Low self-perceived resilience mediates the link between limited access to emotion regulation strategies and non-suicidal self-injury

Abstract

Having limited access to effective emotion regulation (ER) strategies, a construct commonly measured using the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004), has been found to be strongly linked with non-suicidal self-injury (NSSI). However, the mechanisms that explain this association need more in-depth exploration. In this study, we investigated whether a cognitive mechanism (low self-perceived resilience) and/or a behavioral mechanism (insufficient repertoire of strategies used to resist NSSI) explained the association between limited access to ER strategies and the likelihood of recent (past-month) NSSI. Research questions were explored in a sample of 922 university students with a lifetime NSSI history (83.3% female). Study variables were measured using self-report questionnaires. Controlling for anxiety/depression, a path analytic model revealed that only low self-perceived resilience mediated the association between limited access to ER strategies and past-month NSSI. Participants who endorsed having limited access to ER strategies unexpectedly reported a higher number of coping strategies to resist NSSI. The results support the relevance of exploring resiliency beliefs in the association between ER capability and NSSI.

Keywords:

• self-injury
• emotion dysregulation
• resilience beliefs

1.

Difficulties with emotion regulation (ER), broadly defined as deficits in the monitoring, evaluation, modulation, and expression of emotions (Gross, 1998; Thompson, 1994), are widely understood to drive engagement in non-suicidal self-injury (NSSI; ie self-inflicted damage of bodily tissue performed without suicidal intent; Nock & Favazza, 2009). NSSI is, relatedly, most frequently performed and negatively reinforced as a strategy to regulate distressing emotions (Taylor et al., 2018). Accordingly, multiple components of emotion dysregulation have garnered theoretical (see Hasking et al., 2017) and empirical (Wolff et al., 2019) attention as risk factors associated with the onset and maintenance of NSSI behavior. Among these, having difficulty selecting and applying effective strategies to regulate emotions in times of distress is an ER deficit that stands out as a particularly intuitive correlate of NSSI. A recent meta-analysis found that the association between this ER deficit and NSSI was of medium-to-large magnitude, and greater than all other ER constructs explored (Wolff et al., 2019; see also You et al., 2018). Researchers thus concluded that having poorly developed strategies to regulate emotions may be uniquely linked with NSSI risk. While this body of work underscores the salience of this ER construct a risk factor for NSSI, the precise mechanisms that account for the strength of this association need further investigation.

To further explicate this association, paying attention to how the ER construct has been measured in research offers a starting point. As shown in the Wolff et al. (2019) meta-analysis, the “Limited access to ER strategies” subscale from the Difficulties in Emotion Regulation Scale (DERS; Gratz & Roemer, 2004) is most commonly used to measure this construct. The DERS subscale uses items such as “When I’m upset, I believe that there is nothing I can do to make myself feel better” and “When I’m upset, I believe that wallowing in it is all I can do” to measure one’s difficulty in using strategies flexibly and as situationally appropriate in order to modulate emotions (Gratz & Roemer, 2004). The way this ER construct is conceptualized by researchers has varied somewhat in the literature. On the one hand, it may be conceptualized as a behavioral skill deficit, reflecting the individual’s actual experience that they have few strategies at their disposal to regulate emotions (Wolff et al., 2019; You et al., 2018). On the other hand, it may also reflect a cognitive factor, namely the individual’s negative beliefs about their own ER capability (Kiekens et al., 2017; Perez et al., 2012; You et al., 2018). Indeed, besides having difficulty behaviorally enacting strategies to regulate emotions and resist NSSI, holding negative expectancies about one’s ability to regulate emotions may also be important. Held rigidly across situations, such expectancies may generalize into globally negative resiliency beliefs (ie viewing oneself as unable to manage and recover from stress) which, in turn, would likely play a role in sustaining a reliance upon NSSI as a regulation strategy. This hypothesis is consistent with emerging theoretical and empirical elaborations highlighting the role of cognitive processes in NSSI vulnerability. In particular, the cognitive-emotional model of NSSI (Hasking et al., 2017) suggests that negative self-efficacy expectancies—namely the belief that one lacks the ability to cope and will be unable to resist the use of NSSI—may be particularly important in maintaining the behavior. While there is emerging work in support of this model (eg Dawkins et al., 2022), limited research to date has applied this cognitive perspective as a framework for understanding the relationship between this particular ER construct (as measured by the DERS subscale) and NSSI.

In this study, we sought to further unpack the association between the DERS subscale “Limited access to ER strategies” and the likelihood of recent (past-month) NSSI. Specifically, we explored if this association operated indirectly through a behavioral mechanism (ie repertoire of strategies used to resist NSSI) and/or a cognitive mechanism (ie low self-perceived resilience), and we hypothesized a significant indirect effect through the latter. We explored this question in a sample of individuals with a history of NSSI while controlling for anxiety/depression symptoms, because both resilience beliefs (Alessandri et al., 2015) and past-month NSSI (Bentley et al., 2015) are expected to be influenced by one’s levels of internalizing distress.

2. Method

2.1. Participants

A total of 2,811 young adults aged 17–25 years responded to a large survey online in exchange of course credits through a research participant program at a Canadian university. A total of 232 participants were removed due to failing to meet basic quality assurance requirements (ie insufficient questionnaire completion, failing multiple validity screeners embedded in survey, completion time below minimum required). From the remaining 2,579 participants, a total of 922 university students were used in our analyses because they endorsed a lifetime history of NSSI on the Ottawa Self-Injury Inventory (see below) and thus had data on NSSI behaviors (83.3% female, 16.1% male, 0.5% other; 70.0% White; M_age = 18.89; SD_age = 1.52). The distribution of students’ lifetime NSSI frequency was 1–3 times (35.03%), 4–6 times (16.49%), 7–10 times (14.43%), and more than 10 times (33.51%). Participants used a range of NSSI methods, most commonly scratching (68.75%), cutting (67.14%), and hitting (43.82%). The average age of onset was 14.56 years (SD = 2.30). Students were mainly enrolled in their first (68.8%), second (18.5%), and third (7.6%) years of university and came from diverse programs across the sciences, social sciences, and arts. All participants provided their informed consent and study procedures received ethics approval.

From this lifetime NSSI sample, 166 students (84.3% female, 13.9% male, 1.8% other; M_age = 18.77; SD_age = 1.57) reported engaging in NSSI as recently as within the past month. The majority of these students reported self-injuring 1–3 times (78.3%) or 4–6 times (12.0%) in the past month. The majority also used multiple NSSI methods (M = 3.26; SD = 1.62), including scratching (74.1%), cutting (71.1%), hitting (56.6%), biting (39.2%), piercing the skin with sharp/pointy objects (33.7%), and burning (21.5%).

2.2. Measures

Past-month NSSI was assessed using the Ottawa Self-Injury Inventory (Cloutier & Nixon, 2003) based on the question “In the past month, have you purposefully injured yourself without the intention to kill yourself?” (yes = 1, no = 0). An analogous question (“In your lifetime […]”) was used to identify participants’ lifetime NSSI status. The OSI is a validated instrument for NSSI assessment (see Guérin-Marion et al., 2018 for a summary).

Limited access to ER strategies was measured using the corresponding 8-item subscale from the DERS (Gratz & Roemer, 2004). Items from the subscale (α=.86) are rated on a 5-point likert scale, with higher mean scores reflecting greater difficulties accessing ER strategies.

Low self-perceived resilience. The Brief Resilience Scale (Smith et al., 2008) was used as a measure of respondents’ perception of their own ability to manage adversity and stress (α=.84; 6 items; eg I have a hard time making it through stressful events). Items are scored on a 5-point likert scale. Positive items (1, 3, 5) were reverse-scored such that higher mean scores reflected lower self-perceived resilience.

Strategies used to resist NSSI were also measured using the OSI. In response to the question If you are trying to resist hurting yourself, what do you do instead?, participants were asked to provide all strategies they use to resist NSSI urges from a list: “Talk with someone”; “Exercise/sports”; “Reading writing, music, dance”; “Watch television, play video or computer games”; “Do things to relax”; and/or “Do anything to keep hands busy”. Strategies were summed (0–6).

Anxiety/Depression symptoms were measured using the Psychiatric Symptoms Index (Ilfeld, 1976) using a combined mean score (α=.92). This score was included as a covariate in the model given its significant links with both self-perceived resilience (r=.22; p < .001) and past-month NSSI (r=.24; p < .001). The PSI has shown good psychometric properties (Okun et al., 1996).

2.3. Data analysis plan

Analyses were run by way of path analysis in Mplus 8.6. A fully saturated single logit model (using the “link=LOGIT” command) with direct and indirect effects (5000 bootstrap resampling draws) was estimated using maximum likelihood (ML) estimation and full information maximum likelihood (FIML; 0.3–0.6% missing data). Specifically, we designed the multiple-mediation model with low self-perceived resilience and coping strategies to resist NSSI as both mediators of the association between limited access to ER strategies and past-month NSSI. The “INDIRECT” command was used to model indirect effects. Note that while models with binary outcomes can also be run using WLSMV estimation, we chose ML estimation to account for the mixed distributions of endogenous variables in our model (see Muthén et al., 2017). Specifically, paths predicting the binary NSSI variable used logistic regression, while paths predicting the continuous mediating variables (low self-perceived resilience and repertoire of strategies used to resist NSSI) used linear regression. Note that the running of this model in ML and FIML required the use of a numerical integration algorithm. To control for any variance accounted for by internalizing distress, paths between all variables and anxiety/depression symptoms were included.

3. Results

Descriptive statistics are in Table 1. No significant issues with non-normality or multicollinearity were detected; as shown in Table 1, variance inflation factors fell well below recommended cut-offs (ie 1.95–1.02; see Allison, 1999).

Table 1. Descriptive statistics and intercorrelations among study variables

Variable	NSSI in past month (n = 166)		No NSSI in past month (n = 754)		VIF	1.	2.	3.	4.
	M	SD	M	SD
1. Limited Access to ER Strategies	3.38	.89	2.88	.90	1.95	-
2. Low Self-Perceived Resilience	3.53	.82	3.11	.81	1.56	.45**	-
3. Repertoire of Coping Strategies Used to Resist NSSI	3.93	1.64	3.96	1.62	1.02	.11*	.14**	-
4. Anxiety/Depression	2.80	.67	2.36	.66	1.52	.36**	.23**	.14**	-

Notes: Intercorrelations are presented for the full sample (n = 922) with missing data handled using full information maximum likelihood. *p < .05, ** p < .01. VIF = variance inflation factor.

Model results are detailed in Figure 1. Controlling for anxiety/depression, limited access to ER strategies (DERS subscale) was associated with a higher likelihood of past-month NSSI, as shown by a significant total effect, β = .14, SE=.06, 95%CI[.03, .26]. Controlling for anxiety/depression, the model revealed one significant indirect effect; specifically, having limited access to ER strategies was associated with past-month NSSI indirectly through lower self-perceived resilience (standardized indirect effect = .09, p=.021, 95%CI[.02, .17]). Specifically, having limited access to ER strategies was positively linked with low self-perceived resilience (β=.59, SE=.02, 95%CI[.55, .64]), which in turn was linked with a higher risk of past-month NSSI (OR = 1.42, SE=.22, 95%CI_OR[1.07, 2.00]). By contrast, the indirect effect through students’ repertoire of coping strategies used to resist NSSI was not significant (standardized indirect effect=−.01, SE=.01, p=.28, 95%CI[−.02, .002]). Specifically, having more limited access to ER strategies was linked to participants endorsing a more extensive repertoire of coping strategies used to resist NSSI (β=.08, SE=.03, 95%CI[.01, .14]), but repertoire of coping strategies was not linked with NSSI engagement (OR = .92, SE=.06, 95%CI_OR[.82, 1.04]). After accounting for indirect effects, the direct effect of the DERS subscale on NSSI was not significant, β = .06, SE=.07, p=.40, 95%CI[−.08, .20].

Figure 1. Path analytic model.

Graphical representation of the path analytic model. The model shows that Limited Access to Emotion Regulation Strategies is positively linked with past-month non-suicidal self-injury indirectly through low self-perceived resilience.

Path analytic model with standardized path coefficients, predicting past-month NSSI in a sample of participants with a lifetime history of NSSI. The model controls for anxiety/depression. Bootstrapped 95% confidence intervals are represented in brackets for the direct and total effects. Non-significant paths are represented with dotted lines. NSSI=non-suicidal self-injury. DERS=Difficulties in Emotion Regulation Scale. OR=odds ratio.

* p < .05, ** p < .01, *** p < .001

4. Discussion

In a sample of university students with a lifetime history of NSSI, we found that the association between one’s limited access to ER strategies (as measured by the DERS) and the likelihood of past-month NSSI was mediated by a self-perceived sense of low resilience, holding equal the number of coping strategies participants used to resist NSSI and their anxiety/depression symptoms. This suggests that the relationship between this ER construct and NSSI may be explained by a cognitive component (ie the extent to which the individual believes they lack the capacity to cope with distressing emotions), rather than the individual’s actual repertoire of coping strategies. This finding aligns with emerging research on the cognitive-emotional model of NSSI (Dawkins et al., 2022; Hasking et al., 2017) showing that negative self-efficacy expectancies are related to increased risk for NSSI, including a sustained course of NSSI. However, such findings have been mixed (Midkiff et al., 2018) and thus deserve continued exploration.

Contrary to expectations, we found that having a greater repertoire of coping strategies to resist NSSI was unrelated to the likelihood of engaging in past-month NSSI. Moreover, those who reported a higher number of coping strategies to resist NSSI endorsed having a worse capacity to access ER strategies on the DERS subscale (as well as lower self-perceived resilience; see Table 1). One explanation for this finding is that individuals who reported using a higher number of coping strategies to resist NSSI may have done so because they are in fact struggling to find what works for them. They may be struggling to apply ER strategies in the moment, with sustained effort and/or flexibly based on context. Indeed, one limitation of the scale is that it does not measure the perceived helpfulness of the endorsed coping strategies. Nevertheless, this finding may align with the proposition that cessation of NSSI goes beyond expanding one’s behavioral repertoire of ER strategies, and likely hinges upon a broader cognitive-representational shift. In other words, individuals will still struggle to resist NSSI urges in spite of having been taught multiple coping strategies if they hold negative beliefs about their ability to cope with distress. Paying attention to the global negative beliefs that may be compromising the helpfulness of coping strategies used to resist NSSI appears important, including for clinicians working with young people who self-injure.

Study limitations include our cross-sectional design (which precludes conclusions about the true direction of effects), the gender imbalance of the sample (which limits the generalizability of findings to males), and our use of self-report data (which implicates the possibility of respondent bias). Accordingly, future exploration of current hypotheses would benefit from using more gender-balanced samples and alternative longitudinal methods, including ecological momentary assessment (eg daily diaries) for a richer and sequential understanding of the mechanisms under study. Moreover, our list of coping strategies used to resist NSSI was non-exhaustive, did not assess for the perceived helpfulness of strategies, and was taken from the same instrument used to measure NSSI. Therefore, the use of a separate and more comprehensive checklist of coping strategies would have been ideal. It should also be noted that we did not use a measure meant explicitly to measure beliefs about resilience, although this was considered to be implicit.

5. Conclusion

Our findings make the case for conceptualizing the link between the DERS subscale “limited access to ER strategies” and NSSI as nuanced, and as tapping a significant cognitive component. Exploring beliefs-based components in ER capability, such as resiliency beliefs, may hold promise for both NSSI research and intervention with young people struggling with NSSI. Facilitating gradual shifts in individuals’ resiliency beliefs and confidence in their own coping ability through therapeutic intervention (eg cognitive restructuring, process-experiential techniques) may be particularly important in supporting improved emotion regulation and, in turn, NSSI cessation and recovery.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The raw data analyses that support the findings of this study are available upon reasonable request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions.

Additional information

Notes on contributors

Camille Guérin-Marion

Camille Guérin-Marion holds a Ph.D. in Clinical Psychology from the University of Ottawa, located in Ontario, Canada. Her research focuses on non-suicidal self-injury in youth, emotion regulation, and parent-child relationships and attachment.

Jean-François Bureau

Jean-François Bureau is a full professor in the School of Psychology at the University of Ottawa. His research focuses on father-child and mother-child attachment, children’s socioemotional development, and non-suicidal self-injury and psychopathology.

Marie-France Lafontaine

Marie-France Lafontaine is a full professor in the School of Psychology at the University of Ottawa. Her research focuses on adult attachment, intimate partner violence, non-suicidal self-injury, heterosexual and gay couple relationships, and physical health.

Patrick Gaudreau

Patrick Gaudreau is a full professor in the School of Psychology at the University of Ottawa. His research focuses on aspects of personality (perfectionism, optimism), motivation, self-regulation, and coping.