Problems posed by the Werther effect as a ‘net effect’: a comment on recent scholarly work on the effects of 13 Reasons Why

Summary

Recent research estimated that an additional 195 suicides among 10- to 17-year-old youths occurred following the release of the television series 13 Reasons Why. There is an underrecognised aspect in this line of research that this effect represents a net effect based on different possible underlying patterns (e.g. +195/−0 or +395/−200).

13 Reasons Why is a multi-episode fictional series on Netflix about struggles, including suicide, in the lives of high school students in the USA. The first season of the series, which was released on 31 March 2017, sparked a heated scholarly and public debate that focused on the fear that its explicit and graphic depiction of suicide might lead to increases in suicide rates, a phenomenon termed the Werther effect.¹

A recent study now provides evidence for an increase in US suicide rates following the release of season 1. On the basis of a macro-level observational analysis, Bridge and colleagues estimated that an additional 195 (95% CI 168–222) suicide deaths among 10- to 17-year-old youths occurred between 1 April and 31 December 2017 following the series' release.² Note that the confidence interval does not include zero, indicative of a significant Werther effect. The authors appropriately noted that causal claims regarding the underlying mechanism are a limitation of such macro-level observational studies. Nevertheless, the findings confirmed the fears that were raised when the series was first released.

The Werther effect is most often observed on a macro-level as an increase in suicide rates after highly visible stories in the media about persons dying by suicide. The evidence based on studies conducted over recent decades is convincing: the Werther effect is a real phenomenon. However, there is an underrecognised aspect in this line of research (and the related heated debates) that the Werther effect represents only a net effect. That is, Bridge and colleagues estimated 195 additional suicides. But what does this mean? And does it obscure a more complex social phenomenon?

Patterns underlying net effects

It is likely that not all viewers respond to the same media content in the same way, as is already well-known in communication scholarship. Taking the idea of a net effect seriously, other underlying patterns are possible. It is possible, for example, that the series influenced an ultimate outcome involving the potential deaths of n = 205 vulnerable individuals, of whom 200 died by suicide but 5 survived a suicidal crisis as a result of experiencing the suicide portrayal. This +200/−5 pattern results in a net effect of +195. Of course, more extreme patterns are also possible, such as +395/−200, also equalling a net effect of +195.

Research on the beneficial effects of suicide portrayals in the media, known as a Papageno effect, have also been identified.³ In this case, there is a net decrease in suicide rates after specific suicide reports (e.g. following a story about positive coping with a suicidal crisis). Similarly, one study on the effects of media guidelines regarding responsible reporting of suicide in Austria estimated a nationwide impact of following the guidelines, calculated as a significant reduction of 81 suicides (95% CI, −149 to −13) annually.⁴ The standard interpretation assumes that this equals −81 deaths by suicide, elicited by a +0/−81 pattern. However, it is also possible that media content influenced the deaths of, for example, n = 119 vulnerable individuals, with detrimental effects on 19 (death by suicide) and beneficial preventive effects on 100 (survival of a suicidal crisis), again resulting in a net effect of −81.

Consistent with this possibility, our recent individual-level survey study on the effects of the second season of 13 Reasons Why provides evidence for both harmful effects in some and helpful effects in others.⁵ Stated cautiously, it is not unlikely that such a salient media event can elicit both effects. Importantly, even a null net effect (e.g. +10/−10) can be indicative of substantial media effects. This raises the question, do we just want to know the net effects, or does the underlying pattern matter? We argue that it does.

Evaluating underlying patterns: a question of ethics

Our understanding of the underlying pattern is related to important ethical questions, relevant for scholarly work and health professionals. If the net effect is +195/−0 or +0/−81, the moral evaluation is straightforward. However, what about a net decrease of −81 following a +19/−100 pattern? What about a +395/−200 pattern? Is the latter Werther-effect pattern (which gives a net effect of 195, the number of additional suicide deaths reported after the first season of 13 Reasons Why) less serious than a +195/−0 pattern? This is a question of ethics. And this question is not limited to the study of the effects of 13 Reasons Why – it also holds for other media content.

The normative interpretation in current scholarly work on net effects seems to follow a consequentialist perspective: if more individuals are saved (have died), it is a good (bad) media effect. Alternatively, from a deontological perspective, causing the death of any individuals will still pose ethical problems, making it difficult to accept net effects without greater understanding of the underlying pattern. An evaluation following this perspective depends on one's predefined set of rules: what underlying pattern do we accept? We have no definitive answer but we, as a scholarly community, need to discuss this question in an open way.

Future research

Media effects are complex phenomena, as revealed by decades of scholarly work in communication and related disciplines. The Werther and Papageno effects are no exception. The fact that the Werther effect is a net effect operating at the macro-level deserves greater scholarly attention: the same suicide-related media content may elicit harmful and/or helpful effects at the individual level, depending on the interaction of content and audience characteristics, and may or may not translate into macro-level Werther/Papageno effects. We need to supplement the currently dominant methodological paradigm, for example, by conducting: (a) more individual-level (qualitative) interview studies of suicide attempters and loss survivors (i.e., family and friends), focusing on potentially harmful media content; (b) analyses of macro-level suicide rate data using grouping variables that are theoretically related to harmful and helpful effects (is there a Werther effect in some, and a Papageno effect in other, subgroups?), including a test of whether there was an increase in a certain method (Werther) that might have been offset by a decrease (Papageno) in other methods; and, when possible, (c) individual-level surveys with large general population samples both before and after the release of sensitive media content. These surveys may also focus on identification-related concepts: some content elements, such as bullying or sexual assault in 13 Reasons Why, may especially promote harmful identification processes between vulnerable individuals in the audience and the depicted suicide victim in the media content. These studies may also use supplementary data sources such as online information-seeking behaviour (e.g. number of searches for the suicide victim on internet search engines); characters who elicit greater identification may elicit more searches. Although each study type has its limitations, together they could enrich our understanding, hopefully leading to more confident interpretations of macro-level net effects. What is clear is that we need to put more effort into disentangling the underlying patterns.