Identifying therapeutic characteristics of digital social media narratives about suicide: a mixed methods investigation

Abstract

Recently, we found adults who received digital bibliotherapy featuring lived-experience narratives related to suicidal thoughts and behaviors reported lower suicidal thoughts, mediated by increased social connectedness and optimism. This study aimed to identify characteristics of narratives associated with decreased suicidal thinking and increased social connectedness and optimism. 1532 users of a social media platform responded to surveys before and after reading narratives. Mixed-methods content analysis and clustered multidimensional scaling tested whether clusters that shared narrative characteristics were related to suicidal thoughts, social connectedness and optimism. We identified three narrative clusters: Cluster 1: "Encouraging Readers to Live," Cluster 2: "Sharing Personal Stories," and Cluster 3: "Detailed Accounts." Clusters 2 and 3 were associated with greatest reduction in suicidal thoughts, Cluster 2 with the greatest increase in social connectedness, and Cluster 3 with the greatest increase in optimism. Results suggest the optimal narratives for reducing suicidal thoughts are personal, detailed accounts.

Fig. 1. Distributions of Pre-Narrative Outcome Ratings.

Distributions in the top row (A, top left; B, top center; C, top right) are participants who reported that current suicidal thoughts motivated them to read the narrative. Distributions on the bottom row (D, bottom left; E, bottom center; F, bottom right) are participants who did not report that suicidal thoughts motivated them to read the narrative.

Fig. 2. Clustered Multidimensional Scaling Solution Dendrogram and Configuration Plot.

A(left panel) shows a dendrogram illustrating hierarchical relationships between the suicide narratives based on each of the 77 qualitative codes. The different narratives are represented on the horizontal axis, each with a separate letter indicator. The vertical axis, height, represents Euclidean distance, a measure of multidimensional similarity between the narratives. Narratives connected at greater heights are more dissimilar than narratives connected at lower heights. Each of the three clusters is indicated with a colored box. The color of the indicator boxes and each narrative point in the figure corresponds to the narrative’s cluster membership. B(right panel) shows a configuration plot illustrating the similarities between the narratives, based on the 77 subcategory codes, in two-dimensional space. The two-dimensional distance between any pair of narratives is equivalent to the Euclidean distance between those narratives. Narratives appearing closer together are more similar than narratives that are farther apart. The mean qualitative content of each cluster by code is depicted in Fig. 2.

Fig. 3. Mean Percentage of Content within each Code by Story Cluster.

No note included.