A nuanced examination of the relationship between social connectedness and suicidal ideation


The link between suicide-related outcomes, particularly suicidal ideation, and social belongingness has long been discussed, beginning with the writings of Durkheim over a century ago (Durkheim, 1897). Since, more recent theories of suicide have incorporated social belongingness or connectedness as core components in the development of suicidal ideation (Klonsky & May, 2015; Joiner, 2005). While these associations have garnered significant empirical support, the large majority of this research relies on the use of Interpersonal Needs Questionnaire (INQ; Van Orden et al., 2010) to measure social connectedness. The INQ has allowed for instrumental contributions to the suicide literature and suicide risk prediction; however, given the inherent limitations of Likert-type items and the broad nature of INQ items (e.g., “I feel like I belong), nuanced information about feelings of belongingness (i.e., cognitive-affective reactions, relationship-specific experiences), and their impact on suicidal ideation, may be missed. The use of open-ended questions may capture this level of detail, allowing for a more accurate prediction of suicidal ideation. To test this hypothesis, we collected data from an online sample of 464 individuals (136 with a history of suicidal ideation). Participants were asked to provide a text-based response to the question, “How connected do you feel with those around you? How would you describe your relationships?” and complete the INQ-Short Form (which focuses on social connectedness). We utilized an extension of a supervised topic model that simultaneously incorporates topics from text-responses and other predictors (i.e., INQ score) to predict suicidal ideation. Ten topics were extracted. Controlling for all other topics and INQ score, one topic demonstrated a positive coefficient and credible intervals not including zero (3.42 [SD = 0.49]). Representative (stemmed) words for this topic were: “remarri”, “children”, “tend”, “father”, “dad”, “walk”, “relationship”, “divorce” and “afraid”. Controlling for the extracted latent topics, the INQ demonstrated a negative coefficient and credible intervals not including zero (-0.11 [SD=.001]). Results highlight the potential advantages of utilizing text-based responses in the prediction of suicidal ideation above and beyond traditional Likert-type scales. More specifically, findings demonstrate that the discussion of terminated relationships or relationships that invoke feelings of fear may be particularly relevant in the experience of suicidal ideation.

Nov 21, 2019 — Nov 24, 2019
Kenneth Tyler Wilcox
Kenneth Tyler Wilcox
Statistical Consultant

My research interests include integrative data analysis, meta-analysis, topic modeling, Bayesian statistics, multilevel modeling, statistical programming, and psychology.