Colm Sweeney, Edel Ennis, Maurice Mulvenna, Raymond Bond, Áine O'Meara
{"title":"Assessing Patterns of Nonadherence With Reporting Guidelines for Articles Relating to Suicide in National Media.","authors":"Colm Sweeney, Edel Ennis, Maurice Mulvenna, Raymond Bond, Áine O'Meara","doi":"10.1027/0227-5910/a001023","DOIUrl":null,"url":null,"abstract":"<p><p><b></b> <i>Background:</i> Certain forms of media reporting on suicide, particularly those involving celebrity suicides, have been associated with an increase in imitative suicidal behavior. <i>Aims:</i> This research employs data science methodologies, including association rule mining, to identify patterns of nonadherence to reporting guidelines in Irish national media articles about suicide. <i>Methods:</i> This study analyses a dataset of 2,939 articles identified as noncompliant with Samaritans' reporting guidelines, using association rule mining to explore relationships among key categories of nonadherence. Initial analysis was conducted on the full dataset, followed by comparative analysis of celebrity-related and noncelebrity-related subsets to examine potential differences in reporting practices. <i>Results:</i> The most frequent recommendation not adhered to was the provision of helpline information, with higher nonadherence rates observed in celebrity-related articles and an increasing trend from 2019 to 2021. The second most common nonadherence involved the use of the term <i>commit</i>. Among those articles not adhering to two or more recommendations, association rule mining revealed co-occurrence patterns, which differed between celebrity-related media reports and noncelebrity-related reports. The top rules were linkage between nonadherences of the recommendations surrounding Graphic images and Methodology - Suicide Or Self-Harm within celebrity-related media, and linkage between nonadherences of the Methodology - Suicide Or Self-Harm and Location Details recommendations within noncelebrity-related media. <i>Limitations</i><i>:</i> The dataset represents only 20% of guideline-breaking suicide articles, manually tagged with high inter-rater reliability, but excludes positive practices, full guideline coverage, and distinctions between ideation, attempt, or death. <i>Conclusions:</i> Results are discussed in terms of their implications for supporting journalists in adhering to principles of the responsible reporting of suicide within the media, and the need to address celebrity-related and noncelebrity-related reporting separately. Relevant challenges are also discussed.</p>","PeriodicalId":47943,"journal":{"name":"Crisis-The Journal of Crisis Intervention and Suicide Prevention","volume":" ","pages":""},"PeriodicalIF":2.6000,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Crisis-The Journal of Crisis Intervention and Suicide Prevention","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1027/0227-5910/a001023","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"PSYCHIATRY","Score":null,"Total":0}
引用次数: 0
Abstract
Background: Certain forms of media reporting on suicide, particularly those involving celebrity suicides, have been associated with an increase in imitative suicidal behavior. Aims: This research employs data science methodologies, including association rule mining, to identify patterns of nonadherence to reporting guidelines in Irish national media articles about suicide. Methods: This study analyses a dataset of 2,939 articles identified as noncompliant with Samaritans' reporting guidelines, using association rule mining to explore relationships among key categories of nonadherence. Initial analysis was conducted on the full dataset, followed by comparative analysis of celebrity-related and noncelebrity-related subsets to examine potential differences in reporting practices. Results: The most frequent recommendation not adhered to was the provision of helpline information, with higher nonadherence rates observed in celebrity-related articles and an increasing trend from 2019 to 2021. The second most common nonadherence involved the use of the term commit. Among those articles not adhering to two or more recommendations, association rule mining revealed co-occurrence patterns, which differed between celebrity-related media reports and noncelebrity-related reports. The top rules were linkage between nonadherences of the recommendations surrounding Graphic images and Methodology - Suicide Or Self-Harm within celebrity-related media, and linkage between nonadherences of the Methodology - Suicide Or Self-Harm and Location Details recommendations within noncelebrity-related media. Limitations: The dataset represents only 20% of guideline-breaking suicide articles, manually tagged with high inter-rater reliability, but excludes positive practices, full guideline coverage, and distinctions between ideation, attempt, or death. Conclusions: Results are discussed in terms of their implications for supporting journalists in adhering to principles of the responsible reporting of suicide within the media, and the need to address celebrity-related and noncelebrity-related reporting separately. Relevant challenges are also discussed.
期刊介绍:
A must for all who need to keep up on the latest findings from both basic research and practical experience in the fields of suicide prevention and crisis intervention! This well-established periodical’s reputation for publishing important articles on suicidology and crisis intervention from around the world is being further enhanced with the move to 6 issues per year (previously 4) in 2010. But over and above its scientific reputation, Crisis also publishes potentially life-saving information for all those involved in crisis intervention and suicide prevention, making it important reading for clinicians, counselors, hotlines, and crisis intervention centers.