2006 Glasgow University Study - Goth stops self harm

This study from Glasgow University in the British Medical Journal is regularly quoted by the media to demonstrate that goth makes peoples self harm.

Robert Young, Helen Sweeting, and Patrick West, Prevalence of deliberate self harm and attempted suicide within contemporary Goth youth subculture: longitudinal cohort study BMJ 2006 332: 1058-1061. [Download here]

In fact the article proves nothing of the kind as noted in an article on the report from a scientific source New Scientist:

Goth subculture may protect vulnerable children

• 00:01 14 April 2006
• NewScientist.com news service
• Gaia Vince

About half of teenage goths have deliberately harmed themselves or attempted suicide, a new study suggests. But joining the modern subculture – which grew out of the 1980s gothic rock scene – may actually protect vulnerable children, researchers say.

The study followed 1258 young people who were interviewed at ages 11, 13, 15 and 19. It found that of those who considered themselves goths, 53% had self-harmed and 47% had tried to commit suicide. The average prevalence of self-harm among young people in the UK is 7% to 14%. Self-harm includes behaviours such as cutting or burning oneself. And about 6% of young people admit suicide attempts. Some studies suggest the incidence is rising in society.

Researchers at University of Glasgow found that while most self-harmers started the practice at age 12 to 13, they did not become goths until they were a couple of years older, on average.

“One common suggestion is they may be copying subcultural icons or peers [when they self-harm], but our study found that more young people reported self-harm before, rather than after, becoming a goth. This suggests that young people with a tendency to self-harm are attracted to the goth subculture,” says Robert Young, who led the study.

Quick fix

“Rather than posing a risk, it's also possible that by belonging to the goth subculture, young people are gaining valuable social and emotional support from their peers.” But he cautions: “However, the study was based on small numbers and replication is needed to confirm our results.” Only 25 participants felt strongly associated with goth culture.

Self-harming, Young says, is a behaviour that people often employ as a mechanism to deal with negative emotions. “It may be used as a quick-fix. "Some physiological studies suggest, or are compatible with the theory that endorphins [brain chemicals that produce a feeling of well-being] are released after episodes of self-harm," he told New Scientist.

Just 2% of the adolescents in the study identified with goth culture, although 8% said they had identified with it at some point in their lives. But it is a strongly non-violent and accepting subculture, which teens may find offers a supportive environment.

Michael van Beinum, a psychiatrist for children and adolescents, who advised on the study, agrees: “For some young people with mental health problems, a goth subculture may be attractive as it may allow them to find a community within which it may be easier for their distress to be understood.”

The 1980s goth culture grew out of the post-Punk movement and underwent a revival in the mid-1990s. Central to goth belief is the black aesthetic – taking icons that society regards as evil, such as skull imagery, and making them beautiful.

Journal reference: British Medical Journal (vol 332, p 909)

Even beyond this there are problems with the research itself. The results may actually mean that goths are more ready to admit they have a problem with depression and self harm because the subculture is more tolerant.

It was carried out with one age group and in one area in Scotland. Given the frequent verbal buse and bullying goths suffer does this explain some of the results? More importantly as one comment pointed out: "What about the 90% of teenagers who self harm who aren't goths (73 of 81 people in this study)?"

This excellent comment on the BMJ website one of many sums up more problems:

Michelle Phillipov,
Postgraduate Student
Dept of English, University of Adelaide 5005

It is somewhat unexpected that BMJ would publish such an article, which attempts to determine traits associated with the Goth subculture—a grouping which potentially comprises of 100,000s adherents globally—based on a sample of only 15 individuals in a single locality.

While Goth subculture is the article focus, just two references are given, and only one of them academic. There is a failure both to consult other pertinent studies—including Wright [1] and Siegel [2]—and to frame the results with respect to existing knowledge. For example, while Young et al’s results show a significant prevalence of males (about 2:1) in the ‘Goth’ category, Hodkinson—the only academic reference on Goth cited by the authors (see Young et al, ref [5])—clearly states that the Goth subculture is comprised of equal numbers of males and females. The reluctance to address this inconsistency with respect to their own cited reference is not only a significant oversight, it also makes it impossible to determine whether the authors’ findings are the result of male over- or female under-identification as Goth, or if they are simply symptomatic of untenable sample size and study design.

In addition, while Young et al find females at risk of attempted suicide, the known suicide rate for the 15-19yr group in 2000 was 4 to 1 with respect to males (see Young et al, ref [1]). Hence, the results directly contradict what is logically expected.

Furthermore, it is notable that members of the Goth subculture practice self-harm almost exclusively by means of cutting, scratching and scoring, and not through more extreme methods such as punching or self- poisoning. However, the authors unwillingness to consider the context(s) in which cutting, scratching and scoring occur, along with their readiness to subsume potentially diverse practices into a single discourse of pathology, makes it impossible to draw conclusions about the meaning and effects of these behaviours for the Goth subculture.

While the authors seek to link cutting, scratching and scoring to depression, attempted suicide and psychiatric illness, these techniques can in fact be practised for a range of different reasons. For example, within some subcultural contexts, these practices are used to induce fine scarring in decorative patterns on the body; that is, they are used as methods of bodily decoration and adornment, much like tattooing or body piercing. Without an understanding of context, it is impossible to determine whether this is the case for the Goth subculture. However, the tendency for medical and mental health discourse to systematically misrecognise body modification as self-mutilation or self-harm has been noted elsewhere [3].

Unfortunately, studies like the current one, which adopts an epidemiological approach to assessing traits within subgroups, habitually seem to support prevailing popular stereotypes. Perhaps this explains the ready acceptance and promulgation of the findings, whilst disregarding the lack of substantiation and generalisability, as well as the serious methodological flaws.

[1] Wright R. I’d sell you suicide: pop music and moral panic in the age of Marilyn Manson. Popular Music 2000; 19: 365-86.

[2] Siegel C. Goth’s dark empire. Bloomington: Indiana University Press, 2005.

[3] Pitts V. Body modification, self-mutilation and agency in media accounts of a subculture. In Body Modification (Ed. Featherstone M). London: Sage, 2000, pp291-303.

Someone else pointed out statistical flaws:

I have enjoyed reading the paper and all responses. Here, there are few "statistical" commments. There seemed to be several flaws in the design and execution of this study.

1. are the results valid?

Authors don't give us clear description of how subjects were identified and recruited. Did they volunteer? How many refused to participate in the study? The researchers could simply have sampled the participants in such a way as to increase the chance of supporting their hypothesis. Cohort should be representative of a true population for study to be valid. Some researches use randomization procedure in cohort studies in order to choose sample, which is the most accurate picture of population.

2. Follow-up/attrition rate.

Follow-up in this study was probably long enough (8years). However, in 1994, there were 2586 participants, whereas by the year 2002/04 only 1258. Could those followed-up be a biased sample? Loss of participants may effect precision and power of the study. For example, in Table 3 under Model 1 Goth subculture there is wide confidence interval (4,42 to 45,39), which may point out to undepowered study (type II error).

3. Observer Bias?

Outcomes have been assesed using Voice-DISC. Participants were asked varies questions on: suicide, self-harm, identification with subculture. It is not mentioned how raters conducted the interviews, whether they used structured or semi-structured protocol, whether they were blind to the study's hypothesis. The more subjective rating is, the more important blinding becomes. The bias introduced by non-blinding is likely to overestimate findings. It seems that a decision of belonging to one subculture or another was rather subjective (participants' responses were assessed on a five point scale). All in all, no blinding, lack of structured interview, subjectivity can lead to observer bias.

4. Confounders

Reseachers did adjust for some confounders (sex, social class, etc) using logistic regression. However, as somebody has already pointed out, the list of potential confounding factors is long, inc physical illness, obesity, family hx.

5. Results

For prospective studies outcomes are best presented as a relative risk.One can comment then on prevalence or incidence of the disorder. In this study authors use odds ratio.

6. Study's applicability

Is this study applicable in different subcultures? Probably not. In Table 3, which shows relationship between self-harm and type of subculture, majotiry of results are statistically non-significant.

In conclusion,

In cohort studies a group of individuals is followed-up over a period of time. The individuals should be free of the outcome (here: self- harm/attempted suicide)at the beginning of follow-up. The cohort is defined by exposure status(here: Goth subculture), which should be found out before outcome is known. Cohort study measures whether exposure effects the incidence/prevalence of the outcome,i.e. whether identification with Goth subculture increases the risk of self-harm or/and suicide. Researches believe that yes. By reading and critically analysing this study, I have arrived at different conclusions.

The authors reply is interesting:

We were pleased that our paper generated a wide variety of comments. While some contributors raised specific points, others referred to more general issues relating to issues of definition, explanations and generalisability.

The definition of “Goth” is contentious, but covers a wide range of musical tastes, social groupings, and aesthetics. The most relevant distinction here is between contemporary (usually younger) “Baby, Bat or sometimes referred to as Mall (US)” Goths vs. “mature, real or Elder Goths” 1. Our paper, as is clear from the title, refers to younger Goths; the results may not apply to all Goths.

The range of comments in response to our conclusion, that both selection and influence mechanisms may be involve, reflect the lack of evidence on this issue. To highlight this, we drew attention in the press release following publication to the possibility that engagement with Goth subculture could have positive rather than negative consequences for some young people. Our contribution is a first step towards producing an evidence base to test this, rather than relying on media speculation.

Some contributors have suggested that the association between self- harm and Goth subculture may be accounted for by other factors. However this is unlikely, since we adjusted for the strongest and most relevant correlates of self-harm found in other studies of young people. Others have suggested that our results were not valid due to the small numbers involved. We would point out that our paper underwent a formal statistical review before publication. Further, while the media focused on the 25 young people who unambiguously identified as Goth, nearly 8% of our representative sample had identified with Goth subculture, in varying degrees of intensity, and were 3-4 times more likely to self-harm, than the other participants.

It has also been suggested that by adopting a quantitive approach we may have missed contextual factors (this is obviously true of any non- qualitative study), and that the high rate of self-harm found among Goths is a form of decoration, analogous to body modification. We dispute this on two grounds. Firstly, since those who self-harmed were asked why, we know that the majority, regardless of youth subculture did so to relieve anxiety, anger and other negative emotions. Secondly, while cutting could be interpreted as some form of subcultural display, such an argument is difficult to sustain in relation to attempted suicide.

Sadly as I will show in a later post the media ignored the authors intentions with the notable exception of the New Scientist.