Elsevier

Social Science & Medicine

Volume 69, Issue 7, October 2009, Pages 1085-1090
Social Science & Medicine

Copycat effects after media reports on suicide: A population-based ecologic study

https://doi.org/10.1016/j.socscimed.2009.07.041Get rights and content

Abstract

This study aimed to investigate whether the risk of increased suicide occurrence after reports on suicide is associated with the social characteristics of the reported suicides and whether this varies with similarity between the reported suicides and suicides in the population. We collected reports on all 179 individual suicides named in the 13 largest Austrian nationwide newspapers from 1996 to 2006. Information on social status and sociodemographic characteristics of the reported suicides, on certainty of labelling the death as a suicide, and on the suicide methods applied were extracted from the articles. We conducted logistic regression analyses, with the increase of post-report suicides within 28 days after the reports as dependent variable. In model 1, the increase of suicides that matched the reported individual suicide with regard to age group, sex and suicide method was used as outcome variable. In model 2, the increase of suicides that were different from the reported suicide with regard to these characteristics was the outcome. In model 3, the post-report increase of total suicides was the dependent variable. Celebrity status of the reported suicide, age of the reported suicide between 30 and 64 years, and definitive labelling as a suicide were associated with an increased risk of a post-report increase of similar suicides; criminality (i.e. the individual was reported as suspected or convicted of crime) of the reported suicide was associated with a lower risk of a post-report increase. In dissimilar suicides, none of the variables was associated with a post-report increase of suicides. Celebrity status of the reported suicide was the only predictor of a post-report increase of total suicides. The findings support the hypothesis that social variables of reported suicides impact the risk of post-report copycat behaviour. Evidence of copycat effects seemed to be strongest in suicides that were similar to the respective model with regard to age group, sex, and suicide method.

Introduction

The possibility that mass media reporting on suicide has an impact on suicidal behaviour in the population is a controversially discussed public health issue. On one hand, it has been discussed that media discourse supports suicide preventive efforts when it educates the public about the treatment of mental ill-health and suicidality (Hegerl et al., 2006, Niederkrotenthaler and Sonneck, 2007, Sonneck et al., 1994, World Health Organization, 2008). On the other hand, other evidence suggests that suicide-related reporting, especially of non-fictional individual suicides, triggers additional suicides: so-called imitative effects (Cheng et al., 2007, Cheng et al., 2007, Etzersdorfer et al., 2001, Jonas, 1992, Phillips, 1974, Schmidtke and Häfner, 1988, Stack, 1987, Tousignant et al., 2005, Wasserman, 1984). In spite of the wide-spread assumption that copycat effects following media reports are a real phenomenon, there are severe inconsistencies in available research results. A substantial number of studies have not identified any evidence of a post-report increase in suicides (Mercy et al., 2001, Stack, 2005). As Stack notes, such inconsistencies may be due to the large variation in study methodologies (Stack, 2005), which may be attributable to a lack of theoretical guidance in research on the phenomenon (Marsden, 2000, p. 84).

Most frequently, studies on copycat behaviour build on social learning theory (Bandura, 1986, Blood and Pirkis, 2001). Bandura's theory defines identification of the imitator with the model as a necessary prerequisite of imitative behaviour (Bandura, 1986, Blood and Pirkis, 2001). However, concepts of identification have rarely been considered in research on copycat effects (Blood and Pirkis, 2001, Stack, 1992). The theory of differential identification suggests that people tend to identify with someone who is socially superior, as in the case of celebrities (“vertical identification”: Blood and Pirkis, 2001, Stack, 1992). This is supported by the large number of case studies that have found an increase in suicides following reports on celebrity suicides. Furthermore, the concept of differential identification also suggests that social similarity between imitator and model (e.g. same age group and sex) is an important factor in identification and in imitative behaviour (referred to as “horizontal identification”: Blood and Pirkis, 2001, Stack, 1992). Evidence of associations between social status (e.g. celebrity status) and the risk of post-report increase in suicides may be primarily present in subgroups of suicides that are similar to the respective models. We conducted the present study in order to test associations of social model characteristics and reporting characteristics with post-report increases of total suicides and additionally with suicides that were similar and dissimilar to the respective reported suicides.

Section snippets

Social status of the models

The concept of vertical identification suggests that people are more likely to identify with individuals who are socially superior, which is the case in celebrities. By contrast, as suggested by Stack (1987), criminality in the suicide case may indicate a lower social status compared to the majority of the population, and thereby decrease the probability of finding evidence of copycat effects.

Hypothesis (a): Celebrity status is associated with an increased probability of a post-report increase

Material and methods

We obtained daily data on suicides from July 1996 to September 2006, with sex, age groups, and suicide methods recorded according to the International Classification of Diseases-9 (ICD-9) and ICD-10 classifications, from Statistics Austria.

All media reports in the 13 most widely read nationwide Austrian newspapers that included the word “self-murder” or “suicide” were collected from the Austrian Press Agency for the same time period. In total, these newspapers reach regularly 74.2% of the

Results

For a display of the moving count of total suicides and of the total number of media items in the preceding year for each month from July 1997 to September 2006, see Fig. 2.

For 176 (98.3%) of the identified 179 individual suicides, complete information on the independent variables could be extracted. 165 (93.7%) were male and 11 (6.3%) were female. 14 individuals (8.0%) belonged to the age group of adolescents and young adults up to 29 years, 136 (77.3%) were between 30 and 64 years of age, and

Discussion

The present study constitutes the first to investigate associations of social status of reported suicides in terms of both celebrity and criminality status with the probability of a post-report increase of suicides. Based on the theory of vertical identification, celebrities are more appropriate as models for imitation because of their perceived social superiority, whereas criminals may be perceived as open to social disapproval and therefore inferior, which is likely to result in a decreased

Acknowledgements

The Austrian Academy of Sciences supported this study (grant number 70036 to T.N. and 70034 to B. T.).

References (32)

  • E. Durkheim

    Suicide. A study in sociology

    (1997)
  • E. Etzersdorfer et al.

    A dose-response relationship of imitational suicides with newspaper distribution

    Australian and New Zealand Journal of Psychiatry

    (2001)
  • M.S. Gould et al.

    Suicide clusters: an examination of age-specific effects

    American Journal of Public Health

    (1990)
  • B. Hazard Munro

    Statistical methods for health care research

    (2005)
  • U. Hegerl et al.

    The alliance against depression: 2-year evaluation of a community based intervention to reduce suicidality

    Psychological Medicine

    (2006)
  • K. Jonas

    Modelling and suicide: a test of the Werther effect

    British Journal of Social Psychology

    (1992)
  • Cited by (0)

    View full text