RESEARCH ARTICLE
Savouring and Self-compassion as Protective Factors for
Depression
Justin Ford1, Jeffrey J. Klibert1*, Nicholas Tarantino2 & Dorian A. Lamis3
1
Department of Psychology, Georgia Southern University, Statesboro, GA USA
Department of Psychology, Georgia State University, Atlanta, GA USA
3
Emory School of Medicine, Atlanta, GA USA
2
Abstract
Within positive psychology, researchers and clinicians vocalize the need to expand upon how the treatment for
major depressive disorder is conceptualized and implemented. The impetus of the current study was to examine
preliminary criteria for identifying savouring and self-compassion as protective factors for depression. Undergraduate
students (N = 133) completed a series of surveys at two points in time, 5 weeks apart. Results revealed that savouring
and self-compassion were inversely related to depression scores cross-sectionally and prospectively. However, savouring
was the only positive psychological variable to predict changes in depression scores across time. Cross-sectionally,
savouring was also found to moderate the relation between negative life events and depression, such that the strength
of the relation between negative life events and depression decreased when higher savouring was present. However, this
same effect was not significant prospectively. There was no evidence, cross-sectionally or prospectively, that
self-compassion moderated the relation between negative life events and depression. Taken together, results provide
preliminary support for savouring as a protective factor for depressive symptoms. Mental health professionals should
consider teaching savouring strategies to help at-risk clients stimulate and sustain positive affect as a means of preventing
and reducing depressive symptoms. Copyright © 2016 John Wiley & Sons, Ltd.
Received 27 August 2015; Revised 6 May 2016; Accepted 9 May 2016
Keywords
savouring; self-compassion; depression; negative life events
*Correspondence
Jeffrey Klibert, Georgia Southern University, 2670 Southern Dr., Statesboro, GA 30460, USA.
†
Email: [email protected]
Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smi.2687
Introduction
Epidemiological studies characterize major depressive
disorder (MDD) as a considerable societal burden, with
an estimated prevalence rate of 350 million cases
worldwide and an annual incidence rate of 3.0% (Ferrari
et al., 2013a, 2013b; World Health Organization, 2012).
Currently, MDD is the third leading burden of disease,
accounting for 8.2% of years lived with disability and
2.5% of total disability-adjusted life years, a metric derived
from the sum of years of life lost due to premature
mortality and years of life with a disability (Ferrari et al.,
2013a, 2013b). After adjusting for population increases,
MDD is projected to be the leading burden of disease in
2030 (World Health Organization, 2012). MDD presents
with a chronically intermittent life course, and despite
the availability of numerous empirically validated
psychopharmological and psychotherapeutic treatments,
clinicians are continually challenged to find effective
methods for reducing MDD symptomology and
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
preventing future depressive episodes (Farb, Irving,
Anderson, & Segal, 2015). Specifically, research indicates
that between 30% and 40% of individuals report few
beneficial effects associated with antidepressant
medication (Kornstein & Schneider, 2001), and less than
40% of those seeking treatment for the first time achieve
remission of symptoms (Rush et al., 2006).
Consistent with the tenets of prevention science,
barriers to remission are associated with a lack of
consideration for strength-based factors, resources that
promote advances in treatment even in the face of risk
and resistance (Nash & Bowen, 2002). In line with this
position, in the current study, we seek to identify
protective factors for depression and evaluate the
interactive effects between negative life events and two
proposed protective factors in the prediction of
depression across time. It is our hope that the study
results will generate pathways by which clients can
achieve full remission of depressive symptoms and
marshal resilience resources to prevent relapse.
Protective Factors for Depression
Negative life events, undesirable circumstances
that pose a threat to cognitive, behavioural and/or
interpersonal functioning, do not constitute a
sufficient condition by which depressive symptoms
arise (Spinhoven et al., 2011). For instance, 50% of
individuals who experience negative life events do not
report symptoms or problem behaviours associated with
mood difficulties, suggesting that a substantial amount of
people can either successfully navigate through or at least
tolerate the burden of negative life events without
significant mood impairment (Monroe & Hadjiyannakis,
2002). Findings, such as these, prompt researchers to
explore psychological characteristics that alter the course
and nature by which negative life events are associated
with depressive outcomes (Zuess, 2003). However, the
mechanisms by which individuals employ positive
psychological resources to decrease the debilitative effects
of negative life events and minimize depression are
relatively unknown. Investigating these mechanisms is
key in terms of elucidating pathways by which individuals build resilience to depressive outcomes (Bonanno
& Diminich, 2013). We propose that two mechanisms,
savouring and self-compassion, may contribute to an
individual’s ability to modify and counteract the effects
of negative life events as a means to thwart the onset of
depressive symptoms.
Savouring
Savouring is an affective regulation tactic characterized
by the propensity to focus and exert control over how
individuals develop, intensify and sustain positive
emotions (Bryant, 2003; Bryant & Veroff, 2007). For
example, individuals often generate and enhance
positive affect surrounding a vacation by employing
savouring tactics such as anticipation, fantasizing and
proactively constructing an enjoyable itinerary
beforehand. Furthermore, they may also absorb and
block distractions in favour of mindfully immersing
themselves in the enjoyment of the current vacation
activity. The potential benefits of savouring are
well-outlined in the literature. Of note, research
conducted on various populations consistently
highlights savouring in the promotion of psychological well-being and the discontinuance of negative
affective states, most prominently depression. Most
cross-sectional studies report direct negative
relations between savouring and different indices of
depressive psychopathology (e.g., Bryant, 2003).
Similarly, experimental investigations of savouring
interventions reveal marked decreases by which
participants report and express depressive symptoms
(Hurley & Kwon, 2012).
Despite emerging evidence delineating the beneficial
effects of savouring, research has yet to fully disentangle
the links among negative life events, savouring and
depression. However, researchers highlight variables
that promote positive affective functioning, like
savouring, as critical in the deterrence of depressive
J. Ford et al.
symptoms resulting from negative life events
(McMakin, Siegle, & Shirk, 2011). According to the
Broaden and Build Theory (Fredrickson, 2001),
generating and extending positive emotions in the
context of negative life events can undo or negate
cardiovascular aftereffects and lingering emotions
(e.g., frustration, fear) to promote quick recovery
and flourishing. A handful of studies indirectly
support this position. Through the manipulation of
gratitude, a unique savouring response that sustains and
enhances attention to positive emotions/experiences
(Bryant & Veroff, 2007), Krejtz, Nezlek, Michnicka,
Holas, and Rusanowska (2014) found a weaker withinperson relation between daily stress and depressogenic
adjustment for individuals who completed a gratitude
intervention compared with participants in a control
group. This finding suggests savouring strategies may
offset the negative effects of stress to reduce overall risk
to depression. Similarly, savouring through behavioural
expression appears to modify the relation between
negative life events and depression. For example, genuine
behavioural expressions of savouring (i.e., smiling and
laughing) when speaking about the death of spouse/
partner at 6 months post-loss minimized interpersonal
symptoms of grief and depression (i.e., isolation, psychomotor retardation, subjective feelings of hopelessness) at
25 months (Bonanno, 2008; Bonanno & Keltner, 1997).
Together, these findings provide strong support for the
‘undoing effects’ of savouring and suggest higher levels
of savouring may weaken the relation between negative
life events and depression.
Self-compassion
Self-compassion is an emotional regulation tactic
marked by the propensity to activate kind and
non-judgmental attitudes in the face of personal
difficulties, feelings of inadequacy and perceptions of
failure (Neff, 2003a). Individuals who exhibit high
levels of self-compassion often generate positive
emotions (e.g., peacefulness, calmness, contentment)
that contribute to higher rates of psychological
well-being and life satisfaction (Neff, Kirkpatrick &
Rude, 2007a, 2007b). Alternatively, theorists postulate
self-compassion as a protective factor against depression,
a position supported by a large combined effect size for
the inverse relation between self-compassion and
depressive symptoms (MacBeth & Gumley, 2012).
However, research exploring pathways by which
self-compassion exerts its salutary effects on depression
is only just emerging.
When employed, self-compassion can defuse the
effects of negative life events as a means of minimizing
risk for depression (Gilbert, 2006; Neff, 2011). Drawing
from the social mentality theory (Gilbert, 1989),
self-compassion may suppress threat systems commonly
activated during bouts of psychopathology, including
depression. Theorists highlight self-compassion facets,
particularly kindness and forgiveness, as inhibiting agents,
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
J. Ford et al.
variables that sooth stress systems and reduce the onset of
depressive symptoms. Empirical backing for the
protective effects of self-compassion is demonstrated in
a number of recently published works. Self-compassion
moderates how people react to distressing events.
Specifically, higher self-compassion predicts lower reports
of sadness and embarrassment in the face of real,
remembered and imagined negative life events (Leary,
Tate, Adams, Batts Allen, & Hancock, 2007). Finally,
participants asked to complete a writing-based selfcompassion task after recall of a shameful memory report
reduced levels of depression compared with participants
who completed a broad emotion-based expressive writing
task (Johnson & O’Brien, 2013). This finding
supplements previous research by suggesting individuals
high in self-compassion not only recover from negative
life events but do so in a way that minimizes overall risk
to depression. Considering relevant theoretical and
empirical findings, it is expected that high
self-compassion scores will weaken the relation
between negative life events and depression.
Traditional research on depression is grounded in
the identification and the management of risk factors,
an exclusive and restrictive approach that has largely
failed to prevent such difficulties from emerging
(Seligman & Csikszentmihalyi, 2000; Wood & Joseph,
2010). Thus, prevention science theorists stringently
argue for the identification of protective factors as
necessary and advantageous to increase remission rates
of depressive symptoms and decrease the likelihood of
relapse (Sin & Lyubomirsky, 2009; Watson &
Naragon-Gainey, 2010). Moreover, there is an
over-reliance on cross-sectional data, those that fail to
consider the temporal relations between a proposed
predictor and an outcome, in identifying protective
factors for depression. Protective factors are
methodologically defined as conditions that are directly
associated with lower estimates of an outcome (i.e.,
depression) and for which there is evidence that the
condition preceded the outcome (i.e., temporal
precedence; Vagi et al., 2013). The identification of
protective factors is also determined by a variable that
counteracts the effects of stressful and tumultuous life
experiences in the prediction of depression (Schrank,
Brownell, Tylee, & Slade, 2014; Steca et al., 2014).
Overall, consideration for temporal precedence and
offsetting effects is necessary to help researchers and
clinicians develop more effective primary-prevention
approaches to depression (Cairns, Yap, Pilkington, &
Jorm, 2014).
In the current study, we investigate (a) the direct
associations between savouring/self-compassion and
depression, (b) the antecedent effects of savouring
and self-compassion on depression and (c) the
moderating effects of savouring and self-compassion
on the well-established relation between negative life
events and depression. In light of applicable theory
and empirical work, we hypothesized the following:
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
Protective Factors for Depression
(a) negative life events would be positively associated
with depression across time; (b) savouring and
self-compassion would be inversely associated with
depression across time; (c) higher levels of savouring
would reduce the relation between negative life events
and depression; and (d) higher levels of selfcompassion would reduce the relation between
negative life events and depression.
Method
Participants
Two hundred and fifty-one undergraduate students
from a large southeastern university were recruited to
participate in a series of online surveys. The initial
sample of participants consisted of 203 women
(80.9%) and 48 men (19.1%) with a mean age of
21.30 (SD = 3.10) years. The majority of participants
self-identified as European American (n = 136, 54.2%)
or African American (n = 86, 34.3%). Participants were
asked to complete two surveys over a 5-week period.
The attrition rate between administrations was
44.63% with 139 students participating in both surveys.
Of the 139 participants, six cases (4%) were removed
because of missing data, resulting in a final sample of
133. The final sample consisted of 21 men (15.8%)
and 112 women (84.2%), aged 18–27 (M = 21.0,
SD = 1.54). The smaller ratio of men to women is
consistent with other investigations using emerging
adult samples (Johnson & O’Brien, 2013). In terms of
ethnicity, the majority of the sample self-reported as
European American (n = 79, 59.4%), with the
remaining participants identifying as African American
(n = 39, 29.3%), multiracial (n = 10, 7.5%), Asian
American (n = 2, 1.5%), Mexican American (n = 1,
.8%), Native American (n = 1, .8%) and international
students (n = 1, .8%). Most participants indicated their
sexual minority status as heterosexual (90%). About
one in three students indicated that they were from
rural communities, and the remaining participants
were from urban areas. The majority (92%) reported
being single as compared with married or divorced.
All participants received two course credits for
completing the surveys.
Design
The current study employed a short-term prospective
design to examine the predictive validity of negative life
events and positive psychological factors on depression.
Unlike studies utilizing cross-sectional and retrospective
designs, prospective investigations elucidate temporal
relations (Voisin, Hotton, Tan, & DiClemente, 2013),
an essential component in the identification of risk and
protective factors (Cairns et al., 2014).
Participants completed a series of questionnaires at
two time points. The interval between each survey
administration was approximately 5 weeks. Haeffel
et al. (2007) recommend shorter term intervals
Protective Factors for Depression
between survey administrations when examining
longitudinal hypotheses associated with negative life
events. Employing a narrow time frame between
administrations is advantageous as participants are more
likely to recall the frequency and resulting distress
associated with recent negative life events. Overall,
prospective designs, even short-term prospective designs,
engender more accurate and valid results regarding the
stability of important risk/protective-outcome relations
(Ingram, Miranda, & Segal, 1998).
Measures
The Center for Epidemiologic Studies Depression
Scale (CES-D; Radloff, 1977) is a 20-item self-report
scale designed to assess depression symptoms in
adult populations. Each item on the CES-D is
measured on a four-point frequency scale (from
1 = rarely or none of the time to 4 = most or all of
the time) with total scores ranging from 20 to 80.
Higher scores indicate greater levels of depression.
The CES-D has demonstrated good internal
consistency with samples of college students
(α = .87–.91; Hill, Yaroslavsky, & Pettit, 2015). The
CES-D has also demonstrated excellent construct
validity as evidenced by high correlations with
anxiety, affective liability and distress intolerance
(Pearson, Lawless, Brown, & Bravo, 2015).
In the current study, we examined the factor
structure of the CES-D. Confirmatory Factor Analysis
(CFA) revealed good fit for a higher order model with
four established factors (Radloff, 1977; i.e. positive
affect, depressed affect, somatic complaints/activity
inhibition and interpersonal problems) and an
underlying latent depression factor, RMSEA = .09,
CFI = .91 and SRMR = .06, thus providing suitable
evidence for the use of a CES-D total depression score.
Internal consistency scores for the CES-D total score in
first administration (α = .91) and second administration (α = .92) were excellent. The CES-D also
demonstrated a solid test–retest reliability estimate
(r = .74).
The Inventory of College Students’ Recent Life
Experiences (ICSRLE; Kohn, Lafreniere, & Gurevich,
1990) is a 49-item self-report scale designed to assess
the frequency of negative life events in a student’s
college career. Examples of negative life events include
conflicts with significant others, social rejection and
struggling to meet one’s own academic standards. Each
item on the ICSRLE is measured on a four-point scale
(from 1 = not at all part of my life to 4 = very much a
part of my life) with total scores ranging from 49 to
196. Higher scores indicate greater experiences with
negative life events. The ICSRLE has evidenced good
internal consistency (α = .91; Stoltzfus & Farkas, 2012)
and excellent construct validity with other measures
of negative life events (Kohn et al., 1990). In the
current study, the internal consistency scores for the
ICSRLE in first administration (α = .94) and second
J. Ford et al.
administration (α = .95) were excellent. The ICSRLE
also demonstrated a solid test–retest reliability estimate
(r = .72).
The Savouring Beliefs Inventory (SBI; Bryant, 2003)
is a 24-item self-report measure designed to assess the
extent to which individuals savour positive experiences
in their lives. Savouring is a multidimensional
construct that consists of three subscales: anticipation,
reminiscing and in the moment savouring. However,
only a total score was calculated in the current study.
Each item on the SBI is measured on a seven-point
Likert scale (from 1 = strongly disagree to 7 = strongly
agree) with total scores ranging from 24 to 168. High
total scores are reflective of greater savouring beliefs.
The SBI has been found to have good internal
consistency (α = .89; Bryant, 2003). The SBI has also
demonstrated excellent construct validity as evidenced
by high positive correlations with gratification and
self-esteem and inverse correlations with strain and
depression (Bryant, 2003).
In our CFA of the SBI, a bi-factor structure was
found to be the best fitting model and was only a fair
fit overall, RMSEA = .09, CFI = .88 and SRMR = .07.
In this model, items are allowed to load independently
to an underlying savouring factor as well as to three
uncorrelated savouring sub-constructs (anticipation,
reminiscing and in the moment savouring). As with a
higher order model, bi-factor models also generate
evidence for the use of a total score (Neff, 2016). The
internal consistency scores for the SBI total score in
first administration (α = .95) and second administration (α = .96) were excellent. The SBI total score also
demonstrated a solid test–retest reliability estimate
(r = .77).
The Self-Compassion Scale (SCS; Neff, 2003b) is a
26-item self-report measure designed to assess levels
of self-compassion in adult participants. Each item on
the SCS is measured on a five-point scale (from 1 = almost never to 5 = almost always) with total scores
ranging from 26 to 130. Higher total scores on this
scale indicate greater levels of self-compassion and
lower levels of self-judgment. The SCS has
demonstrated good internal consistency (α = .92) and
excellent construct validity as evidenced by high
correlations with self-esteem (Neff, 2003b).
Similar to the SBI, a CFA for the SCS revealed only a
fair fit, RMSEA = .08, CFI = .87 and SRMR = .10, for a
bi-factor model. Because the bi-factor structure
demonstrated comparable fit with other models tested,
we chose to retain it and use a total SCS score for our
analyses. Neff (2016) has recommended this approach
as she has contends that a bi-factor structure is the
most representative model for conceptualizing
self-compassion. The internal consistency scores for
the SCS total score in first administration (α = .94)
and second administration (α = .95) were excellent.
The SCS score has also demonstrated a solid test–retest
reliability estimate (r = .84).
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
J. Ford et al.
Protective Factors for Depression
Procedure
Before initiating data collection, the researchers
submitted data the collection plan to the university’s
Institutional Review Board (IRB), a federally mandated
risk oversight committee. Institutional Review Board
approval was obtained, and ethical guidelines were
followed throughout the data collection process. After
obtaining informed consent, participants volunteered
to answer survey questions anonymously online via
SurveyMonkey.com, which took approximately
50 min to complete. Participants constructed a
personally meaningful and discrete code number that
was used to connect their survey responses from one
administration to the next. After the completion of
the survey, all participants were debriefed and offered
information regarding how to acquire free to low-cost
mental health services in the area, if desired.
Plan of analysis
Two separate hierarchical regressions were used to test
the study’s primary hypotheses, that is, protective
factors associated cross-sectionally and prospectively
with depression. Firstly, to examine cross-sectional
main effects associated with T1 depression, step one
of the regression included the predictor variables of
T1 stressful life events and T1 self-compassion and T1
savouring, adjusting for all covariates (age, race,
gender, marital status, sexual minority status and
whether participants were from rural or urban
communities). To examine cross-sectional interaction
effects, the second step of the regression had all of the
variables in the previous step in addition to two
stressful life events by moderator variable
(self-compassion and savouring) interaction terms.
Longitudinal effects predicting T2 depression were
similarly examined with the addition of controlling
for the effect of T1 depression. Thus, main and
interaction effects in the longitudinal model can be
interpreted as residualized change scores in depression.
If the interaction terms in either models were
significant and explained a significant amount of
additional variance, conditional effects were probed.
In this study, conditional effects are the effects of
stressful life events on depression scores at levels of a
given protective factor (i.e., moderator). One standard
deviation (SD) below the mean, the mean and one
SD above the mean were chosen to represent high,
medium and low levels of a moderator, respectively.
The significance of conditional effects was also tested.
Established SPSS macros was used to test these effects
and aid in constructing figures (Hayes & Matthes,
2009).
An attrition analysis was also performed to
determine if T1 predictor (stressful life events,
self-compassion and savouring), outcome (depression)
and covariate (age, gender, race, sexual minority status,
relationship status and hometown setting) variables
were related to participants’ completion of the
follow-up assessment. This was accomplished by
creating a T2 completion variable (0 = not completed;
1 = completed) and conducting a logistic regression with
all variables predicting T2 completion. We found that
no study variable significantly influenced T2
completion.
Results
Descriptive statistics and bivariate correlations between
main study variables are shown in Table I. As indicated,
study variables were all significantly related to each
other in expected directions. Covariates were associated
with these variables as follows (not tabled). European
American students had lower rates of stressful life
events than non-European American students at both
time points (rs = !.26 to !30; ps < .01). Being female
was associated with a higher degree of savouring at
T2 (r = .20; p < .05). Age, race, sexual minority status
and history of living in a rural community were not
related to main study variables.
Table II presents the results of the hierarchical
regressions. Cross-sectional effects on T1 depression
were first explored through main effects of predictor
variables controlling for all covariates in step one. As
Table I. Descriptives and bivariate correlations between variables (N = 133)
Variable
1
2
3
4
5
6
7
8
1. Stressful life events T1
2. Stressful life events T2
.72
3. Savouring T1
!.27
!.29
4. Savouring T2
!.19
!.27
.77
5. Self-compassion T1
!.42
!.37
.40
.38
6. Self-compassion T2
!.38
!.41
.44
.48
.84
7. Depression T1
.66
.59
!.54
!.43
!.55
!.56
8. Depression T2
.44
.60
!.57
!.66
!.46
!.58
.74
Range
58–151
49–158
75–168
68–168
36–126
35–124
21–72
20–74
Mean (SD)
94.71 (20.65) 95.04 (21.91) 130.18 (22.51) 130.50 (24.26) 76.31 (18.46) 77.26 (19.21) 34.71 (10.07) 35.30 (10.48)
All correlations are significant (p < .01).
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
Protective Factors for Depression
J. Ford et al.
Table II. Summary of hierarchical regression analysis (N = 133)
Variables
Predictors
T1 depression
Stressful life events (SLE)
Savouring
Self-compassion
Interactions
SLE*savouring
SLE*self-compassion
2
R
2
Change in R
B
Predicting T1 depression
Step 1
Step 2
95% CI
B
95% CI
Step 1
B
95% CI
.77
.23
!.14
!.13
.61
[.17, .30]
[!.20, !.08]
[!.21, !.06]
.91
.28
!.01
[.58, 1.24]
[.00, .56]
[!.30, .28]
!.00
!.00
.66
.05
[!.01, !.00]
[!.00, .00]
.55
[.64, .89]
Predicting T2 depression
Step 2
B
95% CI
B
.65
!.03
!.10
!.04
.59
.04
[.46,
[!.11,
[!.16,
[!.12,
.84]
.06]
!.03]
.05]
Step 3
95% CI
.61
.19
!.00
.05
[.40,
[!.23,
[!.32,
[!.28,
.81]
.61]
.32]
.38]
!.00
!.00
.60
.00
[!.01, .00]
[!.00, .00]
All models include covariates. Entries for predictors and interactions are unstandardized B’s. Bolded effects are significant (p < .05). Effects and
confidence intervals (CI) that contain only zeros (.00 and !.00) were rounded to the nearest decimal and represent the direction of the effect
(positive or negative).
shown, savouring and self-compassion were negatively
associated (squared semi-partial correlations [sr2]
= .07 and .04, respectively), and stressful life events
were positively associated (sr2 = .17), with depression.
These three main effects were significant (ps < .05). In
the second step, cross-sectional interaction effects were
added to the model, and the stressful life
events × savouring interaction term was significant
(p < .05). Variance explained in the model in this step
increased from 61% to 66% (p < .05). The interaction
was then probed for conditional effects. Figure 1
demonstrates the predicted effect of stressful life events
on T1 depression when savouring was low, medium or
high. It was found that stressful life events were
significantly and positively related to depressive
symptoms only when savouring was low (B = .43,
95% CI = [.22, 63]) or medium (B = .33, 95% CI =
[.09, 56]) but not high, which is consistent with our hypothesis that higher levels of savouring are a protective
factor that attenuates the stressful life eventsdepression association. However, in the presence of
this effect, similar support was not found for
self-compassion as a moderator. Finally, no covariate
had a significant cross-sectional effect on the outcome
(not tabled).
The second hierarchical regression explored whether
or not our prospective associations predicted changes
in depression over a short period of time. This second
model predicting T2 depression was identical to the
first with exception of controlling for T1 depression.
Table II shows the large effect of T1 depression on T2
depression (sr2 = .53) in step one of the regression
controlling for all covariates. The inclusion of T1 levels
of stressful life events, savouring and self-compassion
in step two explains a significant amount additional
variance (from 55% to 59%; p < .05). Unlike
cross-sectional effects, stressful life events and
self-compassion were no longer significant predictors
of depression (i.e., change in depression). However,
savouring remained significant (sr2 = .03). Thus,
despite such a short period of time, one in which
depressive symptoms are not likely to fluctuate
Figure 1. Association between stressful life events and depression scores at T1 at varying degrees of savouring controlling for all covariates.
The X-axis is centred at the mean of stressful life events, and the ends are !1 standard deviations (SD) and +1 SD, respectively
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
J. Ford et al.
tremendously, savouring still has an impact on
depression, regardless of stressful life events and degree
of self-compassion. Support was not found, nevertheless,
for savouring or self-compassion as buffers of the negative
short-term longitudinal effects of stressful life events on
depression. Demographic covariates did not influence
any longitudinal association.
Discussion
Consistent with the first two hypotheses, stressful life
events were positively related to depression, and
reports of savouring and self-compassion were
inversely related to depression cross-sectionally and
over a small period of time. These effects align well
with previous research (Hurley & Kwon, 2012; Johnson
& O’Brien, 2013). However, when considering the
direct effects of T1 variables on depression at T2,
different conclusions emerged. Of importance,
savouring was the only T1 variable to directly account
for unique variance in T2 depression scores. This
finding satisfies two required conditions, inverse
relation and temporal precedence, to determine
savouring as a protective factor for depression (Vagi
et al., 2013).
This study is the first to consider the prospective
effects of savouring on change in depression scores
and represents a step forward in determining how
positive emotion regulation tactics can minimize the
duration of short-term depressive symptoms.
Consistent with prior research (Carl, Fairholme,
Gallagher, Thompson-Hollands, & Barlow, 2014), our
results suggest individuals who experience short-term
depressive states report difficulties activating savouring,
the ability to generate, intensify and sustain positive
emotions. Without the use of savouring, individuals
who report depressive features may experience delays
in mood recovery, which in turn may increase
vulnerability to more persistent and dysfunctional
depressive outcomes (i.e., depressive episodes).
Considering most individuals report some experience
with mild and transient forms of depression at one
point in his/her life, positive psychological interventions
focusing on savouring may be a viable prevention
strategy. Teaching individuals how to mindfully savour
positive emotions in lieu of depressive symptoms may
generate more intrapersonal and interpersonal support
to alleviate pressing emotional concerns more
expediently. For instance, recalling and sustaining
attention towards empowered life images and memories
(a savouring process termed Memory Building; Bryant &
Veroff, 2007) may help individuals minimize depressive
states by generating greater levels of coping resources
(e.g., optimism, resilience, wisdom), speeding up
cardiovascular recovery and eliciting support from
friends and family (Tugade & Fredrickson, 2007).
Alternatively, negative life events and self-compassion
scores at T1 did not directly explain unique variance in
T2 depression scores after accounting for depression
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
Protective Factors for Depression
scores at T1, which given previous research (e.g., Monroe
& Harkness, 2005; Raes, 2011) and the strong bivariate
associations found in the current study, was surprising.
However, we interpret these non-significant effects with
caution. We cannot rule out the possibility that the time
interval, 5 weeks, between survey administrations was a
limitation. Although similar studies (e.g., Haeffel et al.,
2007) use comparable time intervals, the short window
between survey administrations generated very high
covariance between depression scores at T1 and T2. In
this case, high covariance left little residual variance for
predictors to account for in T2 depression scores.
Residual variance was minimized further after
considering savouring scores. Given these trends, it is
recommended that future research prospectively
evaluate the direct effects of negative life events
and self-compassion on depression using longer
time intervals between survey administrations.
Depressive episodes can persist between 2 and
12 weeks (American Psychiatric Association, 2013).
Therefore, designing longitudinal studies with a 4- to
6-month time interval may be advantageous in
determining whether or not negative life events and
self-compassion directly predict variance in depression
scores.
A third consideration in identifying protective
factors is determining whether or not a proposed
variable counteracts the negative effects of stressful
events to reduce risk for depression (Schrank et al.,
2014; Steca et al., 2014). Thus, we examined the
moderated effects of savouring and self-compassion
on the relation between negative life events and
depression cross-sectionally and prospectively.
Evidence for our third and fourth hypotheses was
mixed.
Cross-sectionally, savouring did interact with
negative life events to account for a small amount of
variance in depression scores. Although the interaction
term appears small, only accounting for 5% of variance
in depression scores, other survey-based studies
regularly report and interpret comparable effects as
practically meaningful (Haeffel & Vargas, 2011).
Notably, our results suggest in instances of high
negative life events, higher levels of savouring buffer
emerging adults against depressive symptoms. Upon
initial examination, this finding supports the undoing
hypothesis of Fredrickson’s Broaden and Build Theory
(Fredrickson, 2001), which states positive affective
stimulation and sustainment are effective solutions in
minimizing the negative effects of stressful events on
different emotional outcomes, like depression.
However, we considered the moderated effects of
savouring further through a prospective analysis.
Surprisingly, results did not reveal a significant
moderated effect for savouring at T1 on the short term
longitudinal relation between negative life events T1
and depression T2. One possible explanation for this
null effect is the patterns by which student samples
Protective Factors for Depression
commonly report on psychological resources.
Specifically, college student samples either experience
higher levels of psychological resources or obscure
difficulties marshalling psychological resources to
create more positive expressions of themselves. As a
result, normal variation within the moderating
variables, as is commonly seen in clinical samples, is
not likely to be achieved in student samples. This may
be problematic given the protective qualities of
savouring, generating and prolonging positive affect,
are particularly useful in thwarting depression among
individuals who tend to minimize the effects of positive
emotions (Carl et al., 2014; McMakin et al., 2011).
Such patterns in reporting may minimize the potential
for detecting stronger interactive effects in the
prediction of depression across a small timeframe.
Future research may need to examine these
associations using clinical samples.
Alternatively, we did not find significant moderated
effects for self-compassion in either the cross-sectional
or prospective models. These findings were quite
surprising given the extensive literature base highlighting
the salutary effects of self-kind and forgiving dispositions
in managing stress and conflict (Gilbert, 2006; Neff,
2011). Two phenomena may explain our disparate
findings. Firstly, we examined the moderated effects of
self-compassion in the presence savouring. It is possible
these two protective factors operate through similar
mechanisms to reduce the association between stressful
life events and depression. However, because savouring
had a greater influence in this regard, the smaller effects
of self-compassion as a buffering agent may have been
masked. Secondly, using a generalized measure of
self-compassion in favour of more delineated
component measures may have limited our ability to
detect significant results. The SCS furnishes a total score
along with six dimensional scores. Given the smaller
sample size and concerns regarding low reliability
estimates (e.g., low number of items in each component
score [n = 4–5]), we decided to use a total score over
underlying component scores. This could be problematic
if these underlying dimensions differentially influence how individuals manage negative life events.
For instance, underlying self-soothing/self-kindness
components of self-compassion readily disarm
overly critical and internalized responses to negative
events (Neff, 2003a), perhaps more effectively than
the common humanity-based and mindfulnessbased components. To remedy these shortcomings,
experimental mixed subjects designs focused more
narrowly on self-soothing/self-kindness components
of self-compassion are warranted.
Outside of the previously noted limitations, there
are also some general limitations worth noting.
Firstly, participants consisted of undergraduate
J. Ford et al.
psychology students enrolled in a regional university
in the southeastern portion of the United States. The
non-clinical nature of our sample generates some
concern of how these findings will generalize to
community and outpatient samples. Another limitation to the current study was the preponderance of
women in our sample. Although this trend is not
uncommon in research surveying college students,
it did restrict our ability to evaluate potential
gender-specific effects in our analyses. Future
research should replicate these findings using more
diverse and inclusive samples to ensure these results
are stable across cultural and developmental groups.
A third limitation was the correlational nature of the
study. Correlational studies are limited given that
causal statements about the relations among
negative life events, depression and positive psychology factors cannot be inferred. Finally, model fit
indices identified by CFA were only fair, especially
for the SCS, which may limit the reliability of our
conclusions. Future research may need to confirm
our conclusions by using different factor structures
of the chosen measures. In particular, verifying and
employing multidimensional factor structures to
the SBI and SCS assessments may generate unique
findings that could strengthen, further clarify or
modify our conclusions.
Nonetheless, these results provide preliminary
support for savouring as a protective factor for
depression. In terms of established criteria (Vagi
et al., 2013), savouring demonstrated an inverse
relation and temporal precedence in predicting
depression scores. In addition, results highlighted
partial support for the position that savouring
moderates the association between negative life events
and depression. At the cross-sectional level, negative
life events were not related to depression scores when
savouring was high. Although this effect was not
replicated in our longitudinal analysis, we believe low
power resulting from a reduced sample size negatively
impacted our ability to detect significant results.
Overall, our results offer more preliminary support
for savouring than self-compassion as a protective
factor. Mental health professionals should consider
implementing savouring strategies (e.g., Memory
Building) as a means of preventing and disrupting the
maintenance of depressive symptoms among at-risk
clients.
Ethical statement
The current study was self-funded.
Conflict of interest
The authors declare that they have no conflict of
interest.
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
J. Ford et al.
Protective Factors for Depression
Gilbert, P. (1989). Human Nature and Suffering. Hove,
REFERENCES
UK: Lawrence Erlbaum Associates.
American Psychiatric Association (2013). Diagnostic
and Statistical Manual of Mental Disorders(5
pp.).
Arlington,
VA:
American
th
ed.
Psychiatric
Association.
resilience: Have we underestimated the human
capacity to thrive after extremely aversive events?.
Psychological Trauma: Theory, Research, Practice, and
Policy, S, 1, 101–113. DOI:10.1037/1942-9681.
S.1.101.
Bonanno, G. A., & Diminich, E. D. (2013). Annual
research review: Positive adjustment to adversity—
trajectories
of
minimal–impact
resilience
and
emergent resilience. Journal of Child Psychology
and
Therapy and Research, 35(3), 217–226. DOI:10.1007/
Gilbert, P. (2006). Evolution and depression: Issues and
implications. Psychological Medicine, 36(3), 287–297.
DOI:10.1017/S0033291705006112.
Psychiatry,
54(4),
378–401.
DOI:10.1111/
jcpp.12021.
Bonanno, G. A., & Keltner, D. (1997). Facial expressions
of emotion and the course of conjugal bereavement.
Journal of Abnormal Psychology, 106(1), 126–137.
DOI:10.1037/0021-843X.106.1.126.
implicit
cognition:
dual-process
theory
A
preliminary
of
cognitive
test
of
a
vulnerability.
Monroe, S. M., & Harkness, K. L. (2005). Life stress, the
‘kindling’
hypothesis,
and
the
recurrence
of
depression:
DOI:10.1016/j.brat.2006.09.003.
perspective. Psychological Review, 112, 417–445.
Haeffel, G. J., & Vargas, I. (2011). Resilience to depressive
Considerations from a life stress
DOI:10.1037/0033-295X.112.2.417.
symptoms: The buffering effects of enhancing cognitive
Nash, J. K., & Bowen, G. L. (2002). Defining and esti-
style and positive life events. Journal of Behavior Therapy
mating risk and protection: An illustration from the
and
School Success Profile. Child & Adolescent Social
Experimental
Psychiatry,
42(1),
13–18.
DOI:10.1016/j.jbtep.2010.09.003.
Work
Hayes, A. F., & Matthes, J. (2009). Computational
Journal,
19,
247–261.
DOI:10.1023/
A:1015532132061.
procedures for probing interactions in OLS and
Neff, K. (2003a). Self-compassion: An alternative
logistic regression: SPSS and SAS implementations.
conceptualization of a healthy attitude toward
Behavior
924–936.
oneself. Self and Identity, 2, 85–101. DOI:10.1080/
Hill, R. M., Yaroslavsky, I., & Pettit, J. W. (2015).
Neff, K. D. (2003b). The development and validation of
Enhancing depression screening to identify college
a scale to measure self-compassion. Self and Identity,
Research
Methods,
41,
15298860309032.
DOI:10.3758/BRM.41.3.924.
students at risk for persistent depressive symptoms.
scale for measuring beliefs about savouring. Journal
Journal of Affective Disorders, 174, 1–6. DOI:10.1016/
0963823031000103489.
development adaptation. Developmental Review, 22,
78–96.
Behaviour Research and Therapy, 45, 1155–1167.
Bryant, F. (2003). Savoring Beliefs Inventory (SBI): A
of Mental Health, 12, 175–196. DOI:10.1080/
s10608-010-9311-5.
Monroe, S., & Hadjiyannakis, K. (2002). Proximal and
distal influences on development: The model of
Haeffel, G. J., Abramson, L. Y., Brazy, P., Shah, J.,
Teachman, B., & Nosek, B. (2007). Explicit and
Bonanno, G. A. (2008). Loss, trauma, and human
investigation of treatment-related effects. Cognitive
2, 223–250. DOI:10.1080/15298860309027.
Neff, K. D. (2011). Self-compassion. New York:
HarperCollins.
j.jad.2014.11.025.
Hurley, D. B., & Kwon, P. (2012). Results of a study to
Neff, K. D. (2016). The Self-compassion Scale is a valid and
Bryant, F. B., & Veroff, J. (2007). Savoring: A New
increase savoring the moment: Differential impact on
theoretically coherent measure of self-compassion.
Model of Positive Experience. Mahwah, NJ, US:
positive and negative outcomes. Journal of Happiness
Mindfulness, 7(1), 264–274. DOI:10.1007/s12671-
Lawrence Erlbaum Associates Publishers.
Studies, 13, 579–588. DOI:10.1007/s10902-011-9280-8.
015-0479-3.
Cairns, K. E., Yap, M. H., Pilkington, P. D., & Jorm, A.
Ingram, R. E., Miranda, J., & Segal, Z. V. (1998).
F. (2014). Risk and protective factors for depression
Cognitive Vulnerability to Depression. New York,
Self-compassion
that adolescents can modify: A systematic review
NY: Guilford Press.
functioning. Journal of Research in Personality, 41,
Neff, K. D., Kirkpatrick, K. L., & Rude, S. S. (2007a).
and
adaptive
psychological
and meta-analysis of longitudinal studies. Journal of
Johnson, E. A., & O’Brien, K. A. (2013). Self-compassion
Affective Disorders, 169, 61–75. DOI:10.1016/j.
soothes the savage ego-threat system: Effects on
Neff, K. D., Rude, S. S., & Kirkpatrick, K. L. (2007b). An
jad.2014.08.006.
negative affect, shame, rumination, and depressive
examination of self-compassion in relation to positive
symptoms. Journal of Social and Clinical Psychology,
psychological functioning and personality traits.
32, 939–963. DOI:10.1521/jscp.2013.32.9.939.
Journal of Research in Personality, 41, 908–916.
Carl, J. R., Fairholme, C. P., Gallagher, M. W.,
Thompson-Hollands, J., & Barlow, D. H. (2014).
The effects of anxiety and depressive symptoms on
daily
positive
emotion
regulation.
Journal
of
Kohn, P. M., Lafreniere, K., & Gurevich, M. (1990). The
139–154. DOI:10.1016/j.jrp.2006.03.004.
DOI:10.1016/j.jrp.2006.08.002.
inventory of college students’ recent life experiences:
Pearson, M. R., Lawless, A. K., Brown, D. B., &
Psychopathology and Behavioral Assessment, 36(2),
A decontaminated hassles scale for a special
Bravo, A. J. (2015). Mindfulness and emotional
224–236. DOI:10.1007/s10862-013-9387-9.
population. Journal of Behavioral Medicine, 13,
outcomes: Identifying subgroups of college students
619–630. DOI:10.1007/BF00844738.
using
Farb, N. S., Irving, J. A., Anderson, A. K., & Segal, Z. V.
(2015). A two-factor model of relapse/recurrence
Kornstein, S. G., & Schneider, R. K. (2001). Clinical
vulnerability in unipolar depression. Journal of
features of treatment-resistant depression. Journal of
Abnormal Psychology, 124, 38–53. DOI:10.1037/
Clinical Psychiatry, 62, 18–25.
abn0000031.
latent
profile
analysis.
Personality
and
Individual Differences, 76, 33–38. DOI:10.1016/j.
paid.2014.11.009.
Radloff, L. S. (1977). The CES-D scale a self-report
Krejtz, I., Nezlek, J. B., Michnicka, A., Holas, P., &
depression scale for research in the general
Ferrari, A. J., Charlson, F. J., Norman, R. E., Patten, S. B.,
Rusanowska, M. (2014). Counting one’s blessings
population. Applied Psychological Measurement, 1,
Freedman, G., Murray, C. J. L., … Whiteford, H.A.
can reduce the impact of daily stress. Journal of
(2013a). Burden of depressive disorders by country,
Happiness Studies, . DOI:10.1007/s10902-014-9578-4.
sex, age, and year: Findings from the Global Burden
Leary, M. R., Tate, E. B., Adams, C. E., Batts Allen, A., &
of Disease Study 2010. PLoS Medicine, 10, e1001547.
Hancock, J. (2007). Self-compassion and reactions to
DOI:10.1371/journal.pmed.1001547.
unpleasant self-relevant events: The implications of
385–401.
Raes, F. (2011). The effect of self-compassion on the
development
of
depression
symptoms
in
a
non-clinical sample. Mindfulness, 2(1), 33–36.
DOI:10.1007/s12671-011-0040-y.
treating oneself kindly. Journal of Personality and
Rush, A. J., Trivedi, M. H., Wisniewski, S. R.,
Norman, R., Patten, S. B., Vos, T., … (2013b).
Social Psychology, 92(5), 887–904. DOI:10.1037/
Nierenberg, A. A., Stewart, J. W., Warden, D., …
Global variation in the prevalence and incidence
0022-3514.92.5.887.
(2006).
Ferrari, A. J., Somerville, A. J., Baxter, A. J.,
Acute
and
longer-term
outcomes
in
of major depressive disorder: A systematic review
MacBeth, A., & Gumley, A. (2012). Exploring
depressed outpatients requiring one or several
of the epidemiological literature. Psychological
compassion: A meta-analysis of the association
treatment steps: A STARD report. The American
Medicine, 43, 471–481.
between
self-compassion
Journal of Psychiatry, 163, 1905–1917. DOI:10.1176/
Clinical
Psychology
Fredrickson, B. L. (2001). The role of positive
emotions in positive psychology: The broaden-
and
Review,
psychopathology.
32,
545–552.
DOI:10.1016/j.cpr.2012.06.003.
appi.ajp.163.11.1905.
Schrank, B., Brownell, T., Tylee, A., & Slade, M. (2014).
and-build theory of positive emotions. American
McMakin, D. L., Siegle, G. J., & Shirk, S. R. (2011).
Positive psychology: An approach to supporting
Psychologist, 56(3), 218–226. DOI:10.1037/0003-
Positive Affect Stimulation and Sustainment (PASS)
recovery in mental illness. East Asian Archives of
066X.56.3.218.
module
Psychiatry, 24, 95–103.
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.
for
depressed
mood:
A
preliminary
Protective Factors for Depression
J. Ford et al.
Seligman, M. P., & Csikszentmihalyi, M. (2000). Positive
Stoltzfus, K. M., & Farkas, K. J. (2012). Alcohol use, daily
psychology: An introduction. American Psychologist,
hassles, and religious coping among students at a
55, 5–14. DOI:10.1037/0003-066X.55.1.5.
religiously affiliated college. Substance Use & Misuse,
Watson, D., & Naragon-Gainey, K. (2010). On the
Sin, N. L., & Lyubomirsky, S. (2009). Enhancing
47, 1134–1142. DOI:10.3109/10826084.2011.644843.
specificity of positive emotional dysfunction in
well-being and alleviating depressivesymptoms with
Tugade, M. M., & Fredrickson, B. L. (2007).
psychopathology: Evidence from the mood and
positive
psychology
interventions:
A
practice-
Regulation
friendly meta-analysis. Journal of Clinical Psychology,
regulation
65, 467–487. DOI:10.1002/jclp.20593.
Journal of Happiness Studies, 8(3), 311–333.
Spinhoven, P., Elzinga, B. M., Hovens, J. M., Roelofs, K.,
of
positive
strategies
that
Review, 35, 1440–1446. DOI:10.1016/j.childyouth.
2013.05.019.
emotions:
Emotion
anxiety
promote
resilience.
Clinical
disorders
and
Psychology
schizophrenia/schizotypy.
Review,
30,
839–848.
DOI:10.1016/j.cpr.2009.11.002.
Wood, A. M., & Joseph, S. (2010). The absence of
DOI:10.1007/s10902-006-9015-4.
van Oppen, P., Zitman, F. G., … (2011). Positive and
Vagi, K. J., Rothman, E. F., Latzman, N. E., Tharp, A. T.,
positive psychological (eudemonic) well-being as a
negative life events and personality traits in predicting
Hall, D. M., & Breiding, M. J. (2013). Beyond
risk factor for depression: A ten year cohort study.
course of depression and anxiety. Acta Psychiatrica
correlates: A review of risk and protective factors for
Journal
Scandinavica, 124, 462–473. DOI:10.1111/j.1600-
adolescent dating violence perpetration. Journal of
DOI:10.1016/j.jad.2009.06.032.
0447.2011.01753.x.
Steca, P., Abela, J. Z., Monzani, D., Greco, A., Hazel, N.
A., & Hankin, B. L. (2014). Cognitive vulnerability to
Youth and Adolescence, 42, 633–649. DOI:10.1007/
D.
R.,
Hotton,
Affective
Disorders,
122,
213–217.
World Health Organization. (2012). Depression Fact
Sheet [Online]. Retrieved June 24, 2015, from
s10964-013-9907-7.
Voisin,
of
A.
L.,
Tan,
K.,
&
http://www.who.int/mediacentre/factsheets/fs369/en/
Zuess, J. (2003). An
integrative
approach to
depressive symptoms in children: The protective role
DiClemente, R. (2013). A longitudinal examina-
of self-efficacy beliefs in a multi-wave longitudinal
tion of risk and protective factors associated with
depression: Part 1-- Etiology. Complementary
study. Journal of Abnormal Child Psychology, 42,
drug use and unsafe sex among young African
Health Practice Review, 8, 9–24. DOI:10.1177/
137–148. DOI:10.1007/s10802-013-9765-5.
American females. Children and Youth Services
1076167502238382.
Stress and Health (2016) © 2016 John Wiley & Sons, Ltd.