Testing the Cycle of Avoidance Model of Non-Helpseeking: Social Comparison and Help-Negation in Deliberate Self-Injury

Self-injurious behaviour among young people is pervasive worldwide and associated with a range of deleterious psychological, physical health, vocational and quality of life outcomes. Furthermore, help-seeking rates are consistently low and those evidencing the most need are often the most likely to negate the help of both professionals and informal sources of support. The measurement and operationalisation of these phenomena has been impeded by methodological and definitional issues in the research field and the lack of a unified empirically validated theory of help-seeking. Help-negation effects have been established for components of suicidality including suicidal ideation, but this phenomenon has been relatively understudied in the context of deliberate self-injury (DSI) specifically. This thesis explored variables related to help-seeking for self-injury and developed an experimental behavioural task paradigm to empirically test a threshold theory of help-seeking, the Cycle of Avoidance (COA, Biddle et al., 2007). This theory suggests there is a threshold point for “real” distress that must be crossed prior to help-seeking initiation and a range of strategies are employed to delay this recognition and keep perceptions of distress in the “normal” frame (Biddle et al., 2007). The studies in this thesis endeavoured to apply the COA theory to address the preponderance of atheoretical studies in the help-seeking field and to also use a multi-method approach to assess help-seeking to mitigate prior methodological limitations such as shared method variance. A novel application of a psychophysical method to quantify help-seeking thresholds for DSI was undertaken. This approach, adapted from sensory perception research, offered several advantages over traditional self-report measures, including precise threshold quantification, mitigation of cognitive and response biases, timelimited responding and the experimental manipulation of variables. Study 1 was a proof-ofconcept study in which undergraduate university students (N = 183; the same participant group and dataset as used for Study 2) completed self-report questionnaires assessing help seeking and its correlates and completed a novel psychophysical computer task. The task depicted hypothetical scratches which measured threshold intentions for self and other helpseeking for this behaviour. Results showed 67% demonstrated a threshold, requiring on average 2.4 and 1.77 injuries for self and others respectively before endorsing help-seeking. This offered empirical support for a threshold theory of help-seeking and evidence of differences in thresholds for seeking help for self-versus others. Self-report questionnaires correlated with the threshold data, further validating the paradigm. Consistent with help negation effects, greater frequency of DSI was related to higher thresholds for seeking help, as was seeking help from no-one for suicidal ideation and personal/emotional problems. Individuals with greater social comparison tendencies had larger self-other threshold differences, endorsing help for others' self-injury more readily than for their own. Study 2 utilised the data captured from the same participant sample as Study 1, with both studies drawing from a single data collection time point and presents a different set of analyses conducted on the same data. Study 2 sought to examine whether social comparison processes and other variables contributed to help-negation. Participants completed self-report questionnaires assessing DSI and help-seeking variables. Hierarchical multiple regression analysis was used to identify predictors of help-seeking intentions from mental health professionals for suicidal ideation. A help-negation relationship was observed in that a greater history of self-injury was associated with lowered intentions to seek help. The tendency to rate one’s mental health as comparatively worse than others significantly contributed to the prediction of help-seeking intentions. However, contrary to expectations, participants’ ratings of comparatively worse mental health were associated with lower helpseeking intentions. Lower prior help-seeking, higher frequency of DSI and the worse participants’ saw their mental health in comparison to others were all associated with lower help-seeking intentions for suicidal thoughts. Study 3 was conducted using a university participant pool as in Studies 1 and 2, but these data were collected at a different time point, making it a separate and distinct phase of the research. This study used the same behavioural threshold method as in Study 1 but in addition to identifying correlates of help-seeking thresholds, it tested whether manipulation of normative information in the vignettes describing the self-injury could shift these thresholds in line with the COA theory. Three hundred and forty-six undergraduate university student participants completed self-report questionnaires measuring correlates of help-seeking and were randomized into two vignettebased conditions. One condition characterised DSI (a single scratch injury) as a “normal” response to cope with distress and the second as “abnormal” behaviour. Participants then completed the computer response task viewing images depicting deliberate superficial scratch injuries to capture behaviour-based help-seeking thresholds for self-injury. A help-seeking threshold was found for 47% of the sample. Higher thresholds for help-seeking were associated with lowered self-reported help-seeking intentions. However, no differences were observed in thresholds between the two normative information conditions. Endorsing helpseeking intentions for others was significantly higher than for self, with 35% of the variance in help-seeking intentions explained by whether this was self or other help-seeking. Inverse relationships between distress, self-injury engagement, and help-seeking thresholds indicated the presence of help-negation in this sample. Moderation analyses indicated that perceived mental health comparison weakly moderated links between psychological distress, selfinjury, and help-seeking intentions, yet the direction of these relationships were incongruent with theory-based expectations. The combined results established help-seeking thresholds for self-injury using a psychophysical behavioural task which aligned with a central assumption of the COA theory (Biddle et al., 2007). The establishment of self-other differential helpseeking thresholds provides empirical support for the role of social comparison processes in normalising symptoms and potentially producing delays in help-seeking. Results also suggest that some tenets of the COA require further testing. Specifically, the inability to move thresholds via simple vignette manipulations which framed self-injury as either “normal” or “abnormal” raises questions about the importance of this categorical process in help avoidance. Unique factors related to functionality of self-injury, nuanced stigma and shame and refractory help-negation effects may mean the “real” and “normal” distress categorisations in determining help-seeking decisions in the COA are insufficient to shift thresholds for help-seeking. The research highlights the necessity to adapt existing theories to better represent multidimensional barriers and facilitators specific to sensitive behaviours like self-injury and to consider non-linear pathways to seeking help.