Trends in Explicit Portrayal of Suicidal Behavior in Popular U.S. Movies, 1950-2006

Trends in movie suicidal behavior portrayal may reflect greater societal acceptance of suicide with potential adverse effects on adolescents. To assess the potential for such adverse effects, explicit portrayals of suicidal behavior and the ratings of films were coded in top-grossing U. S. movies from 1950-2006 (N=855). Suicidal behavior portrayal in films increased linearly from 1950 to 2006. From 1968-1984, movies rated R by the Motion Picture Association of America had 5 times more highly explicit suicide behavior portrayals than did G/PG films. After the adoption of the PG-13 category in 1985, PG-13 and R films were indistinguishable on this measure. The results indicate the need for further study of the effects of suicidal behavior portrayals on adolescent movie audiences.

Suicide is the third leading cause of mortality for young people ages 15 to 24 in the U. S. (Centers for Disease Control and Prevention, 2009) with the rate nearly tripling from 1960 to 1990 (Centers for Disease Control and Prevention, 1994). Many explanations have been suggested for this rise, including changes in demographics (Brenner, 1973; Brenner, 1979; Easterlin, 1980; Holinger, Offer, Barter et al., 1994) and increases in various forms of social disorganization (Stack, 2000). For example, the size of the baby boom generation may have increased strain on this demographic cohort (Holinger, Offer, Barter et al., 1994) and rising rates of divorce have been linked toyouth suicide rates (Stack, 1990; Stack, 1981). Growing rates of major depression in adolescents and young adultsmay also have contributed to the rise in youth suicide(Wickramaratne & Weissman, 1996).

Aside from the above explanations, there also has been a steady increase in societal acceptance of suicide (Romer & Jamieson, 2003). Approval of suicide has increased for all age groups since the General Social Survey began tracking it in 1977 with young people ages 18 to 24 reporting more acceptance than those over the age of 50 (Romer & Jamieson, 2003). Furthermore, adolescent acceptance of suicide is related to reports of actually planning a suicide. In a recent study, young Americans were over 14 times more likely to have reported planning to kill themselves if they believed that suicide was an acceptable solution to life problems (Joe, Romer, & Jamieson, 2007).

Just as the decline in portrayal of tobacco use in movies paralleled reductions in cigarette consumption (Jamieson & Romer, 2010), a rise in movie suicidal behavior portrayal may reflect greater acceptance of suicide in the culture. Since adolescents are avid movie consumers and movie sales are large [worldwide box office was $29.9 billion in 2009 (MPAA, 2009)], this is cause for concern.

Media portrayal can influence the acceptance of and imitation of behavior. According to Bandura’s(2002) Social Cognitive Theory of Mass Communication (SCTMC), the more a behavior is shown as a potential solution to a personal problem, the more likely it is to be seen as normative. Indeed, repeated portrayal in the media of an otherwise proscribed behavior can cultivate beliefs about the acceptance of the behavior in the world (Bandura, 2002). The SCTMC also predicts that exposure to a novel behavior in the media creates representations in memory that make it more available for enactment. In particular, exposure to highly explicit suicidal behavior portrayal that dramatizes the attempted or completed taking of a character’s life can suggest novel and more lethal methods of suicide than might otherwise be contemplated (Fekete & Macsai, 1990).

Considerable research indicates that exposure to suicidal behavior portrayals can lead to suicide contagion (Kunrath, Baumert, & Ladwig, 2010; Insel & Gould, 2008; Phillips, Lesyna, & Paight,1992). Adolescents are particularly susceptible to this form of media influence (Gould & Shaffer, 1986; Gould, 1990; Gould, Jamieson, & Romer 2003; Romer, Jamieson, & Jamieson, 2006). Adolescents with a variety of risk factors, such as psychiatric illness, family history of suicidal behavior, poor communication between child and parent, and stressful life events, may be especially vulnerable to media portrayals of suicide (Gould, Fisher, Parides et al., 1996; Gould, Greenberg, Velting et al., 2003).

Although there is stronger evidence for non-fictional than fictional portrayals influencing suicide (Centers for Disease Control and Prevention 2001; Gould, Jamieson, & Romer 2003; Pirkis & Blood, 2001), fictional portrayals also have been shown to produce contagion (Fowler, 1986; Gould & Shaffer, 1986; Gould, Shaffer, & Kleinman,1988; Holding, 1974; Holding, 1975; Schmidtke & Hafner, 1988). For example, Hawton and colleagues (Hawton, Simkin, Deekset al., 1999) reported a 17% increase in self-poisoning in the week following portrayal of such a method in a television drama. Indeed, reports from U.K. emergency rooms indicated 20% of patients who were interviewed reported they had seen the show and indicated it had influenced their choice of method.

Industry Efforts to Control Adolescent Exposure to Harmful Content

A major driver of suicidal behavior portrayal in U. S. films could have been the lifting of restrictions against such content when the “production code” was abandoned by the industry in 1968 (Nalkur, Jamieson, & Romer, 2010). The code prohibited explicit portrayal of a wide range of content including violence directed at others or the self. Beginning in 1968 however, the Motion Picture Association of America (MPAA) no longer restricted this content but instead provided parents with ratings information about its presence. To implement this system, the MPAA set up the Classification and Rating Administration (CARA) to rate all films distributed in the U. S. CARA employs a Board of unnamed parents who are charged with the general guideline to:

“…reflect what they believe would be the majority view of their fellow American parents in assigning a rating to a film. This gives the system a built-in mechanism to continually evolve with modern parental concerns. For instance, as society has grown increasingly concerned about drug use since the 1970s, the rating system, too, has reflected that growing parental concern.” (Classification and Rating Administration, 2010)

In implementing this charge, the Board originally assigned ratings using a three-tiered system: A general category (G) was distinguished from a parental guidance category (PG) to warn parents about films that might not be appropriate for all children. In addition, a restricted (R) rating was assigned to films that should not be seen by those under the age of 17 unless accompanied by a parent or other adult. In 1984, an additional rating category was added to differentiate among movies that had previously been rated as PG. This PG-13 rating was meant to warn parents that a movie might not be appropriate for a child under age 13.

Because this system of ratings could serve to reduce the effects on adolescents of exposure to highly explicit forms of suicidal behavior in films, it is important to examine trends in suicidal behavior content as a function of film ratings. This is especially critical because the fact that the CARA Board is sponsored and paid by funds from the movie industry, it is subject to an obvious conflict of interest in its assignment of ratings. Rather than rating films solely on the basis of potential adverse effects of movie content, the Board may also be sensitive to the economic expectations of the industry. Consistent with this concern, there has been a strong rise in explicit movie violence in general over this period of which explicit suicidal behavior portrayal may be a part. Furthermore, the rating system has been found to assign films with highly graphic violence that would have formerly been rated as R to the less restrictive PG-13 category (Nalkur, Jamieson, & Romer, 2010). Because the PG-13 category has the potential to draw a larger audience than an R-rated film, placing a film in the R category is likely to reduce its box office sales. While the assignment of the PG-13 rating to films containing highly explicit violence has been documented, it not clear whether a similar pattern has developed in regard to explicit forms of suicidal behavior.

The Current Study

In this study, we examine trends in explicit suicidal behavior portrayal in top-grossing movies since 1950 to determine whether portrayal of suicidal behavior has increased since the removal of the production code in 1968. Very little research has examined trends in suicidal behavior portrayal in popular films. Trewavas, Hasking, & McCallister 2010 examined a sample of 41 movies and found that non-suicidal self-injury was prominently featured. We have previously found that all forms of explicit violence have increased in popular films since the abandonment of the production code (Nalkur, Jamieson, & Romer, 2010). However, little is known about the effectiveness of the MPAA rating system for shielding children and adolescents from the explicit portrayal of suicidal behavior. If the CARA Board were sensitive to the potential effects on children and adolescents of exposure to explicit portrayals of suicidal behavior, it would assign films with this content to the R category.

We hypothesized that consistent with the rise in societal acceptance of suicide, popular films would increasingly display suicidal behavior and second, that just as with explicit violence in popular PG-13 and R rated movies from 1985 to 2006(Nalkur, Jamieson, & Romer, 2010), adolescent accessible PG-13 movies would contain a large amount of explicit suicidal behavior content that would previously have been rated R. At the same time, we did not expect to see much change in the least restrictive G/PG categories.

We also recognized that not all portrayals of suicidal behavior might pose the same risk of contagion for young people. It might be expected that highly explicit portrayals that showed both the method and its use would be particularly susceptible to imitation (Insel & Gould, 2008; Fekete & Macsai, 1990). The age of the character committing the suicidal act may influence identification with the victim (Phillips, 1979; Schmidtke & Hafner, 1988). If trends in movie suicidal behavior portrayal involved primarily older adults (over age 40), it might pose less of a risk of contagion for a younger audience. Generalizing from the media violence literature, we also would not expect comedic suicidal portrayals to be as influential on an audience as more serious (non-comedic/dramatic) portrayals(King, 2000; Potter & Warren, 1998). Hence, we examined the ages of movie characters and the comedic context of the suicidal behavior portrayal.

We examined trends in suicidal behavior portrayal in top-grossing films from 1950 to 2006 and for highly explicit suicidal behavior content by movie rating since the introduction of the MPAA rating system in 1968 and also since the first full year of the PG-13 rating in 1985. In particular, we evaluated whether highly explicit portrayal of suicidal behavior has become more available to younger audiences since the introduction of the PG-13 category. We examined these questions using a database of 855 movies (15 per year) that had been content coded for portrayals of suicidal behavior among the top-30 grossing U.S. movies from 1950 to 2006.

Method

Movie Sample

The top-30 grossing movies per year from 1950 to 2006 were identified from Variety magazine’s annual lists. To identify trends in amount and explicitness of suicidal behavior, a representative half-sample of these movies was selected for content coding (N= 855). Every second movie rank was included in the sample with the starting rank (1st or 2nd) determined randomly. For the roughly 5% of cases in which movies were not available for purchase, the next ranking movie was substituted. Movie rating was identified from movie VHS tapes and DVD video boxes. The study did not involve human subjects, thus I.R.B. approval was not sought.

The Suicidal Behavior Explicitness Scale

Suicidal behavior was defined as actions in which a character had the “option of living but attempted or completed the taking of his/her life.” Verbal references to suicidal behavior were not coded as suicidal behavior portrayal. A 5-point suicide explicitness behavior scale adapted from Jamieson (2003, p.119) was used to measure explicit suicidal behavior portrayal (see Table 1). This scale was designed to be theoretically consistent with Bandura’s SCTMC (2002) in that higher levels of explicitness corresponded with more modeled suicidal behavior. Levels 1 and 2 on this scale represent two levels of implicit suicide behavior portrayal in which the act is implied but not shown. Levels 3 and 4 represent increasing levels of more explicit portrayal in which at least some of the act is shown.

Suicide Behavior Coding

The coding unit for movies was the five-minute segment. For example, a two-hour movie contained 24 segments. Suicidal behavior coding proceeded in two steps. In the first step, a pool of 23 undergraduate coders evaluated segments for the presence of suicidal behavior. To assess reliability, we used Krippendorff’s (2004) alpha formula, which controls for chance agreement between multiple coders and can handle both nominal and rating scale data such as was used in this study. By this formula, coders achieved a high level of reliability for identifying suicidal behavior content (Yes = 1; No =0;K = .77).

Coders identified 104 out of the 855 movies (12.2%) as having suicidal behavior content, not including references to suicidal behavior. Eight undergraduate coders were trained to evaluate explicit portrayal of suicidal behavior using 18 practice movies containing such portrayal. Each coder then viewed films across the entire study period to evaluate the explicitness of any suicidal behavior segments within those films. For each instance of suicide behavior portrayal, the explicitness of the portrayal was rated using the 1 to 4 scale in Table 1 (K = .84). If more than one occurrence of suicide behavior portrayal appeared in a 5-minute segment, the one with the highest explicitness rating was retained. We calculated the average suicidal behavior explicitness rating (0-4) over all 5-minute movie segments per film and then averaged these scores by year (N =57). As described below, we also evaluated trends over time by MPAA rating category for films released after 1968.

The age of up to 8 maincharacters in each film was coded using a scale with five categories (under age 10, 10-20, 21-40, 41-65, over age 65, K = .80). These ages were then linked to any main characters displaying suicidal behavior, approximately 71% of the portrayals that were identified. As we report below, although there were very few portrayals in the adolescent age range and none under age 10, most of the portrayals by main characters were under the age of 41. Hence, the age scale was recoded to reflect younger(1 = < age 41) versus older (0 = > age 40) suicidal behavior portrayals. Coders also rated each suicidal behavior for its comedic impact, which was defined as “the suicide takes on a comedic role (i.e., meant to get laughs), or is not a serious or realistic suicide attempt.” A parody of a suicide was also considered a comedic suicide. This measure was also highly reliable (K =.70).

Statistical Analyses

Stata 11 (StataCorp, 2009) was used for all analyses. To identify trends for the 57-year study period, we regressed mean annual suicidal behavior portrayal scores on year using ordinary least squares. We tested linear, quadratic, and cubic trends as well as a trend that was stationary from 1950 to 1968 but linear thereafter. All tests of regression parameters used robust standard errors to protect against failures in assumptions of independence and heteroscedasticity in errors (Huber, 1967; White, 1982). We also tested but found no autocorrelation in the series. All tests were two tailed and 95% confidence intervals (CI) were reported.

To assess changes over time in overall as well as the most explicit portrayals of suicide behavior (categories 3 and 4 in Table 1), we examined trends in these portrayals in individual films grouped according to three eras in the study period: the pre-rating era from 1950 to 1967 (N=270), the initial ratings era from 1968 to 1984 (N=242), and the PG-13 era from 1985 to 2006 (N=327). To test the hypothesis that the PG-13 category would permit more explicit suicidal behavior portrayals to be shown to children under age 17 than was the case prior to its introduction, we examined the percentage of films with highly explicit suicidal behavior portrayals in each MPAA rating category across the three ratings eras (see Table 2 for the number of films in each rating category). We collapsed films with ratings of G and PG into one category because there were too few G-rated films to analyze separately. For this analysis, we could not identify a rating for 16 films released after 1968, and these films were treated as missing from the ratings analysis. We compared mean levels of suicidal behavior portrayal between rating categories within and across rating eras using paired and independent t tests. Logistic regression analyses were also conducted to compare films in different rating categories across the three ratings eras with a dependent variable that was coded 1 for films with highly explicit suicidal behavior portrayals (values of 3 or 4) versus 0 for films without highly explicit portrayals of suicidal behavior (values of 0, 1, or 2). Finally, we examined trends over time for the presence of comedic portrayals and for suicidal behavior portrayal by characters ages 10-40.

Results

Trends in Suicidal Behavior Explicitness

The mean suicidal behavior portrayal score across the 57 years was .018 (inter quartile range .007-.027; SD = .016). The best fitting trend was linear with an r2 of .285 (b = .00051, CI = .00029, .00073, p< .001). As seen in Figure 1, this score more than tripled from .0037 in 1950 to .0324 in 2006.[1] The rise in mean suicidal behavior portrayal, shown in Table 2, coincided with increases in the most explicit levels of portrayal in the films across the three eras, rising from 1.1% of all sampled films during 1950 to 1967, to 6.3% during the first ratings era, and then 8.8% after 1985, χ²(2) = 16.815, p< .001. Thus, the rise in portrayal cannot be solely attributed to an increase in implicit levels of portrayal.