The Press and Suicides in the 19th Century: Investigating Possible Imitative Effects in Five Territories of the Austro-Hungarian Empire

Abstract

During the 19th century, suicide rates increased in many countries. The press may have contributed to this increase, even though empirical evidence is lacking in this regard. We assessed suicide statistics within five territories of the Austro-Hungarian Empire between 1871 and 1910 and combined these data with a content analysis of suicide reporting in five newspapers, each appearing in one of the five territories. The analysis revealed a covariation between the quantity of reporting and the number of suicides within all five regions. Furthermore, the quantity of reporting significantly predicted the following year’s suicides. Although the causal order of suicides and the quantity of reporting should be assessed with caution, evidence is consistent with the idea that the press may have contributed to the establishment of suicide as a mass phenomenon. The findings also support contemporary guidelines for journalists, especially the notion of avoiding undue repetition of suicide stories.

Keywords

suicide Werther effect 19th century media guidelines suicide reporting

Among others, Durkheim (1897/2014) documented that the suicide rates increased massively in many countries during the 19th century. For example, within a period of a few decades, suicide became a mass phenomenon in the Austro-Hungarian Empire during the second half of the 19th century (Sonneck, Kapusta, Tomandl, & Voracek, 2012). Importantly for this study, contemporary scholarship has revealed suicide reporting to be a risk factor for imitative suicides, termed the “Werther effect” (Phillips, 1974; Stack, 2005). Although some studies have investigated the content of suicide reporting in the nineteenth and beginning of the 20th century (Richardson, 2015; Wasserman, Stack, & Reeves, 1994), little is known about the nature of the relationship between suicide reporting and suicide rates.

A recent study utilizing a content analysis of suicide reporting between 1819 and 1944 revealed a covariation between suicide rates and the quantity of reporting (Arendt, 2018). The analysis indicated that the suicide rate mirrored the peaks and troughs in the quantity of suicide reporting over time. Furthermore, the quantity of suicide reporting explained the subsequent year’s suicide rates above and beyond the information contained in the past values of suicide rates alone—a pattern consistent with a long-term Werther effect.

This study contributes to this line of research in two important ways. First, we provide an in-depth analysis of the time period related to strong increases in suicide rates. This appeared in the second half of the 19th century in many countries (Durkheim, 1897/2014), including the Austro-Hungarian Empire (Sonneck et al., 2012). The study cited above investigated a much longer time period (i.e., 1819–1944), including both World Wars, which elicited substantial peaks and troughs in both time series. Thus, it is still unclear whether there is a reliable covariation between the quantity of reporting and suicide rates in the second half of the 19th century and the beginning of the 20th century before World War I. Second, the previous work focused on the covariation within one geographic region (i.e., the geographic region of the present state of Austria). As a supplement to this evidence, we tested whether the covariation between the quantity of reporting and suicide rates held true within five different territories of the Austro-Hungarian Empire. We assessed the annual number of suicides within these five territories and the quantity of reporting on suicide in five newspapers between 1871 and 1910, a period for which there were reliable suicide data available for all five regions. Importantly, each newspaper’s place of publication corresponded to one of the five territories. Durkheim (1897/2014, p. 145) had already theorized that if press-induced imitative effects did contribute to macro-level suicide rates, then the size of this effect must have been influenced by the press’ penetration of readers’ daily lives. The territory-newspaper-matching approach allowed us to investigate Durkheim’s idea.

As a supplement to its contribution to media and suicide history research (and thus to our broader theoretical understanding of the phenomenon of suicide), this study also has important, practical present-day implications. As we will outline later, our findings support suicide experts’ recommendations for the media on how to report on suicide responsibly, especially regarding the notion of avoiding the prominent placement and undue repetition of suicide stories.

Method

This study utilized a content analysis of reporting on suicide. We investigated the quantity of suicide reporting in five different newspapers published in five territories of the Austro-Hungarian Empire. We used official statistics (annual number of suicides) and data from our content analysis to assess whether there was a covariation between the quantity of reporting and the number of suicides within each of the five territories.

Number of Suicides

Annual statistics were taken from Kuttelwascher (1912), who presents the annual number of suicides for the territories of the Austro-Hungarian Empire between 1871 and 1910.

Content Analysis

We used the archive “AustriaN Newspapers Online (ANNO),” run by the Austrian National Library (ANNO, 2017). The ANNO archive allows access to scans of historic newspapers. We content-analyzed the news coverage of five newspapers (Wiener Zeitung, Grazer Volksblatt, Salzburger Volksblatt, Innsbrucker Nachrichten, and Linzer Tages-Post). Each of these five newspapers appeared in the largest city within five different territories of the Austro-Hungarian Empire (Lower Austria including the capital Vienna, Styria including the city of Graz, Salzburg including the city of Salzburg, Tyrol including the city of Innsbruck, and Upper Austria including the city of Linz). We selected these five territories and newspapers because suicide statistics were available for these five territories for the whole observation period (i.e., 1871–1910), and ANNO allowed for an automated content analysis during the whole observation period for these five newspapers (i.e., there was a newspaper appearing throughout the whole observation period and each newspaper was digitally available in the archive). Each newspaper used five newspapers used the name of the largest city of the specific territory in its brand name and was an important newspaper in each of the territories (see Melischek & Seethalter, 2016, for a summary article on the press in the Austro-Hungarian Empire during that time).

We used an automated software-based identification procedure to identify newspaper issues in which suicide had been covered. ANNO allows for keyword searches. We used the search term Selbstmord (suicide) and investigated on how many days this term appeared in each newspaper per year.

Statistical Analysis

We use descriptive statistics to assess the overall trend. Pearson correlation coefficients were used to assess the strength of association between the number of suicides and the quantity of reporting. To test whether there was evidence for a specific causal order, we tested for Granger causality (Granger, 1969). We relied on a hierarchical regression analysis: The quantity of reporting in a given newspaper was assumed to “Granger-cause” suicide rates in a given territory if the quantity of reporting (t – 1; the predictor variable added in Step 2 of the hierarchical model) provided statistically significant information about the future values of the number of suicides (t; outcome variable) when simultaneously controlling for the lagged values of the number of suicides (t – 1; the predictor variable added in Step 1 of the hierarchical model). If the analysis indicated that the quantity of suicide reporting explained the subsequent year’s number of suicides above and beyond the information contained in the past values of the number of suicides alone, this finding would increase the confidence of the causal interpretation consistent with a long-term Werther effect. The change in R² of Step 2 of the hierarchical model tested whether the quantity of reporting Granger-caused the number of suicides.

We also tested for the reverse causal order, that is, whether the number of suicides (t − 1) provided statistically significant information about the future values of the quantity of reporting (t) when simultaneously controlling for the lagged values of the quantity of reporting (t − 1). This finding would increase confidence regarding the causal interpretation that suicides that have occurred, in reality, passed journalists’ gates and found their place in the daily news coverage (see Shoemaker & Vos, 2009, for a review of gatekeeping research). This result would be consistent with the mirror metaphor of the press.

Results

We predicted that there would be a covariation between the number of suicides and the quantity of reporting within all five territories. In a first step, we looked at the descriptive data on the number of suicides and the quantity of reporting over time. As can be found in Figure 1(a), Lower Austria had the highest number of suicides during the whole observation period. This is not surprising because it is a large territory including the capital Vienna. Interestingly, there was a strong increase in Lower Austria’s suicide rate during the early 1870s. Afterwards, the number of suicides continuously increased until 1910. The number of suicides also increased in the other territories, but without a similar abrupt increase in the early 1870s.

Figure 1.

Number of suicides (a) and the quantity of reporting (b) in five different territories between 1871 and 1910 (absolute numbers). Corresponding newspapers: Lower Austria (Wiener Zeitung), Styria (Grazer Volksblatt), Salzburg (Salzburger Volksblatt), Tyrol (Innsbrucker Nachrichten), and Upper Austria (Linzer Tages-Post). The quantity of reporting indicates on how many days the term Selbstmord (suicide) appeared in a given newspaper per year.

Figure 1(b) presents the results of the content analysis. The figure visualizes how many days the term Selbstmord (suicide) appeared in each newspaper per year. Interestingly, Wiener Zeitung (Lower Austria) showed an increase during the late 1870s, a few years following Lower Austria’s abrupt increase in the number of suicides. The press in the other territories also increasingly reported on suicide, albeit that this increase occurred continuously over a longer period. In general, at the end of the observation period, the press reported on suicide, as a rough estimate, on every second day throughout the year. This is a high number, indicating a highly mediated attention and public salience.

To assess whether there was a covariation between the number of suicides and the quantity of reporting, Figure 2 presents the time series for both z-standardized variables for each of the five territories (i.e., official statistics for the specific territory, the quantity of reporting measured for the corresponding newspaper). As can be seen in Figure 2(a) to (e), there were similar trends in all five territories, indicated by significant correlations between the number of suicides and the quantity of reporting: Lower Austria, r(38) = .73, p < .001; Styria, r(38) = .87, p < .001; Salzburg, r(38) = .78, p < .001; Tyrol, r(38) = .54, p < .001; and Upper Austria, r(38) = .56, p < .001. Of interest, in the Salzburg territory, both the number of suicides as well as the quantity of reporting increased in the last years of the observation period, thus increasing confidence in the unique covariation patterns within the territory.

Figure 2.

Covariation between the number of suicides in a given territory (official statistics) and the quantity of reporting on suicide in the corresponding newspaper (z-standardized variables). This figure shows the covariation patterns separately for all five territories (a to e).

We found supporting evidence for the assumption that the quantity of reporting Granger-caused the number of suicides in Styria, ΔF(1, 36) = 21.28, p < .001; Salzburg, ΔF(1, 36) = 18.54, p < .001; and Tyrol, ΔF(1, 36) = 4.53, p = .040. This implicates that when predicting the number of suicides (t) simultaneously with the number of suicides (t − 1) and the quantity of reporting (t − 1), the quantity of reporting (t − 1) added statistically significant explanatory power. Although the coefficients in Lower Austria, ΔF(1, 36) = 0.64, p = .428, and especially in Upper Austria, ΔF(1, 36) = 1.77, p = .192, pointed in the right direction (i.e., consistent with the pattern of a long-term Werther effect), they failed to achieve statistical significance.

We also found evidence for a reversed causal order in Lower Austria, ΔF(1, 36) = 4.79, p = .035; Styria, ΔF(1, 36) = 11.37, p = .002; and Salzburg, ΔF(1, 36) = 6.86, p = .013, consistent with the assumption that the press reported more frequently on suicides (mediated reality) because of the higher number of suicides in reality, supporting the mirror metaphor. No evidence for the mirror metaphor could be obtained for Tyrol, ΔF(1, 36) = 0.78, p = .384, or for Upper Austria, ΔF(1, 36) = 0.98, p = .329.

Additional Analysis

There was no evidence that suicide reporting in Wiener Zeitung Granger-caused Lower Austria’s number of suicides. This is an important finding, given the importance of this territory for the 19th century suicide phenomenon in Austria (see Sonneck et al., 2012). Therefore, we decided to collect more data for this specific territory to increase the generalizability of our findings. It should be noted that we now present the results of a post hoc analysis conducted after the planned analysis reported above had been performed.

We content-analyzed the suicide reporting of two additional newspapers appearing in the Lower Austrian territory: Neue Freie Presse and Das Vaterland. We applied the same procedure as for the other newspapers (see method section). Both newspapers appeared throughout the whole observation period. Consistent with the findings from Wiener Zeitung, there was no evidence that the quantity of reporting Granger-caused the number of suicides, either for Neue Freie Presse, ΔF(1, 36) = 0.22, p = .641, or for Das Vaterland, ΔF(1, 36) = 0.01, p = .928.

Discussion

Suicide rates increased in many countries during the 19th century (Ortmayr, 1990; Thomas & Gunnell, 2010). One previous study provided preliminary evidence that there was a covariation between the quantity of reporting and suicide rates (Arendt, 2018). This study adds to this line of research by providing additional, supporting empirical evidence. An analysis of the time period of strong increases in suicide rates revealed a covariation between the quantity of reporting and the number of suicides. This finding held true within five different geographic regions.

The inspection of the time series and a test for Granger causality are consistent with the assumption that the quantity of suicide reporting contributed to the year-on-year variations in the number of suicides. However, Granger causality could only be revealed for the territories of Salzburg, Tyrol, and Styria. Importantly, there was no evidence that the quantity of reporting Granger-caused the number of suicides in Lower Austria (including the capital Vienna) and Upper Austria. The absence of Ganger causality in Lower Austria is especially striking due to the importance of this territory for Austrian suicide statistics (see Sonneck et al., 2012).

It appears that the press Granger-caused the number of suicides in the (rural) provinces, but not in Lower Austria, which is dominated by the (urban) capital Vienna. Although speculative, we now offer post hoc theorizing on this finding, which may stimulate future research: Suicide research identified the level of industrialization as an important phenomenon for suicide rates (Stack, 1979, 2000). According to Durkheim (1897/2014), the level of industrialization erodes the ties of the individual to society (see also Stack, 2000). Importantly, as industrialization processes increase, economic opportunities tend to be greater in cities. This leads to processes of urbanization. Consequently, industrialization may also elicit negative consequences on social ties to kin, friends, and the church. People must leave their familiar environments. It has also been noted that industrialization is related to individuals questioning their religion, thus leading to secularization (Stack, 1979, 2000).

Assuming that the weakening of the bond between the individual and society was especially strong in the capital Vienna, we offer the following tentative idea: In the Vienna-dominated Lower Austrian territory, processes of modernization (i.e., industrialization, urbanization, and secularization) contributed to the establishment of suicide as a mass phenomenon—without the press being a contributing causal factor within the observation period of this study. A high quantity of reporting might not have been able to increase the public’s salience of suicide any further. Conversely, in the provinces, which were more strongly dominated by rural areas, suicide was less of a publicly salient topic and modernization processes set in much later. In the rural-province context, the press may have made suicide publicly more salient, adding suicide to an individual’s relevant set of possible (detrimental) problem-solving strategies, ultimately leading to imitative effects.

This may have happened via two mechanisms: (a) via the priming of the concept of suicide, that is, increasing the chronic accessibility of suicide through frequent exposure in the public’s mind (see Bargh, 2006, for a review of priming). If a concept such as suicide is chronically accessible, the likelihood of it coming into a person’s mind and thoughts when faced with a situation regarding a solution to a negative life event is higher; and (b) via influencing perceived social norms: High media attention on suicide may have increased perceived descriptive norms (“Many people die by suicide”) or even perceived injunctive norms (“Many people think suicide is acceptable”). Both norm concepts have been shown to be causally related to behavior (see Fishbein & Ajzen, 2010).

We want to emphasize again that this post hoc theorizing is tentative. Future studies need to provide more evidence for the suicide-rate interaction with processes of modernization. We understand this post hoc theorizing as a starting point for future studies, and not as definitive knowledge.

Implications for Media Guidelines

As a supplement to this study’s contribution to media and suicide history research (and thus to our broader theoretical understanding of the phenomenon of suicide), the evidence also has present-day implications. Findings support contemporary media recommendations for responsible reporting on suicide (Jamieson, Hall-Jamieson, & Romer, 2003; Tomandl, Sonneck, & Stein, 2008; World Health Organisation [WHO], 2017), especially regarding the notion of avoiding the prominent placement and undue repetition of suicide stories. International and national guidelines typically target a variety of content elements (see Scherr, Arendt, & Schäfer, 2017) such as avoiding undue attention being placed on suicides; avoiding romanticizing, glamourizing, or heroicizing suicidal behaviors; avoiding detailed descriptions of the methods employed and detailed descriptions of the location; taking special care when reporting about celebrity suicides; and providing alternatives to suicide (e.g., contacting crisis-intervention centers). Among this myriad of content elements, the quantity of reporting as measured in this study is most strongly relevant to the first element: undue attention. In fact, this study found that the newspapers tended to report on suicide every second day at the end of our observation period. This is a high level of media attention on suicide, possibly eliciting a high degree of public attention.

It is important to note that it is not the goal of suicide-prevention efforts to displace the topic of suicide completely from media and public attention. Of course, suicide should not be a taboo. A tabooziation may hinder suicide-prevention efforts (e.g., talking about suicidal ideation can have beneficial consequences in suicidal individuals; see Sonneck et al., 2012). Conversely, the media might act as an agent to break this taboo. Consistent with this notion, research on the suicide-protective Papageno effect shows that positive content elements (e.g., stories about individuals successfully coping with a suicidal crisis) can contribute to a reduction in the suicide rate (Niederkrotenthaler et al., 2010).

Taken together, journalists should make Papageno-related content elements more salient in their reporting. Therefore, when reporting on suicide, survival/success stories, references to counseling services by telephone or chat, epidemiological facts, and research evidence debunking widely shared suicide myths are important content elements of responsible reporting. However, these positive content elements should be separated from undue attention: Journalists should avoid prominent placement and undue repetition of suicide stories due to possible negative imitative effects. This study adds to this line of research by providing empirical evidence for a 40-year period. This is an important supplement to the existing suicide literature, which largely tested “short-term” Werther effects (i.e., effects occurring within a few weeks after publication) in present-day media environments.

Limitations

The study has some limitations. First, suicide-rate statistics are not an exact mirror of reality. Although the official statistics of the Austro-Hungarian Empire were well developed, there might be factors influencing statistics in an undesirable way (e.g., not reporting a death as a suicide because of religious or legal reasons, different validity criteria for suicide statistics in different territories). Second, the specific newspapers were selected because they were published within the five selected territories. Although the readership was concentrated in this region, it is possible that some readers living outside the specific territory also read the newspapers. We have no market research data for the 19th century press. However, due to barriers such as those related to transportation, we assume that these papers were largely read by readers living in the specific regions. Third, the test for Granger causality should be interpreted with caution due to the low number of observation years. Unfortunately, we were limited to the 40-year time period because this was the time period where suicide statistics were available for the five territories (Kuttelwascher, 1912). Fourth, there were other unknown factors influencing the quantity of reporting that we did not measure. For example, Salzburger Volksblatt’s quantity of reporting was substantially reduced in 1904. This decrease can hardly be explained by the number of suicides. It is possible that specific events such as the death by suicide of Rudolf, the Crown Prince of Austria (son of Emperor Franz Joseph I)—monarchy-loyal Austrian newspapers did not report on this event as a suicide but only as a tragic death—influenced the reporting on suicide differently for different newspapers (e.g., due to the editor’s opinions). This is left for a future study. Fifth, the analysis was restricted to territories of the Austro-Hungarian Empire. It is up to future studies to prove whether the findings of this study can be generalized to other contexts.

Conclusion

Despite these limitations, this study provides additional supporting evidence for a covariation between the quantity of reporting on suicide and macro-level suicide data. Although this study is not able to prove a causal effect—partly due to the impossibility of experimentation back in the 19th century—the evidence is consistent with the idea that the press may have contributed to the establishment of suicide as a mass phenomenon. However, we want to emphasize that a lack of confidence exists regarding the causal interpretation. Additional empirical data are required to be better able to judge the nature of the relationship back in the 19th century.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Author Biography

Florian Arendt has a position as an “Akademischer Rat” (post-doc level) at the chair of empirical communication science at the University of Munich (LMU). His primary research interests focus on health communication.

References

AustriaN Newspapers Online. (2017). Austrian Newspapers Online. http://anno.onb.ac.at/index.htm.

Arendt, F. (2018). Reporting on suicide between 1819 and 1944: Suicide rates, the press, and possible long-term Werther effects in Austria. Crisis: The Journal of Crisis Intervention and Suicide Prevention, 39(5), 344–352.

Bargh

(2006) What have we been priming all these years? On the development, mechanisms, and ecology of nonconscious social behavior. European Journal of Social Psychology 36: 147–168.

Durkheim

(1897/2014) Der Selbstmord [Suicide], Berlin, Germany: Suhrkamp.

Fishbein

Ajzen

(2010) Predicting and changing behavior: The reasoned action approach, New York, NY: Taylor and Francis.

Granger

(1969) Investigating causal relations by econometric models and cross-spectral methods. Econometrica 37: 424–438.

Jamieson

Hall-Jamieson

Romer

(2003) The responsible reporting of suicide in print journalism. American Behavioral Scientist 46: 1643–1660.

Kuttelwascher

(1912) Selbstmord und Selbstmordstatistik in Österreich [Suicide and suicide statistics in Austria]. Statistische Monatsschrift 38: 267–351.

Melischek

Seethalter

(2016) Die Tagespresse der franzisko-josephinischen Ära [The press during the reign of Franz Joseph I]. In: Karmasin

Oggolder

(eds) Österreichische Mediengeschichte: Band 1, Wiesbaden, Germany: Springer, pp. 167–192.

10.

Niederkrotenthaler

Voracek

Herberth

Till

Strauss

Etzersdorfer

Sonneck

(2010) Role of media reports in completed and prevented suicide: Werther v. Papageno effects. The British Journal of Psychiatry 197: 234–243.

11.

Ortmayr

(1990) Selbstmord in Österreich 1819–1988 [Suicide in Austria 1819–1988]. Zeitgeschichte 17: 209–225.

12.

Phillips

D. P.

(1974) The influence of suggestion on suicide: Substantive and theoretical implications of the Werther effect. American Sociological Review 39: 340–354.

13.

Richardson

(2015) A history of suicide reporting in Canadian newspapers, 1844–1990. Canadian Journal of Communication 40: 425–445.

14.

Scherr

Arendt

Schäfer

(2017) Supporting reporting: On the positive effects of text- and video-based awareness material on responsible journalistic suicide news writing. Archives of Suicide Research 21: 646–658.

15.

Shoemaker

Vos

(2009) Gatekeeping theory, New York, NY: Routledge.

16.

Sonneck, G., Kapusta, N., Tomandl, G., & Voracek, M. (Eds.). (2012). Krisenintervention und Suizidverhütung [Crisis intervention and suicide prevention]. Vienna, Austria: Facultas.

17.

Stack

(1979) Durkheim’s theory of fatalistic suicide: A cross-national approach. The Journal of Social Psychology 107: 161–168.

18.

Stack

(2000) Suicide: A 15-year review of the sociological literature part II: Modernization and social integration perspectives. Suicide & Life-Threatening Behavior 30: 163–176.

19.

Stack

(2005) Suicide in the media: A quantitative review of studies based on nonfictional stories. Suicide & Life-Threatening Behavior 35: 121–133.

20.

Thomas

Gunnell

(2010) Suicide in England and Wales 1861–2007: A time-trends analysis. International Journal of Epidemiology 39: 1464–1475.

21.

Tomandl, G., Sonneck, G., & Stein, C. (2008). Leitfaden zur Berichterstattung über Suizid [Guidelines for news coverage about suicides]. Vienna, Austria: Kriseninterventionszentrum. http://www.suizidpraevention.at/downloads.html.

22.

Wasserman

Stack

Reeves

(1994) Suicide and the media: The New York Times’s presentation of front-page suicide stories between 1910 and 1920. Journal of Communication 44: 64–83.

23.

World Health Organisation. (2017). Preventing suicide: A resource for media professionals. http://www.who.int/mental_health/suicide-prevention/resource_booklet_2017/en/.