Music and Suicidality: A Quantitative Review and Extension

654 Suicide and Life-Threatening Behavior 42(6) December 2012 Ó 2012 The American Association of Suicidology DOI: 10.1111/j.1943-278X.2012.00120.x Music and Suicidality: A Quantitative Review and Extension STEVEN STACK, PHD, DAVID L ESTER, PHD, AND JONATHAN S. ROSENBERG , BA This article provides the first quantitative review of the literature on music and suicidality. Multivariate logistic regression techniques are applied to 90 findings from 21 studies. Investigations employing ecological data on suicide completions are 19.2 times more apt than other studies to report a link between music and suicide. More recent and studies with large samples are also more apt than their counterparts to report significant results. Further, none of the studies based on experimental research designs found a link between music and suicide ideation, prompting us to do a brief content analysis of 24 suicide songs versus 24 nonsuicide songs from the same album. Using Linguistic Inquiry and Word Count software, we found no difference in the content of the suicide songs and controls, including the percentage of sad words, negative affect, and mentions of death, thus providing an explanation for nonfindings from experimental research. In summary, ecologically based (which capture at-risk persons not in typical school-based samples) and more recent investigations (which have used superior or new methodologies) tend to demonstrate a linkage between music and suicidality. Experimental research is needed with a control group of songs from an alternative genre with low suicidogenic content. With the advent of new developments including the Great Recession and aging of the baby boomer cohort, the suicide rate among adults is rising, recently reaching 15.2/100,000. In addition to 35,000 suicides a year, there are well over 300,000 emergency department admissions for suicide attempts annually (Crosby, Han, Orteag, Parks, & Gfroerer, 2011). While there are many risk and protective factors for suicide, the influence of the media has received considerable attention (Pirkis & Blood, 2001; Stack, 2005, 2009). However, nearly all of the over 130 STEVEN STACK, Wayne State University, Detroit, MI, USA; DAVID LESTER and JONATHAN S. ROSENBERG, The Richard Stockton College of New Jersey, Galloway, NJ, USA. Address correspondence to Steven Stack, Department of Psychiatry & Criminology, 3258 Faculty Administration Building, Wayne State University, Detroit, MI 48202; E-mail: steven_ stack@hotmail.com known investigations deal with the news media and, to a lesser extent, depictions of suicide in the cinema and in television series (Pirkis & Blood, 2001; Stack, 2005, 2009). The possible impact of suicidality as expressed in the medium of song or in musical subcultures has received scant attention. Musical genres and the subcultures associated with them have often been linked by both scholars and journalists to suicide risk. For example, scholarly work on American terrorists has underscored the importance of white power rock; journalists and concerned citizens have attributed individual acts of suicide to songs such as Ozzie Osbourne’s Suicide Solution; and even mass homicide followed by suicide, such as the case of Columbine High School, has been connected to musical influences such as industrial and heavy metal (Hamm, 2004; Litman & Farberow, 1994; Snell & Hodgetts, 2007). Quantitative research has supported a link between a STACK AND LESTER musical genre and suicide (e.g., Pimental, Gouveia, de Santana, Chaves, & Rodrigues, 2009; Stack & Gundlach, 1992; Young, Sweeting, & West, 2006), but other studies report little or no support for such a connection (e.g., Ballard & Coates, 1995; Rustad, Snall, Jobes, Safer, & Peterson, 2003). Some studies report split findings such as a gendered relationship between music and suicidality (Burge, Goldblat, & Lester, 2002; Martin, Clarke, & Pearce, 1993; Scheel & Westfield, 1999). To date, these mixed reports have not been synthesized or fully reviewed. In a general review of the effect of fictional media on suicide, only four studies on music are analyzed (Pirkis & Blood, 2001). The present article fills this gap in the literature. As we shall see, the weakest support for a link between suicidality and music is work based on experimental research on individuals. There, a suicide song is often presumed to have more of an impact on audience suicidality than a nonsuicide song from the same genre. An untested assumption of this work is that the content of suicide and nonsuicide songs from the same genre (as defined by the researchers) is sufficiently different so that the former would be expected to differentially impact the suicidality of the audience. We tested this assumption in an extension to the quantitative literature review by comparing the content of a sample of 24 suicide songs and 24 nonsuicide songs from the same artist/album. First, however, we briefly review the two theoretical perspectives that have guided most of the work on music and suicide: the subcultural and psychological/theoretical perspectives. These are not the only possible perspectives that one could use to explain a link between music and suicide (e.g., see the unapplied perspectives on music and emotion, Gagnon & Peretz, 2003), but they are the ones that have guided existing work. MUSIC-ORIENTED SUBCULTURES Youth in the United States and United Kingdom spend between 2.5 and 4 hours 655 each day listening to music (Baker & Bor, 2008). In general, music is second only to watching movies and television in terms of the amount of time people spend on leisure activities in the United States. On average, people spend 18 hours per week listening to music compared to only 8 hours on reading newspapers, books, and magazines (Rentfrow, Goldberg, & Zilca, 2011). The patterns in the narratives in popular music offer substantial opportunities for social learning: they shape, question, and/or reinforce the public’s values, beliefs, attitudes, and behaviors (Stack, 2002; Young et al., 2006). Two broad theoretical perspectives have been employed in research on the association between music and suicide. One is sociological and argues that the impact of music on suicide is best understood in relation to a subculture where music reflects as well as reinforces the values, attitudes, and behaviors of its members (e.g., Gaines, 1991; Stack, Gundlach, & Reeves, 1994; Weinstein, 1991; Young et al., 2006). The other is psychological and focuses on particular songs about suicide or the individual personalities of the audience. In this latter view, suicide songs are thought to influence listeners independent of their membership in a music subculture, or that individual personality traits attract at-risk persons to the music independent of social influences (e.g., Baker & Bor, 2008; Peterson, Safer, & Jobes, 2008). These two views are not mutually exclusive. For example, members of a particular musicbased subculture may be initially drawn to a music genre based on individual differences. Thereafter, social influences may intensify adherence to the norms of the subculture. Not all music-based subcultures are suicidogenic or equally suicidogenic. The literature has tended to focus on only a few subcultures or genres thought to be among the more high-risk groupings. Favorite selections have included the more dismal thrash and death metal songs/subcultures of the 1980s and subsequent times. These songs illustrate themes of individual (e.g., hopelessness, depression, pessimism, alienation) and societal (e.g., war, pollution) chaos (Wein- 656 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW stein, 1991; Young et al., 2006). They are marked by a recurrent theme of hopelessness that nothing can be done to resolve issues. Hopelessness is a key risk factor for suicide (Stack, 1998; Stack et al., 1994; Weinstein, 1991). In the case of heavy metal, a content analysis of over 4,000 songs determined that personal and societal chaos is a recurrent theme in the lyrics (Weinstein, 1991). The more current Goth and Emo subcultures are new cases in point (Baker & Bor, 2008; Young et al., 2006). Membership in a musical subculture is characterized by such indicators as liking the music, number of hours spent listening to the music, memorizing the lyrics of the tunes, collecting CDs, attending concerts, collecting memorabilia and symbols such as posters of a band, adopting the style of clothes and grooming of the subculture, hanging out in clubs or music stores frequented by like-minded persons, and playing or aspiring to play in or assist a band (Lacourse, Claes, & Villeneuve, 2001; Recours, Aussaguel, & Trujillo, 2009; Snell & Hodgetts, 2007; Weinstein, 1991). Such activities expose the individual to the subculture and reinforce its tenets. defines suicide in different ways including suicide rates, suicide attempts, and suicide ideation. Our mission is to ascertain whether these methodological differences make a difference in what researchers find in the relationship between music and suicide. Standard bibliographic sources were searched (MEDLINE and Sociological Abstracts) under the keywords ‘‘suicide’’ and ‘‘music.’’ Over 1,000 hits were achieved, but most investigations did not meet the criteria for inclusion in the review: (1) the study must measure some standard aspect of suicidality such as suicide completions, suicide attempts, suicide ideation, and suicide acceptability; (2) the study must measure some aspect of music including such variables as radio market share of a musical genre, musical preference (e.g., for heavy metal, punk, blues, opera, and so on), and music worship or involvement (such as owning posters, going to concerts); and (3) an empirical test of the relationship between a measure of suicide and a measure of music must be reported (e.g., a correlation, t test of differences in means, or results from a regression analysis). As investigations that met these criteria were found, a grandfathering search was done for additional studies cited in their bibliographies. A total of 19 papers containing 21 studies or investigations met the criteria for inclusion in the study. They date from 1992 to 2011. The main findings of the central studies reviewed here are summarized in Table 1. A preliminary review of this body of work noted some variation in positive results associated with how suicidality is measured. The results based on studies using suicide completions tend to be the strongest, while results based on experimental research on suicide ideation tend to be the weakest. The table follows this order. SYSTEMATIC LITERATURE REVIEW: MUSIC AND SUICIDALITY It is important to note that the constructs of both music and suicide are defined and measured differently in existing research studies. Suicide research has defined music in a variety of terms. In studies of urban- or state-level suicide rates, for example, music refers to such concepts as radio market share of country music and heavy metal magazine subscription rates (e.g., Stack & Gundlach, 1992). In a larger number of studies, music is measured in terms of musical preferences for a variety of music genres (e.g., Young et al., 2006). In still other studies, researchers select what they believe to be a representative song and measure subjects’ reactions after exposure to the song (e.g., Rustad et al., 2003). In like manner, the research to be reviewed Suicide Completions, Suicide Rates Research based on actual suicide rates, following an ecological model, generally finds significant associations between music and suicide. The body of ecological research Research on Music and Suicidality: Study, Measure of Music, Sample Size, and Key Findings by Suicide Outcome Variable: (A) Rates, (B) Attempts, (C) Ideation, and (D) Acceptability (N = 21 studies) Study type, first author (year) Measure of music (A) Studies on Suicide Completions: Suicide Rates Gundlach (2010) Number of radio plays of Whiskey Lullaby Lester (2011) Proportion low notes in national anthems Stack (1992) Country radio market share Sample size 108 cities 18 nations 49 cities Stack (1994) Heavy metal magazine subscriptions 50 states Maguire (1994) Country radio market share 49 cities 49 cities 93 college students 1,255 youth 1,255 youth 1,255 youth 1,255 youth 175 college students For every 124.5 plays of the song, there was one additional male suicide The greater the proportion of low notes, the higher the nation’s suicide rate The greater the country market share, the greater the White suicide rate The greater the magazine subscriptions, the greater the youth suicide rate (but not middle age rate) Country radio market share unrelated to suicide rates (errors in suicide data) Country market share affects None of five preference factors predicted attempts At age 19, gothic ID increases odds of previous suicide attempt by 16.37 times Predicts self-harm Predicts self-harm Predicts self-harm 104 college students Exposure to neither a rap suicide song nor a heavy metal suicide song was associated with elevated suicide ideation relative to nonsuicide songs No group differences in perceived self-risk for suicide 133 college students No group differences in perceived self-risk for suicide 657 Stack (1995) Country radio market share (B) Studies on Suicide Attempts Lester (1996) Factor scores, music preference Young (2006) Heavy identification (ID) with gothic subculture ID with punk ID heavy metal ID with mosher (C) Studies on Suicide Ideation (C1) Experimental Designs Ballard (1995) Exposure to a suicide song versus nonsuicide song of same genre Rustad (2003) Exposure to suicide versus nonsuicide heavy metal music videos by same artist (Pearl Jam) Rustad (2003) Exposure to heavy metal lyrics of three suicide songs versus lyrics three nonsuicide songs, same artists Key finding STACK AND LESTER TABLE 1 658 TABLE 1 (continued) Study type, first author (year) Sample size 77 high school (HS) students 77 HS students 77 HS students Burge (2002) Heavy metal/classic rock/ alternative/punk factor Country/rap (negative)/pop factor Heavy metal/classic rock/ alternative/punk factor Country/rap (negative)/pop Factor Lacourse (2001) Heavy metal preference 275 HS students Canada Lester (1996) Factor scores for groups of musical preferences 93 college students Martin (1993) Preference for rock/metal 227 HS students Australia Pimental (2009) Preference for conventional versus alternative music 200 college students Brazil Scheel (1999) Preference for heavy metal 121 HS students (C2) Nonexperimental Designs Burge (2001) Burge (2001) Burge (2002) 77 HS students Key finding Music factor predicts suicide ideation Music factor predicts suicide ideation Factor not predictive of suicide ideation for girls or boys Music factor predicts suicide ideation for girls, not boys At the bivariate level, metal predicted suicide ideation/risk for girls only At the bivariate level, metal worship predicted suicide ideation for girls only At the multivariate level, metal indicators failed to predict suicide ideation/risk Those with musical preference for heavy metal were more apt to have considered suicide in the past A musical preference for country/western/ show tunes was unrelated to suicide ideation For girls, 66% rock/metal fans versus 35% other fans thought about suicide in past For boys, no relationship was found Preference for conventional music versus alternative music lowers suicide risk items on the Reasons for Living Scale 74% female metal fans versus 35% nonfans have considered suicide. For boys, 42% versus 15%, not significant MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW Measure of music (continued) Study type, first author (year) Measure of music Sample size (D) Studies on Suicide Acceptability (D1) Nonexperimental Designs Burge (2001) Heavy metal/classic rock/ alternative/punk factor Scheel (1999) Preference for heavy metal (strongly liked or liked) 77 HS students Music factor increases suicide acceptability 121 HS students Heavy metal preference lowered moral objections to suicide Male fans in particular had fewer moral objections than nonmale fans Heavy metal fanship lowered acceptability at the bivariate level Heavy metal fanship lowered religiosity which, in turn, affected suicide acceptability Blues fanship affects White suicide acceptability indirectly through lowering religiosity (but not for Blacks) Opera fans were 2.45 times more apt than others to approve of suicide in the context of dishonor Stack (1998) Like heavy metal music 833 adults Stack (1999) Like blues music very much 961 adults Stack (2002) Like opera 845 adults Exposure to suicide versus nonsuicide songs, same artist Exposure to suicide versus nonsuicide music video, same artist 133 college students (D2) Experimental Research Rustad (2003) Rustad (2003) 104 college students Key finding STACK AND LESTER TABLE 1 No group differences were found in the acceptability of suicide No group differences were found in the acceptability of suicide 659 660 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW covers several musical subcultures and measures including the strength of the country and western subculture, the heavy metal subculture, and also the proportion of low notes in national anthems. Gundlach and Stack (2010) focused on the effect of the number of plays of the popular country song, Whiskey Lullaby, on urban suicide rates. The song concerns the suicide of a soldier who, coming home from war, finds that his wife is having an affair with another man. He ‘‘puts the bottle to his head and pulls the trigger.’’ The wife suffers extreme guilt and ultimately suicides as well. She is found dead, clinging to his picture, and is buried beside him under a willow. The song resonated with the country audience and rose to number three on the charts. It received the Country Music Association’s award for Best Song of the Year in 2005. In an analysis of male suicide rates in 108 cities, it was found that for every 124.5 plays of the song, there was one additional male suicide (Stack & Bowman, 2012, p. 271). The message of the song may resonate with suicidal men seeking revenge on unfaithful women. Stack and Gundlach (1992) reported a significant association between country music radio market share and White suicide rates (but not Black rates) in a set of 49 metropolitan areas. These findings were based on 1985 data, before country and western music fully proliferated beyond its historical demographic roots (e.g., working class, rural, South). The results were interpreted in terms of both selection and subcultural effects, not so much the often sad themes in country music per se. For example, country fans were considerably more apt than nonfans to be divorced (27% vs. 18%) and to own guns (62% vs. 40%), two prime risk factors for completed suicide (Stack & Gundlach, 1995). Mediator variables often need to be taken into account to assess the full impact of music on suicidality. For example, on the ecological level, there is also a powerful indirect effect of the country music radio market share on urban divorce rates, which in turn affect White suicide rates (Stack & Gund- lach, 1995). Maguire and Snipes (1994) were not able to replicate the Stack and Gundlach findings, but this was apparently due to large measurement errors they made in their calculation of suicide rates (Stack & Gundlach, 1994). Stack et al. (1994) explored the link between the strength of the heavy metal subculture and suicide rates in the 50 states. The strength of the heavy metal subculture was measured as the rate of subscriptions to a major magazine, Metal Edge. The strength of the subculture predicted youth suicide rates (but not those of the middle aged) in the 50 states and did so independent of five socioeconomic predictors of suicide rates. Finally, Lester and Gunn (2011) explored a hypothesis from Rihmer (1997) that sad themes in national anthems were associated with low notes and high national suicide rates. Indeed, the greater the proportion of low notes in national anthems, the higher the national suicide rates. The proportion of low notes is a relatively unexplored area in suicide and music. It is noteworthy that classic heavy metal songs, a genre thought to nurture suicidal tendencies through a theme of hopelessness, are generally played in minor keys, the keys of melancholy, and almost never in major keys (Arnett, 1991). As a brief aside, there is a body of research on the influence of music on emotions which could be employed in future work on suicide following the findings of Lester and Gunn (2011). Music has long been known to have an impact on mood (Peretz & Sloboda, 2005; Peretz & Zattore, 2005). Research has found that the mode (major–minor) and tempo (fast–slow) affect the judgments of the happiness versus sadness of the music, with slow music and a minor key resulting in higher ratings of sadness (Gagnon and Peretz (2003). The link between music and mood has implications for psychological treatment. Importantly, the use of music in therapy is a well-established practice (e.g., Darnley-Smith & Pately, 2003), with a professional organization, The American Music Therapy Association STACK AND LESTER (http://www.musictherapy.org), and specialized journals (e.g., Journal of Music Therapy). However, the evidence and theory from this body of work on music and emotion has yet to be systematically applied to suicide studies. Suicide Attempts Only two investigations explored the link between music indicators and suicide attempts. Young et al. (2006) found strong identification with several youth subcultures on the risk of suicide attempts as well as serious self-injury. Their focus was on the Goth subculture, representing a subgenre of punk. In a longitudinal study of 1,258 youth beginning at age 11, those who identified with the Goth subculture (heavy identification with or ‘‘I am one’’) were fully 16.37 times more apt to report at least one previous suicide attempt at age 19 than non-Goths. This association was independent of controls including depression level, substance abuse indicators, family dysfunctions, social class, and gender. Those reporting ‘‘quite a bit’’ of identification with the subculture were just 3.54 times more apt to have attempted suicide. The degree of identification with being a Goth was rather an important consideration. High identification with the Goth subculture was the most important predictor of a suicide attempt. Goth identification was a significant independent predictor of deliberate self-harm (odds ratio = 14.2) as was identification with three other music subcultures: punk (OR = 4.4), heavy metal (OR = 3.6), and mosher (OR = 3.5) (Young et al., 2006). The causal nature of the association between Goth and suicide was not fully clear since some Goths attempted suicide prior to joining the Goth subculture. This investigation exemplifies the need to assess new subgenres in the changing world of youth music subcultures, as well as the need for longitudinal research for assisting in unraveling cause and effect. In a cross-sectional study, Lester and Whipple (1996) found no association among any of five musical preference factors (factor 661 scores on which several or more musical preferences loaded) and suicide attempts. However, new and emerging musical preferences thought to be high in suicide risk, such as death metal, black metal, and Goth/industrial/punk were not included on the list of musical preferences. The sample size of 93 may have been too small to generate enough persons who identified with the newer musical subcultures. Suicide Ideation The ten studies on suicide ideation include those based on experimental designs and others based on cross-sectional survey research. The three experimental investigations generally exposed a group of students to a suicide song or set of suicide songs (as defined by the investigators) and compared their suicidality afterward to a control group of students exposed to a song from the same genre but without reference to suicide. Ballard and Coates (1995) conducted an experiment where each of 175 students at Appalachian State University was exposed to either a suicide song (heavy metal or rap) or a nonsuicide song (also heavy metal or rap). The heavy metal suicide song was Falling Asleep and the nonsuicide song metal song was Of Wolf and Man. The rap suicide song was Ever So Clear, while the rap nonsuicide song was Sunny Meadows. There was no difference between groups on suicide ideation measured after such exposure. Two additional experimental studies were undertaken: one on songs and one on videos as a venue. In the first investigation, students were exposed to three suicide songs (including Metallica’s Fade to Black). After such exposure, their suicide ideation level was compared to that of controls who were exposed to three songs from the same artists without any specific reference to suicide. In the other investigation, a music video with (Jeremy) and without reference to suicide (Even Flow) were employed. Both music videos were by the same artist, Pearl Jam (Rustad et al., 2003). Both studies found no significant difference between the 662 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW experimental and control groups in suicide ideation. Seven nonexperimental, survey investigations found mixed support for links between music and suicide ideation. Lester and Whipple (1996), in a study based on 93 college students, measured musical preferences by the extent to which the students liked each of 19 selected musical forms. The data on musical preferences were reduced to five factor scores. The factor best representing heavy metal predicted past suicide ideation, but not present ideation. The factor representing country/western/ show tunes was unrelated to suicide ideation. The factor representing folk/environmental/alternative rock/ new age predicted past suicide ideation. Two additional investigations using a very similar methodology yielded mixed results and a gendered effect (Burge & Lester, 2001; Burge et al., 2002). Three non-U.S. studies found some support for a music–suicidality linkage. A Brazilian study differentiated conventional music genres from alternative ones. The college students preferring conventional genres were significantly less apt than those reporting a preference for alternative music genres to report suicide ideation on the Reasons for Living Scale (Pimental et al., 2009). Martin et al. (1993) explored the relationship between musical preference and suicide ideation among 227 high school students in Australia. Rock/metal fans were more apt than others to have thought about suicide, but the relationship was gendered. A significant association existed for girls (66% vs. 35% for rock/metal vs. nonrock/metal fans), but the difference between categories of male fans, while in the expected direction, did not meet the requirements for statistical significance (31% vs. 14% for rock/metal vs. nonrock/ metal fans). Mixed results were also found in a study of 275 high school students in Canada (Lacourse et al., 2001). tures/preference and suicide attitudes or the degree of acceptability of suicide. In two investigations (Stack, 1998, 1999), an index of suicide acceptability was created out of four questions from the General Social Surveys: Is suicide acceptable in the case of (1) a terminal illness, (2) bankruptcy, (3) dishonor, and (4) tired of living. Liking heavy metal music predicted suicide acceptability at both the bivariate and multivariate analysis of this index (Stack, 1998). Liking blues music predicted suicide acceptability only indirectly through blues’ being associated with lowered religiosity (Stack, 1999). A study of 121 high school students in the Midwest found a gendered relationship between metal fanship and suicide attitudes. Based on the questions regarding the moral objections for suicide in the Reasons for Living scale, students who strongly liked or liked heavy metal music were significantly higher in suicide acceptability (Scheel & Westfield, 1999). However, male fans had higher suicide acceptability scores than nonmetal male fans. In contrast, there was no significant difference between female metal and female nonmetal fans. As in keeping with work on suicide ideation, two experimental studies found no association between exposure to heavy metal music in the laboratory and subsequent suicide acceptability (Rustad et al., 2003). Suicide Acceptability Six articles containing seven studies explored the link between musical subcul- QUANTITATIVE REVIEW OF WORK ON MUSIC AND SUICIDALITY The 21 studies contained a total of 90 findings relating or not relating an indicator of music with an indicator of suicidality. Herein, the finding is the unit of analysis. Following previous work on quantitative reviews (e.g., Stack, 2005), the dependent variable is whether or not a finding represents a significant association between music and suicidality, where a significant finding = 1 and nonsignificant findings = 0. Briefly, 40% of the findings reported a significant association, while 60% did not. Since the dependent STACK AND LESTER variable is a dichotomy, logistic regression techniques are appropriate (Pampel, 2000). A series of independent variables were constructed and used in an effort to predict the presence or absence of reported significant findings. First, following the organization of Table 1, a distinction was made among studies’ outcome measures. Four corresponding variables were created: suicidality measured as a suicide rate (0,1), suicide attempt (0,1), suicide ideation (0,1), and suicide acceptability (0,1). Second, the type of music investigated in the study was coded as a series of binary variables for music types where there were at least six findings: heavy metal (0,1) and country (0,1). In addition, we use a binary variable (0,1) to distinguish between studies that use a factor score where subgenres of music load on the same factor (e.g., Lester & Whipple, 1996), and studies that use a simplified measure of a particular music preference or identification (e.g., Stack, 2002). 663 Audience Characteristics The gender of the audience was coded as male (0,1) or female (0,1), and both female and male served as the benchmark category. Analysis Strategy Studies were differentiated on two design issues. Multivariate findings (1) were distinguished from those derived from bivariate models (0). A distinction was drawn between work based on experimental studies with an experimental and control group (1) and all other studies (0). In addition, sample size was used to predict significant findings, given that all else being equal, larger samples generate a greater chance for significance. Finally, a control was introduced for year of publication. Possibly given methodological and theoretical advances, more recent studies might differ in their findings from older investigations. TABLE 2 The Effect of Study Characteristics on the Reporting of a Significant Finding on Music and Suicide; 20 Studies, 90 Findings, Pearson Correlation Coefficients Variable Dependent variable in study Suicide rates versus all other suicide measures Suicide attempts versus all other suicide measures Suicide ideation versus all other suicide measures Suicide acceptability versus all other suicide measures Music Genre Country music Heavy metal music Use of clusters or factor scores for musical preference Audience Characteristics Males Females Analysis Strategies Multivariate analysis Experimental research design Sample size Year of Publication *p < .05. Correlation between music and the presence of an association with suicidality .249* .147 ).082 .052 ).01 ).031 ).246* ).06 .025 .009 ).218* .334* .316* 664 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW The zero order correlations between each of the above independent variables and a study reporting a music effect on suicidality are presented in Table 2. Studies exploring suicide rates are more likely than studies based on attempts/ideation/acceptability to report a significant association between suicidality and music (r = .249, p < .05). However, the type of suicidality studied does not predict significant music–suicidality associations for the three other categories of investigations (attempts, ideation, and acceptability). more likely than others to report a significant association between music and suicidality. Analysis Characteristics Findings based on a multivariate analysis were no more likely to be significant than those findings coming from bivariate analysis. Investigations based on experimental research designs were less apt than others to report significant findings. Indeed, under closer inspection, all findings from experimental studies on suicide ideation and suicide acceptability were insignificant. As anticipated, the larger the sample size, the more apt a study will report significant associations between music and suicide (r = .334, p < .05). There was also a significant association between year of publication and the probability of reporting a significant finding (r = .316, p < .05). Four significant predictors of positive findings were entered into a multivariate logistic regression model. The results are provided in Table 3. Given invariance in the experimental design variable (all such studies reported insignificant findings), it could not be used as a reliable predictor. From the odds ratios, investigations based on suicide rates are 19.47 times more apt than their counterparts to report significant findings. Larger studies are also significantly more apt to report significant findings: for every additional respondent, the odds of a significant Music Genre The correlation between a measure of country music and suicidality was insignificant. In addition, studies based on heavy metal exposure or identification were, on average, unrelated to suicidality. However, studies that used factor scores instead of a simplified measure of musical preferences or other measures of musical influences, were significantly less apt to report significant findings (r = ).246, p < .05). Audience Characteristics Studies based on males were no more or less apt to report significant relationships between music and suicidality than studies based on females and the total populations. Studies based on females were also not any TABLE 3 The Effect of Study Characteristics on the Odds of Reporting Significant Findings between Music and Suicide: Multiple Logistic Regression Results ( N ¼ 21 studies, 90 findings) Independent Variable (study characteristics) B SE Wald chi-square test Odds Ratio Study based on suicide rates Study uses factor scores of music preference Sample size of study Year of publication Constant term Model chi-square test Nagelkerke r2 Percentage of cases correctly classified 2.96* 0.30 0.002* 0.17* )360.1* 26.2* 0.342 75.56% 1.01 0.60 0.001 0.08 171.1 8.57 0.25 5.13 4.39 4.42 19.27 1.36 1.0025 1.19 – *p < .05. STACK AND LESTER outcome are raised by 0.25%. As anticipated, later studies are more apt than earlier ones to report significant findings. Each year raises the odds of a significant finding by 19%. However, controlling for the other independent variables, the use of factor scores as a measure of musical preference is no longer a significant predictor of study outcomes. The model provides a good fit as witnessed by model chi-square test. Fully 75.56% of the 90 findings are correctly classified. From the Nagelkerke r2, 34.2% of the variance is explained. The weakest results for building a link between music and suicide are from the several experimental studies of suicide ideation and suicide acceptability (e.g., Rustad et al., 2003). For example, there is no evidence whatsoever in these investigations that exposure to a suicide song or music video enhances suicide ideation compared to exposure to a nonsuicide song of the same genre. In an effort to explain these negative results, we turn to a brief content analysis of a sample of suicide songs and compare their nature to nonsuicide songs. If the content of songs in a musical genre (e.g., themes of sadness, negative affect, and death) do not vary between its suicide songs and nonsuicide songs, we would expect the nil findings in this body of experimental research. No study to date has performed such a content analysis. We briefly fill this gap in the research. EXTENSION: CONTENT OF SUICIDE SONGS VERSUS NONSUICIDE SONGS Method A Google search was carried out using ‘‘suicide songs’’ as a keyword. Some caution needs to be exercised in using Google as a research tool. There is a bias when providers and people who post on Google can artificially inflate relevance of data. This potential problem was at least partially circumvented by the inspection of a series of posts on famous suicide songs from a variety of 665 sources. The first five lists of suicide songs compiled by a variety of sources, including music enthusiasts, were selected for further review. These Internet-based lists of potential songs for inclusion included Wikipedia’s list of 108 suicide songs (category: songs about suicide; en.wikipedia.org), Top Ten Songs About Suicide (media/gunaxin.com), Top 10 Most Depressing Rock Songs (listserve.com/ 2007/11/05/top-ten-most-depressing-rocksongs), Don’t Try to Wake Me Up in the Morning: 36 Songs to Soundtrack Your Suicide (http://www.avclub.com), and Songs About Suicide (http://www.eopinions.com). The lists were scrutinized for suicide songs that met four criteria: (1) the song had to be on at least one, and preferably two, of the five lists scrutinized, most of the songs were on at least two lists; (2) the song was published and aired after 1965; (3) the song needed to have been played in the United States; and (4) the song had to have mention of a completed suicide, suicide plan, suicidal ideation, or at least a very strong suggestion of suicidality. The lyrics of all the songs were read, and the songs were listened to. Twenty-four songs made the final list (see Appendix). Examples include Pearl Jam’s Jeremy and Metallica’s suicidogenic Fade to Black (e.g., ‘‘There is nothing more for me, need the end to set me free’’), a famous suicide song in the heavy metal subculture. It was listed as the favorite song by a majority of metal fans in the 1990s (Arnett, 1991). Both songs were used in previous research (e.g., Rustad et al., 2003). Comparison songs were chosen by using the adjoining song on the same CD as the suicide song. The control songs are also listed in the Appendix. In this study we followed general procedures in quantitative narrative analysis, specifically those adopted in suicide narrative analysis (Franzosi, 2010; Lester & McSwain, 2011; Stirman & Pennebaker, 2001). The content of each song was run through the Linguistic Inquiry and Word Count (LIWC) program devised by Pennebaker, Francis, and Booth (2001). This program analyzes written text for 74 language variables. The program can count pronouns (such as ‘‘I’’ 666 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW and ‘‘you’’), emotional words (negative and positive), and words indicating particular content (such as death and dying). The scores are expressed as percentages of the total number of words. The content of the suicide songs was compared with the content of the comparison songs using paired-sample t tests. systematic review of the literature exists on the subject. The present study fills this gap by providing both a systematic narrative review and a quantitative review. Studies based on suicide rates are fully 19 times more likely than other studies to report significant associations between indicators of music involvement and suicidality. Studies with large samples and more recently published investigations are also more apt to report significant results. A key finding is that studies based on ecological data, or suicide rates, are more apt to generate significant findings than other studies. Ecological studies on suicide rates (completed suicides) typically are carried out on the general population, which includes persons in and out of school, while other investigations are based on high school and college student populations. To the extent that members of at risk music subcultures (e.g., goth, death metal, punk) are more likely to drop out of high school, never attend college, and be truant (Gaines, 1991; Weinstein, 1991), investigations of school populations may miss the most at-risk youth and be less apt to find significant associations. The relative lack of significant findings at the individual level for a link between country music and suicide may also be an artifact of overreliance on student samples. Country music has an older audience. For middle-aged subcultures (country music), where suicide risk may come later in the life cycle from stressful life events such as divorce or job loss, the most at-risk population is not high school or college students. However, most work on suicide ideation is based on surveys of young students. On close inspection, some recent investigations follow innovative methodologies, which may help explain the trend toward the reporting of more significant associations between music and suicidality. Nearly all older investigations do not measure the characteristics of music per se (key, low notes), but rather focus on lyrics or fanship. Recently, Lester and Gunn (2011) looked beyond lyrics and focused on the RESULTS Of the 74 variables studied, only two were statistically significant at the 5% level of significance: words per sentence and sad words. The suicide songs had more words per sentence (171.3 vs. 99.4, SDs = 105.6 and 74.9, respectively, t = 2.61, df = 23, twotailed p = .016) and had a higher percentage of sad words (1.23 vs. 0.56, SDs = 1.58 and 0.83, respectively, t = 2.17, p = .041). However, using the Bonferroni technique to correct for multiple statistical tests, none of the differences were significant. In particular, the suicide songs did not differ from the comparison songs in overall positive emotions; overall negative emotions; anger; anxiety; references to the past, present, or future; metaphysical issues; and religion or death, content that might be expected to appear in songs about suicide. Hence, these findings offer a tentative explanation for why previous investigations that exposed subjects to songs from the same album/genre did not find any association between a suicide song and suicidality. Songs from the same album are apt to carry many similar moods and some cognitive messages even if one is alleged to be a ‘‘suicide song.’’ CONCLUSIONS While there have been over 130 investigations to date on media effects on suicidality, the vast majority have been concerned with news reports and, to a lesser extent, with film. In contrast, the link between music and suicide has received less attention, and no STACK AND LESTER percentage of low notes in national anthems. This new measure predicts national suicide rates. Future work is needed to build on Lester and Gunn’s brief analysis by incorporating the empirical and theoretical work found in the literature on music and emotions (e.g., Gagnon & Peretz, 2003; Peretz & Sloboda, 2005). In a similar vein, Young et al. (2006) explored a relatively unstudied suicidogenic culture, Goth culture, and found that Goths are 16 times more apt to have attempted suicide than others. In addition, they measured the degree of identification with the music culture, and found that the higher the degree of identification, the higher the risk of suicide attempts. Finally, Young et al. (2006) conducted an 8-year longitudinal investigation to determine whether suicidal behavior occurs before, after, or both before and after entry into a suicidogenic culture. Such designs can address issues of causality, but are rare in music and suicide studies. Future work is needed on new music–oriented subcultures to the extent that old ones have become main stream or changed in their nature. Emo culture is another example of an under researched suicidogenic youth culture (Baker & Bor, 2008). All findings on the influence of music on suicide ideation and suicide acceptability that were based on experimental research designs were insignificant. Students exposed to suicide songs were not significantly higher in suicidality than those exposed to nonsuicide songs. This nil finding provided a point of departure for our brief auxiliary study. By comparing the lyrical content of suicide songs versus nonsuicide songs by the same artist and album, we determined that the suicide songs did not differ from their counterparts. References to death, the percent of words representing sadness, and so forth were the same between groups. Future work is needed to expose a control group to a song that is markedly different in its nature with, for example, no mention of death and zero percent of sad words. In this fashion, the impact of short-term exposure to a suicide song could be fully assessed. 667 There are legal implications of the present findings on suicide song content. A number of lawsuits have been filed against several band’s suicide songs (e.g., suicide songs by Black Sabbath and Judas Priest; Litman & Farberow, 1994; Stack et al., 1994). Based on our analysis, we determined that the level of dangerousness in the lyrical content of suicide songs does not differ from nonsuicide songs on the same album. This suggests that dangerousness of songs from a particular musical genre is relatively constant across songs of that genre. Hopelessness, for example, is a major theme of heavy metal songs (Weinstein, 1991). This does not mean that suicide songs are, indeed, less dangerous to triggering actual suicides overall, but the main thrust of danger may be the genre itself. Currently, Goth and Emo fans, for example, may be at higher risk of suicide given personal and cultural tendencies toward hopelessness (Baker & Bor, 2008; Young et al., 2006). The full impact of exposure to actual specific suicide songs recommending suicide as a solution may be best understood by taking into account the greater picture of membership in a suicidogenic cultural system. It is important to take into account mediators of links between music and suicide. For example, membership in several musical subcultures was associated with low involvement in organized religion, an involvement thought to protect against suicidality (Stack, 2000). Arnett (1991) was the first to note that heavy metal fans in the 1980s were three times more likely than other youth to be agnostics or atheists (33% vs. 11%). Sometimes the effect of music on suicide risk in the studies reviewed herein is entirely indirect through religion (e.g., Stack, 1998, 1999). Without checking for this and other indirect effects, investigations may miss the full impact of music on suicidality. At the macro level, country music’s main impact on urban suicide rates is its association with divorce rates (Stack & Gundlach,1995). Several studies failed to check for indirect effects where their correlation matrices suggested them. For example, drug use was one of the strongest predictors of suicide 668 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW risk in the multivariate models in a Canadian study by Lacourse et al. (2001). In that study, most bivariate relationships between metal indicators and drug use were significant. This suggests that metal may affect suicide risk indirectly through affecting drug use. Many music-based subcultures with suicidal tendencies are small and are often working class oriented (Definis-Gojanovic, Gugic, & Sutlovic, 2009; Young et al., 2006). As such, there may not be enough members for a meaningful analysis in the typical samples based on 80 to 200 students. To fully capture the highly suicidogenic persons in small music subcultures, a new sampling strategy is necessary. Such strategies may involve investigations based at concerts where the bands of the musical genre play (e.g., Weinstein, 1991), samples drawn from Web pages dedicated to the music subculture, and music stores or clubs frequented by members of the subculture (e.g., KahnHarris, 2007). Future work is needed to integrate work on music as a medium to the study of music and suicide. People often become fixated on a piece of music, playing a particular song over and over again for hours and, therefore, music may have a different psychological impact than, say, reading a novel. This type of fixation may be more common in particular types of individuals, such as those with Asperger’s syndrome (http:// www.psychforums.com/asperger-syndrome/ topic62867.html). Music may run through a person’s mind, almost like an obsessive thought (Brown, 2006) and may also constitute the content of an hallucination (Hermesh et al., 2004), but such fixations may not always be pathological (Hymer, 1984). Research at the individual level could be improved by adopting new measures of the degree of fanship in a music subculture. For example, one might use the number of hours spent listening to a musical genre or the number of concerts attended. An Internet-based study determined that the self-identified heavy metal fans attended, on average, 16 concerts a year, but there was a large disparity among the fans with a standard deviation of 19 concerts (Recours et al., 2009). Some went to as many as 100 concerts a year! While a preference for heavy metal is a commonly used measure, the degree of involvement in heavy metal culture needs to be measured as well. In a similar fashion, the degree of exposure to suicidogenic songs could be measured by content analyses of the favorite song/artist of respondents. Primark, Douglas, Fine, and Dalton (2009) have developed a methodology for estimating the amount of exposure to degrading sexual lyrics based on the favorite artist of the respondent. Such exposure is a significant predictor of a range of actual self-reported sexual behaviors in youth. The often found link between select musical subcultures and suicide is echoed in the broader literature on the relationship between music in general and select musical subcultures in particular and risk positive cognitions, emotions, and behaviors (e.g., Fischer, Greitemeyer, Kastenmuller, Vogrinic, & Sauer, 2011; Primark, Kraemer, Fine, & Dalton, 2009). A meta-analysis on risk glorifying media exposure determined that exposure to risk glorifying music, in particular, tended to be associated with risky behaviors, cognitions, and emotions. The average effect size across studies of the association between music indicators and such behaviors as drug addiction, alcohol abuse, and delinquency was g = 0.39 (Fischer et al., 2011). Several reviews and a meta-analysis have determined that the number of selfreported voluntary exposures to movies where the stars smoke is often the best single predictor of smoking initiation and continuation among teens (e.g., Morgenstern et al., 2011). In a similar vein, the number of selfreported exposures to movies with incidents of drug use, the greater the odds of drug use among adults (Hunt, Sweeting, Sargent, Lewars, Young, & West, 2011). Given such findings on other risk behaviors, future research is needed to assess the degree of voluntary exposure to a list of well-known suicide songs to see whether cumulative exposure predicts suicidality. Research on STACK AND LESTER music and suicide tends to measure exposure in qualitative terms such as one’s musical preference, not the degree of voluntary (nonexperimental) exposure over time to the most dangerous songs. Finally, future work should address ethical issues and cult worship. Individual-based research on music and suicide can be filled with ethical quandaries, including perceptions that exposure to suicidal music content as well as questions regarding the subject’s suicidal behavior can place subjects at risk. However, available research on this matter suggests minimal danger to subjects from 669 such exposure; repeated questions on suicide ideation asked to the same adolescent or adult subjects over time indicate that there is no relationship between being repeatedly asked about suicidal thoughts and subsequent increased suicidal behavior (Cukrowitz, Smith, & Poindexter, 2010; Mathias et al., 2012). In addition, future research could expand on our reference to musical subcultures. It is not only the social construct but the idealization of the musician (singer/band) itself that can go as far as fanatical cult worship (e.g., Kurt Cobain–copy cat suicides). REFERENCES ARNETT, J. (1991). Adolescents and heavy metal music: From the mouths of the metal heads. Youth and Society, 23, 76–98. BAKER, F., & BOR, W. (2008). Can music preference indicate mental health status in young people? Australian Psychiatry, 16, 284–288. *BALLARD, M. E., & COATES, S. (1995). The immediate effects of homicidal, suicidal, and nonviolent heavy metal and rap songs on the moods of college students. Youth and Society, 27, 148–168. BROWN, S. (2006). The perpetual music track. Journal of Consciousness Studies, 13, 43–62. *BURGE, M., GOLDBLAT, C., & LESTER, D. (2002). Music preferences and suicidality: A comment on Stack. Death Studies, 26, 501–504. *BURGE, M., & LESTER, D. (2001). Predicting suicide ideation in high school students. Psychological Reports, 89, 283–284. CROSBY, A. E., HAN, B., ORTEAG, L. V. A. G., PARKS, S. E., & GFROERER, J. (2011). Suicidal thoughts and behaviors among adults aged >18 years, U.S., 2008–2009. MMWR Surveillance Summaries: Morbidity and Mortality Weekly Report, 60, 1–22. CUKROWITZ, K., SMITH, P., & POINDEXTER, E. (2010). The effect of participating in suicide research: Does participating in a research protocol on suicide and psychiatric symptoms increase suicide ideation and attempts? Suicide and LifeThreatening Behavior, 40, 535–543. DARNLEY-SMITH, R., & PATELY, H. M. (2003). Music therapy. Thousand Oaks, CA: Sage. DEFINIS-GOJANOVIC, M., GUGIC, D., & SUTLOVIC, D. (2009). Suicide and emo youth subculture. Collegium Antropologicum, 33(Suppl 2), 173–175. FISCHER, P., GREITEMEYER, T., KASTENMULLER, A., VOGRINIC, C., & SAUER, A. (2011). The effects of risk glorifying media exposure on risk positive cognitions, emotions, and behaviors: A meta analytic review. Psychological Bulletin, 167, 367–390. FRANZOSI, R. (2010). Quantitative narrative analysis. Beverly Hills, CA: Sage. GAGNON, L., & PERETZ, I. (2003). Mode and temp relative contributions to ‘‘happy-sad’’ judgments in equitone melodies. Cognition and Emotion, 17, 25–40. GAINES, D. (1991). Teenage wasteland. New York: Pantheon. *GUNDLACH, J., & STACK, S. (2010). The effect of a popular country song on suicide: Whiskey Lullaby. Unpublished manuscript. HAMM, M. S. (2004). Apocalyptic violence: The seduction of terrorist subcultures. Theoretical Criminology, 8, 323–339. HERMESH, H., KONAS, K., SHILOH, R., REUVEN, D., MAROM, S., WEIZMAN, A., et al. (2004). Musical hallucinations. Journal of Clinical Psychiatry, 65, 191–197. HUNT, K., SWEETING, H., SARGENT, J., LEWARS, H., YOUNG, R., & WEST, P. (2011). Is there an association between seeing incidents of alcohol or drug use in films and young Scottish adults’ own alcohol or drug use? BMC Public Health, 11, 259. HYMER, S. M. (1984). Absorption as a therapeutic agent. Journal of Contemporary Psychotherapy, 14, 93–108. KAHN-HARRIS, K. (2007). Extreme metal: Music and culture on the edge. London: Berg. *LACOURSE, E., CLAES, M., & VILLENEUVE, M. (2001). Heavy metal music and adolescent suicide risk. Journal of Youth and Adolescence, 30, 321–332. 670 MUSIC AND SUICIDALITY: QUANTITATIVE REVIEW *LESTER, D., & GUNN, J. F. (2011). National anthems and suicide rates. Psychological Reports, 108, 43–44. LESTER, D., & MCSWAIN, S. (2011). A text analysis of the poems of Sylvia Plath. Psychological Reports, 109, 73–76. *LESTER, D., & WHIPPLE, M. (1996). Music preference, depression, suicidal preoccupation, and personality. Suicide and Life-Threatening Behavior, 26, 68–70. LITMAN, R., & FARBEROW, N. (1994). Poprock music as precipitating cause in youth suicide. Journal of Forensic Science, 39, 494–499. *MAGUIRE, E., & SNIPES, J. (1994). Reassessing the link between country music and suicide. Social Forces, 72, 1239–1243. *MARTIN, G., CLARKE, M., & PEARCE, C. (1993). Adolescent suicide: Music preference as an indicator of vulnerability. Journal of the American Academy of Child and Adolescent Psychiatry, 32, 530– 535. MATHIAS, C. W., FURR, R. M., SHEFTALL, A. H., HILL-CAPTURCZAK, N., PAIGE, C., & DOUGHERTY, D. M. (2012). What’s the harm in asking about suicidal ideation? Suicide and LifeThreatening Behavior, 43, 341–351. MORGENSTERN, M., POELEN, E. A. P., SCHOLTE, R., KARLSDOTTIR, S., JONSSON, S. H., MATHIS, F., et al. (2011). Smoking in the movies and adolescent smoking: Cross cultural study in six European countries. Thorax, 66, 875–883. PAMPEL, F. (2000). Logistic regression. Thousand Oaks, CA: Sage. PENNEBAKER, J. W., FRANCIS, M. E., & BOOTH, R. J. (2001). Linguistic inquiry and word count. Mahwah, NJ: Erlbaum. PERETZ, I., & SLOBODA, J. (2005). Music and the emotional brain. Annals of the New York Academy of Science, 1060, 409–411. PERETZ, I., & ZATTORE, R. J. (2005). Brain organization for music processing. Annual Review of Psychology, 56, 89–114. PETERSON, R. J., SAFER, M. A., & JOBES, D. A. (2008). The impact of suicidal rock music lyrics on youth: An investigation of individual differences. Archives of Suicide Research, 12, 161– 169. *PIMENTAL, C. E., GOUVEIA, V. V., DE SANTANA, N. L., CHAVES, W. A., & RODRIGUES, C. A. (2009). Preferencia musical e risco de suicido entre jovens. Journal of Brasilian Psychiatry, 58, 26– 33. PIRKIS, J., & BLOOD, R. W. (2001). Suicide and the media, Part II: Portrayal in fictional media. Crisis, 22, 155–162. PRIMARK, B. A., DOUGLAS, E. L., FINE, M. J., & DALTON, M. A. (2009). Exposure to sexual lyrics and sexual experience among urban adolescents. American Journal of Preventive Medicine, 36, 317–323. PRIMARK, B. A., KRAEMER, K. L., FINE, M. J., & DALTON, M. A. (2009). Media exposure and marijuana and alcohol use. Substance Use and Misuse, 44, 722–739. RECOURS, R., AUSSAGUEL, F., & TRUJILLO, N. (2009). Metal music and mental health in France. Culture, Medicine and Psychiatry, 33, 473– 488. RENTFROW, P. J., GOLDBERG, L. R., & ZILCA, R. (2011). Listening, watching, and reading: The structure and correlates of entertainment preferences. Journal of Personality, 79, 223–257. RIHMER, Z. (1997). Studies of suicide and suicidal behavior in Hungary. In L.L. JUDD, B. SALETU & V. FILIP (Eds.), Basic and clinical science of mental and addictive disorders (pp. 171–174). Basel, Switzerland: Karger. *RUSTAD, R., SNALL, J. E., JOBES, D. A., SAFER, M., & PETERSON, R. (2003). The impact of rock videos and music with suicidal content on thoughts and attitudes toward suicide. Suicide and Life-Threatening Behavior, 33, 120–131. *SCHEEL, K. R., & WESTFIELD, J. (1999). Heavy metal music and adolescent suicidality: An empirical investigation. Adolescence, 34, 253–273. SNELL, D., & HODGETTS, D. (2007). Heavy metal, identity, and social negotiation in a community of practice. Journal of Community and Applied Social Psychology, 17, 430–445. *STACK, S. (1998). Heavy metal, religiosity and suicide acceptability. Suicide and Life-Threatening Behavior, 28, 388–394. *STACK, S. (1999). Blues fans and suicide acceptability. Death Studies, 24, 223–231. STACK, S. (2000). Suicide: A fifteen year review of the sociological literature. Part II: Modernization & social integration perspectives. Suicide and Life-Threatening Behavior, 30, 163–176. *STACK, S. (2002). Opera subculture and suicide for honor. Death Studies, 26, 431–437. STACK, S. (2005). Suicide in the media: A quantitative review of studies based on nonfictional stories. Suicide and Life-Threatening Behavior, 35, 121–133. STACK, S. (2009). Copycat effects of fictional suicides: A meta analysis. In S. STACK & D. LESTER (Eds.), Suicide in the creative arts (pp. 231–244). New York: Nova Science. STACK, S., & BOWMAN, B. (2012). Suicide movies, social patterns: 1900–2009. Cambridge: Hogrefe. *STACK, S., & GUNDLACH, J. (1992). The effect of country music on suicide. Social Forces, 71, 211–218. STACK, S., & GUNDLACH, J. (1994). Country music and suicide: A reply to Maguire and Snipes. Social Forces, 72, 1245–1248. STACK AND LESTER 671 *STACK, S., & GUNDLACH, J. (1995). Country music and suicide: Individual, indirect, and interaction effects. Social Forces, 74, 331–335. *STACK, S., GUNDLACH, J., & REEVES, J. (1994). The heavy metal subculture and suicide. Suicide and Life-Threatening Behavior, 24, 15–23. STIRMAN, S. W., & PENNEBAKER, J. W. (2001). Word use in the poetry of suicidal and nonsuicidal poets. Psychosomatic Medicine, 63, 517– 522. WEINSTEIN, D. (1991). Heavy metal: A cultural sociology. New York: Macmillan. *YOUNG, R., SWEETING, H., & WEST, P. (2006). Prevalence of deliberate self harm and attempted suicide within contemporary Goth youth subculture: Longitudinal cohort study. British Medical Journal, 332, 1058–1061. *Study in quantitative review. Manuscript Received: January 25, 2012 Revision Accepted: July 1, 2012 APPENDIX Group 1. Metallica 2. Megadeath 3. Ozzy Osbourne 4. Blink 182 5. Papa Roach 6. Third Eye Blind 7. Good Charlotte 8. 3 Doors Down 9. Pearl Jam 10. P.O.D 11. Joy Division 12. Pantera 13. The Replacements 14. Damien Jurado 15. Eminem 16. The Smiths 17. Suicide 18. Elliott Smith 19. Eminem 20. Gilbert O’Sullivan 21. Slipknot 22. Sleater-Kinney 23. Unwritten Law 24. Eels Suicide song Control song Fade to Black A tout le monde Suicide Solution Adam’s Song Last Resort Jumper Hold On Loser Jeremy Youth of the Nation Exercise One Suicide note, part one The Ledge Tonight I Will Retire Space Bound Asleep Frankie Teardrop King’s Crossing Stan Alone Again Wait and Bleed Jumpers Teenage Suicide Elizabeth in the Bathroom For Whom the Bell Tolls Elysian Fields Mr. Crowley All the Small Things Broken Home Semi-charmed Life Riot Girl Duck and Run Once BOOM! Ice Age 13 Steps to Nowhere Nightclub Jitters Great Today Going Through Changes The boy with the Thorn in His Side Ghost Rider A Fond Farewell Kill You Ooh-wakka-doo-wakka-day Spit It Out What’s Mine Is Yours Sorry Going to Your Funeral