sion from ideation to potentially lethal lessness. In contrast, suicide ideation was
attempts as distinct processes with distinct elevated in the subgroup high on both pain
explanations and predictors. A new theory and hopelessness. Additionally, connected-
of suicide positioned within the ideation-to- ness protected against escalation of ideation
action framework is the Three-Step Theory in those high on both pain and hopeless-
(3ST; Klonsky & May, 2015). The 3ST ness, but was minimally related to ideation
(Figure 1) makes three central claims. First, among everyone else. A number of studies
the combination of pain and hopelessness is have also supported the importance of the
capability for suicide in enhancing the asso-
ciation between suicidal desire and suicidal
behavior (e.g., Dhingra et al., 2015).
Finally, Klonsky and May (2015) found that
dispositional, acquired, and practical con-
tributors to the capacity for suicide each
predicted suicide attempt history over and
above current and past suicide ideation.
Thus, based on this preliminary research, it
appears that the 3ST has great potential to
advance suicide research and prevention.

This Study

Few studies have examined the central

hypotheses postulated by the 3ST. There-
fore, the present research aimed to empiri-
cally test the 3ST in a large sample of U.K.
university students. It was predicted that (a)
pain and hopelessness will interact to predict
current suicide ideation and will predict sui-
Figure 1. Illustration of the Three-Step Theory (3ST) cide ideation better than a comparison
of suicide (Klonsky & May, 2015). model consisting of high burdensomeness
DHINGRA ET AL. 3

and low belongingness, (b) connectedness “yes” to the following question taken from
will protect against the escalation of suicide The Self-Injurious Thoughts and Beha-
ideation among those high on both pain and viours Interview (SITBI; Nock, Holmberg,
hopelessness, and (c) suicide capacity will Photos, & Michel, 2007): “Have you ever
distinguish lifetime suicide attempters from made an actual attempt to kill yourself in
those with histories of ideation but not which you had at least some intent to die?”
attempts. Should the theory be validated, the The SITBI suicide attempt subscale has
3ST will help generate more parsimonious demonstrated strong interrater reliability
risk assessment tools as well as more tar- (average k = 0.99), test–retest reliability (av-
geted intervention approaches. erage k = 0.70), and construct validity, as
demonstrated by strong relations with other
measures of suicide attempt (k = .65; Nock
METHODS et al., 2007).
Suicidal Desire. Eight items from the
Sample Beck Scale for Suicide Ideation (BSS; Beck
& Steer, 1991) were used to index suicidal
Participants (N = 665) were 475 desire. Item selection and their use as an
female and 190 male university students index of suicidal desire are now explained.
recruited from various faculties from two Some items assess suicidal desire, whereas
U.K. universities. Participants were aged some index other variables such as prepara-
between 17 and 67 years (M = 24.2; tory actions (e.g., “. . . finished or completed
SD = 8.11). Regarding ethnicity, 78.9% of my preparations for committing suicide”) or
participants reported they were Caucasian, perceived capability to make a suicide
12.1% Asian, 3.4% Black, 3.4% Mixed, and attempt (e.g., “I have the courage or ability
2.2% Other. Additionally, most students to commit suicide”). The former is
(83.3%) described their sexual orientation addressed by Steps 1 and 2 of the 3ST,
as heterosexual/straight. whereas, the latter is addressed by Step 3.
Thus, to test Steps 1 and 2 of the 3ST, fac-
Procedure tor analysis was used to identify a subset of
BSS items that index suicidal desire uncon-
The research protocol was reviewed founded by items relevant to preparations
and approved by the institutional ethics or capability. This approach is similar to
panels of both participating universities. previously published factor-analyses that
Participants were recruited via an e-mail distinguished ideation items assessing suici-
invite to participate in a study of suicide. dal desire from those assessing preparation
Within this e-mail, it was made clear to (Beck, Brown, & Steer, 1997; Beck, Kovacs,
potential participants that they did not need & Weissman, 1979).
to have experienced suicidal thoughts and In this study, a factor analysis (princi-
behaviors to take part. Participants were ple-axis factoring) with promax rotation
required to consent before the survey was yielded a 3-factor solution accounting for
presented online. Participation in this study 53.6% of variance. Factor 1, with an eigen-
was voluntary and no inducements or obli- value of 7.2, assessed suicidal desire. This
gations were used. All participants were factor included items 1–7 and 9; each of
debriefed and given phone numbers for these items had loadings above .4 with no
local mental health services. cross-loadings on other factors. This factor
was similar to the suicidal desire factors
Measures identified in previous studies (Beck et al.,
1979, 1997). Thus, items 1–7 and 9 were
Suicide Attempts. Suicide attempts summed to form a suicidal desire scale,
were recorded if a respondent answered which was used for subsequent analyses
4 3ST UK UNIVERSITY

(a = .89). Factors 2 and 3 were less robust degree to which each item is true for them
(eigenvalues of 1.7 and 1.3, respectively), on a 7-point Likert scale.
contained items peripherally related or Suicide Capacity. The 6-item Suicide
unrelated to suicidal desire, and were not Capacity Scale (SCS-3; Klonsky & May,
used in this study. 2015) assesses three contributors to the
Psychological Pain. The 13-item Scale capacity to enact a potentially lethal suicide
of Psychache measured current emotional attempt: Dispositional Capacity (i.e., long-
or mental pain as conceptualized by Shneid- standing pattern of low fear of pain or
man (1993). Items are rated on a 5-point death), Acquired Capacity (i.e., fear of pain
Likert scale. The psychometric properties or death had decreased over time), and
of this measure have been established (Hol- Practical Capacity (i.e., access to and
den, Mehta, Cunningham, & McLeod, knowledge of suicide methods). Items are
2001). In this sample, coefficient alpha was rated on a 7-point Likert scale. The SCS-3
very high (a = .96). has been shown to reliably differentiate sui-
Hopelessness. The Beck Hopelessness cide attempters from suicide ideators
Scale (BHS; Beck, Weissman, Lester, & (Klonsky & May, 2015). Coefficient alpha
Trexler, 1974) is a widely used scale that in the present sample was satisfactory
assesses hopelessness within the past week. (a = .72).
The BHS includes 20 items, which are
rated as true or false. In this sample, coeffi-
cient alpha was excellent (a = .90). RESULTS
Burdensomeness and Belonging-
ness. The 12-item version of the Interper- Descriptive statistics and intercorrela-
sonal Needs Questionnaire (INQ; Van tions for the study variables are presented
Orden, Witte, Gordon, Bender, & Joiner, in Table 1. All variables had skewness and
2008) was used to measure participants’ kurtosis statistics within normal limits (i.e.,
beliefs about the extent to which they feel <|2|). At least one suicide attempt was
connected to others (i.e., belongingness; reported by 24% of participants, and 72.4%
a = .82) and the degree to which they feel reported a lifetime history of suicide idea-
they are a burden to others (i.e., burden- tion. Described below are results addressing
someness; a = .86). Participants indicate the each step of the 3ST.

TABLE 1
Means, Standard Deviations, and Intercorrelations for Study Variables
Correlations

Variable M SD 1 2 3 4 5 6 7

1. Suicidal desire 2.09 3.07 —

2. Suicide attempta 0.25 0.43 .43 —
3. Pain (Psychache) 30.86 13.47 .64 .37 —
4. Hopelessness 6.89 5.30 .67 .31 .65 —
5. Belongingness (Low) 18.64 7.39 .51 .29 .58 .65 —
6. Burdensomeness 21.12 9.96 .67 .36 .65 .66 .58 —
7. SCS-3 25.94 7.46 .39 .33 .33 .24 .26 .33 —

SCS-3 = Suicide Capacity Scale.

All correlations presented are statistically reliable at p < .001.
a
The mean of .25 indicates that 25% of participants had a history of at least one suicide
attempt. Because this variable is dichotomous, point-biserial correlations are reported.
DHINGRA ET AL. 5

Step 1: Pain and Hopelessness Combine Next, we examined whether the inter-
to Predict Suicidal Desire action between pain and hopelessness would
appear consistently across demographic sub-
First, we report the direct effects of groups. As expected, the interaction was
pain and hopelessness on suicidal desire. As statistically reliable in both men (t = 4.24,
expected, both pain (r = .64) and hopelessness p < .001) and women (t = 7.73, p < .001),
(r = .67) exhibited robust correlations with as well as in participants both less than
suicidal desire. Also, as expected, pain and 35 years old (t = 5.91, p < .001) and at or
hopelessness were strongly related (r = .65). >35 years old (t = 3.05, p = .003).
Second, as a direct test of the 3ST’s Finally, we wanted to evaluate Step 1
Step 1, we examined whether pain and of the 3ST in comparison with a highly
hopelessness interacted to predict suicidal cited, well-researched model. We, therefore,
desire. The interaction was significant also examined the variables hypothesized by
(t = 6.42, p < .001), and the full model the Interpersonal Theory of Suicide (Joiner,
accounted for 56% of the variance in suici- 2005; Van Orden et al., 2010) to explain
dal desire. The interaction term itself suicidal desire. This comparison model
explained an additional 3% of variance over specifies that thwarted belongingness and
and above the main effects. However, perceived burdensomeness interact to
because the interaction term correlates explain suicidal desire. Results indicate that
strongly with both main effect terms, the these variables indeed interacted signifi-
3% of unique variance added may not fully cantly to predict suicidal desire (t = 4.12,
capture the interactive nature of pain and p < .001), although the full model including
hopelessness in predicting suicide ideation. the interaction term accounted for 49% of
Thus, to illustrate the potential clinical sig- the variance in suicidal desire—which rep-
nificance of this pattern, median splits were resents 7% less variance explained than Step
used to create low and high subgroups for 1 of the 3ST.
pain and hopelessness. As seen in Figure 2,
suicide ideation is negligible in subgroups Step 2: Connectedness Protects Against
with (a) low pain and hopelessness or (b) Escalation of Suicidal Desire
either high pain or high hopelessness, but is
elevated in the subgroup, (c) reporting both The 3ST hypothesizes that connect-
high pain and high hopelessness. edness can foster a desire to live even among

5.00

4.00
Mean Suicide Desire

3.00

2.00

1.00

0.00
Low P and H Either High P or High H Both High P and High H
Group

Figure 2. Interactive effects of pain (P) and hopelessness (H) on suicidal desire.
6 3ST UK UNIVERSITY

those with both pain and hopelessness, and with histories of suicide attempts (n = 161).
thus protect against higher levels of suicidal As predicted, suicide capacity robustly dis-
desire among this at-risk subgroup. We con- tinguished suicide ideators and attempters:
ducted two analyses to evaluate this hypoth- d = .72, p < .001. Similarly, dispositional
esis. First, we examined whether the (d = .29, p < .005), acquired (d = .48,
relationship of connectedness to suicidal p < .001), and practical d = .87, p < .001)
desire is particularly strong in the subgroup contributors to suicide capacity each distin-
of at-risk participants high on both pain and guished ideators from attempters. Addition-
hopelessness. We utilized the belongingness ally, overall suicide capacity predicted
scale of the INQ to assess connectedness. As suicide attempt history when controlling for
predicted, among the subgroup high on current suicidal desire (partial rpb = .21,
both pain and hopelessness (n = 237), there p < .001). Similar findings were found for
was a robust correlation between connected- practical (partial rpb = .26, p < .001) and
ness and increased suicidal desire (r = .34, acquired (partial rpb = .11, p = .02) contrib-
p < .001). However, contrary to expecta- utors to suicide capacity, but not disposi-
tions, this relationship was not lower among tional contributors (partial rpb = .06,
participants without both pain and hopeless- p = .23). In contrast to overall suicide
ness (n = 396, r = .41, p < .001). capacity—and consistent with the 3ST—
Second, we wanted to test the more pain, hopelessness, thwarted belongingness,
specific hypothesis that suicidal desire is and perceived burdensomeness each failed
buffered in individuals with combined pain to predict suicide attempt status over and
and hopelessness when connectedness above suicidal desire (all point-biserial cor-
exceeds pain. To do this, we standardized relations <.09, all ps > .09).
scores for pain and connectedness, and then
subtracted connectedness scores from pain.
Thus, positive scores indicated that pain DISCUSSION
exceeds connectedness, whereas negative
scores indicated that connectedness exceeds As previously noted, our ability to
pain. If our hypothesis is correct, this differ- better predict and prevent suicide likely
ence score should be a particularly powerful hinges upon a better understanding of the
predictor of suicidal desire in the combined transition from suicidal thoughts to action.
pain and hopelessness subgroup, and less The aim of the present study, therefore,
predictive of suicidal desire in everyone else. was to empirically test the Three-Step The-
As predicted, the correlation of suicidal ory (3ST; Klonsky & May, 2015). In brief,
desire with the pain-connectedness differ- the 3ST proposes that (a) suicide ideation
ence score was strong (r = .46, p < .001) develops due to a combination of high
among those high on both pain and hope- levels of both pain and hopelessness, (b)
lessness. However, contrary to expectations, connectedness is a key protective factor
this correlation was similar to that observed against escalating ideation in those high on
among everyone else (r = .47, p < .001). both pain and hopelessness, and (c) the pro-
gression from suicide ideation to attempts
Step 3: Dispositional, Acquired, and occurs when someone has the capacity to
Practical Contributors to Capacity for face the pain and fear inherent in attempt-
Suicide Differentiate Attempters from ing to end one’s own life. Data drawn from
Ideators our large sample of U.K. university students
provide partial support for the theory.
T-tests were used to compare suicide The first tenet of the 3ST, that sui-
capacity (SCS-3 scores) between (a) partici- cide ideation is driven by the combination
pants with histories of suicide ideation but of pain and hopelessness, was strongly sup-
not attempts (n = 301), and (b) participants ported. These two variables significantly
DHINGRA ET AL. 7

interacted, explaining 56% of the variance components may be missing from existing
in suicidal desire. Moreover, this interaction measures. An alternative explanation for the
was statistically reliable in both men and disparity between our study findings and
women, as well as in participants both those of Klonsky and May (2015) is that the
below age 35 and at or above age 35. Fur- theory could have better explanatory power
thermore, the model including pain, hope- for certain subsets of individuals. This sug-
lessness, and their interaction explained 7% gestion is consistent with recent work using
more variability in suicide ideation than a latent class analysis to distinguish subclasses
comparison model including the variables of individuals with self-injurious thoughts
hypothesized by the Interpersonal Theory and behavior who display different symptom
of Suicide (Joiner, 2005). Not only are patterns and risk trajectories over time
these results consistent with the predictions (Dhingra, Boduszek, & Klonsky, 2016;
of the 3ST, but also with research indicat- Logan, Hall, & Karch, 2011). Future
ing that suicide attempts are more often research testing the 3ST constructs across
motivated by pain and hopelessness than by different subsets of individuals would, there-
any other factor (May & Klonsky, 2013). fore, help to further specify the generaliz-
Partial support for the second tenet— ability of the 3ST to different populations.
that connectedness plays a particularly The third key tenet of the 3ST, that
important protective role against the escala- suicide capacity explains the progression
tion of suicide ideation in those at greatest from suicide ideation to attempts, was sup-
risk for ideation (i.e., those high on both ported. Results indicated that suicide capac-
pain and hopelessness)—was also found. ity robustly distinguished suicide ideators
Specifically, connectedness, as well as the and attempters. This finding is consistent
degree to which participants’ connectedness with Joiner’s (2005) and O’Connor’s (2011)
exceeded their pain, predicted lower suicide emphasis on acquired capability, but also
ideation among those with combined pain supports the relevance of dispositional and
and hopelessness. However, contrary to our practical contributors to suicide capacity.
prediction, and in contrast to the findings However, it is important to note that the
of Klonsky and May (2015), this relation- effect was small and less reliable for disposi-
ship was not lower among participants with- tional contributors and that much variabil-
out combined pain and hopelessness. It is ity in the progression from suicide ideation
worth noting that Klonsky and May (2015) to attempts remain unexplained. Conse-
suggest a broad definition of connectedness quently, it is necessary to explore novel fac-
that can involve connection not only to tors that may contribute to action among
other people, but also to a valued job, pro- ideators (Glenn & Nock, 2014). In addi-
ject, role, interest, or any sense of perceived tion, the measure of suicide capacity used is
purpose or meaning that keeps one invested quite brief and does not cover the full
in living. This is important to highlight as breadth of dispositional, acquired, and prac-
the measures used in the present study tical contributors to capacity. Consequently,
relates to interpersonal connectedness, it is important for future research to
rather than connectedness in a broader develop more comprehensive measures of
sense. Consequently, it is questionable suicide capacity. Finally, and consistent with
whether the construct of connectedness was the 3ST, pain, hopelessness, thwarted
adequately assessed, and it may be that uni- belongingness, and perceived burdensome-
versity students have more interpersonal ness, all failed to predict suicide attempt
connections than individuals in the general status over and above suicidal desire.
population due to their student status. As The results should be interpreted in
such, it is important that future studies light of potential limitations. First, the anal-
develop and validate broader measures of ysis was based on retrospective self-reports,
connectedness and examine what which may contain inaccuracies due to
8 3ST UK UNIVERSITY

reporting biases or forgetting. Conse- appraisal would suggest that health care pro-
quently, there is a need for future studies to fessionals should explicitly address the
be designed in ways that move us from tests degree to which individuals are currently
of correlation to the examination of risk experiencing feelings of pain and hopeless-
factors, mechanisms, moderators, and com- ness, as well as the degree to which they feel
plex interactions. Prospective and longitudi- connected to others and have capability for
nal studies will be particularly important as lethal self-injury. Addressing the factors that
a next step to test the usefulness of the 3ST involve or increase an individual’s capability
constructs. Second, although the 3ST posits for suicide, in particular, may be important,
specific and testable mechanisms that differ- as capability appears to facilitate the transi-
entiate suicide ideators from suicide attemp- tion from ideation to acts. In addition, when
ters, as yet, it does not make any considered alongside previous research (e.g.,
predictions in terms of when the transition Boduszek & Dhingra, 2015; O’Connor, Fra-
from thought to action will occur, or how ser, Whyte, MacHale, & Masterton, 2008),
this transition is expected to unfold. Given findings suggest that prevention efforts tar-
that the transition from thoughts to action is geting a specific type of hopelessness, namely
typically rapid (Nock, 2010), this constitutes low levels of positive future thinking, may be
an important direction for future research. It particularly useful. To this end, we suggest
is also important to note that there may be the use of behavioral and cognitive tech-
important differences between those who niques (e.g., behavior activation strategies) to
make nonfatal (as in participants in our sam- assist individuals to encode and access partic-
ple) and those who make fatal suicide ular memories (MacLeod & Moore, 2000)
attempts. As only a minority of attempters and to develop positive schemas (Padesky,
die by suicide, and the majority of those who 1994). It is anticipated that through such
do die by suicide do so on their first attempt interventions, at-risk individuals (i.e., those
(Nock et al., 2008), further research is high on pain and hopelessness) may be able
needed to determine whether individuals to access positive memories more frequently,
who have made nonfatal attempts provide with greater speed and specificity.
knowledge that is also relevant to under- In general, the 3ST suggests that any
standing and preventing suicide death. intervention—regardless of type (e.g., medi-
Despite the aforementioned consider- cation vs. psychotherapy) or level (e.g., indi-
ations, the results presented here comple- vidual vs. community)—will succeed to the
ment and extend previous research by testing extent it decreases pain, increases hope,
the 3ST in a large sample of U.K. university improves connection, and/or reduces sui-
students. The results indicate that the 3ST is cide capacity. Future research should exam-
a promising conceptual framework for guid- ine these four variables as potential
ing future research on the development of mechanisms of change in the treatment of
suicide ideation and behavior. suicide risk and develop and evaluate inter-
The findings also have important clin- ventions for suicide risk that specifically tar-
ical implications. Applying our results to risk get one or more of these variables.

REFERENCES

BECK, A. T., BROWN, G. K., & STEER, R. A. The scale for suicidal ideation. Journal of
(1997). Psychometric characteristics of the Scale Consulting and Clinical Psychology, 47, 343–
for Suicide Ideation with psychiatric outpatients. 352. https://doi.org/10.1037/0022-006X.47.2.
Behaviour Research and Therapy, 35(11), 1039–1046. 343
https://doi.org/10.1016/S0005-7967(97)00073-9 BECK, A. T., & STEER, R. A. (1991). Man-
BECK, A. T., KOVACS, M., & WEISSMAN, ual for the beck scale for suicide ideation. San Anto-
A. (1979). Assessment of suicidal intention: nio, TX: Psychological Corporation.
DHINGRA ET AL. 9

BECK, A. T., WEISSMAN, A., LESTER, D., LOGAN, J., HALL, J., & KARCH, D. (2011).
& TREXLER, L. (1974). The measurement of pes- Suicide categories by patterns of known risk fac-
simism: The Hopelessness Scale. Journal of Con- tors: A latent class analysis. Archives of General
sulting and Clinical Psychology, 42, 861–865. Psychiatry, 68(9), 935–941. https://doi.org/10.
https://doi.org/10.1037/h0037562 1001/archgenpsychiatry.2011.85
BODUSZEK, D., & DHINGRA, K. (2015). MACLEOD, A. K., & MOORE, R. (2000).
Construct validity of the Beck Hopelessness Positive thinking revisited: Positive cognitions,
Scale (BHS) among university students: A multi- well- being and mental health. Clinical Psychology
trait–multimethod approach. Psychological Assess- and Psychotherapy, 7, 1–10. https://doi.org/10.
ment, 28(10), 1325–1330. 1002/(ISSN)1099-0879
DHINGRA, K., BODUSZEK, D., & KLONSKY, D. MAY, A. M., & KLONSKY, E. D. (2013).
(2016). Empirically derived subgroups of self-injur- Assessing motivations for suicide attempts:
ious thoughts and behavior: Application of latent Development and psychometric properties of the
class analysis. Suicide and Life-Threatening Behavior, inventory of motivations for suicide attempts.
46, 486–499. https://doi.org/10.1111/sltb.12232 Suicide and Life-Threatening Behavior, 43(5), 532–
DHINGRA, K., BODUSZEK, D., & O’CON- 546.
NOR, R. (2015). Differentiating suicide attempters MAY, A. M., & KLONSKY, E. D. (2016).
from suicide ideators using the Integrated Moti- What distinguishes suicide attempters from sui-
vational-Volitional Model of suicidal behavior. cide ideators? A meta-analysis of potential fac-
Journal of Affective Disorders, 186, 211–218. tors. Clinical Psychology: Science and Practice, 23,
https://doi.org/10.1016/j.jad.2015.07.007 5–20. https://doi.org/10.1111/cpsp.12136.
EISENBERG, D., HUNT, J. B., & SPEER, N. NOCK, M. K. (2010). Self-injury. Annual
(2013). Mental health in American colleges and Review of Clinical Psychology, 6, 339–363. https://d
universities: Variation across student subgroups oi.org/10.1146/annurev.clinpsy.121208.131258
and across campuses. Journal of Nervous and Men- NOCK, M. K., BORGES, G., BROMET, E. J.,
tal Disease, 201(1), 60–67. https://doi.org/10. CHA, C. B., KESSLER, R. C., & LEE, S. (2008).
1097/NMD.0b013e31827ab077 Suicide and suicidal behavior. Epidemiologic
GLENN, C. R., & NOCK, M. K. (2014). Reviews, 30, 133–154. https://doi.org/10.1093/
Improving the short-term prediction of suicidal epirev/mxn002
behavior. American Journal of Preventive Medicine, NOCK, M. K., HOLMBERG, E. B., PHOTOS,
47(3), S176–S180. https://doi.org/10.1016/j.ame V. I., & MICHEL, B. D. (2007). The self-injurious
pre.2014.06.004 thoughts and behaviors interview: Development,
HARROD, C. S., GOSS, C. W., STALLONES, reliability, and validity in an adolescent sample.
L., & DI GUISEPPI, C. (2014). Interventions for pri- Psychological Assessment, 19, 309–317. https://doi.
mary prevention of suicide in university and other org/10.1037/1040-3590.19.3.309
post-secondary educational settings. Cochrane O’CONNOR, R. C. (2011). The integrated
Database of Systematic Reviews, 10, Cd009439. motivational-volitional model of suicide behav-
HOLDEN, R., MEHTA, K., CUNNINGHAM, E. ior. Crisis, 32, 295–298. https://doi.org/10.1027/
J., & MCLEOD, L. D. (2001). Development and 0227-5910/a000120
preliminary validation of a scale of psychache. O’CONNOR, R. C., FRASER, L., WHYTE,
Canadian Journal of Behavioural Science, 33(4), M. C., MACHALE, S., & MASTERTON, G. (2008).
224–232. https://doi.org/10.1037/h0087144 A comparison of specific positive future
JOINER, T. (2005). Why people die by suicide. expectancies and global hopelessness as predic-
Cambridge, MA: Harvard University Press. tors of suicidal ideation in a prospective study of
KLONSKY, E. D., & MAY, A. M. (2014). repeat self-harmers. Journal of Affective Disorders,
Differentiating suicide attempters from suicide 110(3), 207–214. https://doi.org/10.1016/j.jad.
ideators: A critical frontier for suicidology 2008.01.008
research. Suicide and Life-Threatening Behavior, 1, O’CONNOR, R. C., & NOCK, M. K.
1–5. https://doi.org/10.1111/sltb.12068 (2014). The psychology of suicidal behavior.
KLONSKY, E. D., & MAY, A. M. (2015). Lancet Psychiatry, 1, 73–85. https://doi.org/10.
The three-step theory (3ST): A new theory of sui- 1016/S2215-0366(14)70222-6
cide rooted in the “ideation-to-action” framework. ONS (2016). Total number of deaths by sui-
International Journal of Cognitive Therapy, 8(2), cide or undetermined intent for Students aged 18 and
114–129. https://doi.org/10.1521/ijct.2015.8.2.114 above in England and Wales, 2014. Retrieved
KLONSKY, E. D., QUI, T., & SAFFER, B. Y. November 15, 2016, from https://www.ons.gov.
(2017). Recent advances in differentiating suicide uk/peoplepopulationandcommunity/healthandsoc
attempters from suicide ideators. Current Opinion ialcare/causesofdeath/adhocs/005732totalnumbe
in Psychiatry, 30, 15–20. https://doi.org/10.1097/ rofdeathsbysuicideorundeterminedintentforstude
YCO.0000000000000294 ntsaged18andaboveinenglandandwales2014.
10 3ST UK UNIVERSITY

PADESKY, C. A. (1994). Schema change JR, (2008). Suicidal desire and the capability for
processes in cognitive therapy. Clinical Psychology suicide: Tests of the interpersonal-psychological
and Psychotherapy, 1, 267–278. https://doi.org/10. theory of suicidal behavior among adults. Journal
1002/cpp.5640010502 of Consulting and Clinical Psychology, 76(1), 72–83.
ROBINS, A., & FISKE, A. (2009). Explaining https://doi.org/10.1037/0022-006X.76.1.72
the relation between religiousness and reduced WONG, M. M., BROWER, K. J., & ZUCKER,
suicidal behavior: Social support rather than R. A. (2011). Sleep problems, suicidal ideation,
specific beliefs. Suicide and Life-Threatening and self-harm behaviors in adolescence. Journal
Behavior, 39, 386–395. https://doi.org/10. of Psychiatric Research, 45(3), 505–511. https://doi.
1521/suli.2009.39.4.386 org/10.1016/j.jpsychires.2010.09.005
SHNEIDMAN, E. S. (1993). Suicide as psy- World Health Organization (2016). World
chache: A clinical approach to self-destructive behav- Health Organization (WHO) Suicide data.
ior. Northfield, NJ: Jason Aronson. Retrieved June 8, 2016, from http://www.who.
VAN ORDEN, K. A., WITTE, T. K., int/mental_health/prevention/suicide/suicide
CUKROWICZ, K. C., BRAITHWAITE, S. R., SELBY, prevent/en/
E. A., & JOINER, T. E. (2010). The interpersonal
theory of suicide. Psychological Review, 117(2), Manuscript Received: November 30, 2016
575–600. https://doi.org/10.1037/a0018697 Revision Accepted: November 6, 2017
VAN ORDEN, K. A., WITTE, T. K., GOR-
DON, K. H., BENDER, T. W., & JOINER, T. E.,