A Pilot Case-Control Study of the Social Media Activity Following Cluster and Non-Cluster Suicides in Australia

by Phillip Cheuk Fung Law, Lay San Too, Nicole T.M. Hill, Jo Robinson, Madelyn Gould, Jo-An Occipinti, Matthew J. Spittal, Katrina Witt, Mark Sinyor, Benedikt Till, Nathaniel Osgood, Ante Prodan, Rifat Zahan and Jane Pirkis

Published 29 December 2021


An estimated 6% of suicides occur in clusters; 6% of suicides by young people and 2% of those by adults. 'Contagion', the social transmission of suicidal behaviour following exposure to the suicide of another, is a mechanism by which suicide clusters are thought to occur. There is strong evidence that media can play a role in contagion, but little is known about how social media may play a role in suicide clusters.

Research and findings

This pilot study presented a methodological approach to examining whether Facebook activity following cluster and non-cluster suicides differed.

Two suicide clusters were identified by statistically scanning data obtained from the National Coronial Information System (NCIS). Each cluster case was then matched to 10 non-cluster suicides.

Public Facebook accounts of three of the 48 cluster cases were found, and 20 of the 480 non-cluster controls, as well as their public friends-list accounts. Social media activity was retrieved, and text segments that referred to the deceased and were written during the month after the date of death were included in the analysis.

Text segments were analysed in a number of ways:

  • Examined for “putatively harmful” and “putatively protective” content, for example discussion of the suicide method vs. messages discouraging suicidal acts: No “putatively harmful” or “putatively protective” content following any suicides was found.
  • Concept mapping: “Family” and “son” concepts were more common for cluster cases and “xx”, “sorry” and “loss” concepts were more common for non-cluster controls.
  • Word-emotion association and sentiment analysis: there were twice as many surprise and disgust-associated words for cluster cases.
  • Reactions-to-posts ratio (using emoji reactions): posts about non-cluster controls were four times as receptive as those about cluster cases.


These findings suggest there may be differences in social media posts between cluster and non-cluster suicides, that may contribute to contagion in cluster cases. There were a number of limitations of this study, such as the use of the Facebook platform only and public profiles, resulting in relatively limited number of text segments that may not represent the full scope of social media dialogue following suicide. The paper presents a new way to conduct this type of analysis of social media posts in examining contagion that can be built upon for future research.

Note: this paper includes example terms related to method.

#chatsafe guidelines are available to guide the safe online discussion of suicide for young people.