Narratives and Mental Illness: Understanding the Factors That Impact Stigmatizing Attitudes and Behavioral Intentions

Abstract

Entertainment television has been explored to reduce stigmatizing attitudes toward mental illness by incorporating positive stories about characters with mental illness. Guided by mediated contact theory and the extended elaboration likelihood model, this study examines whether exposure and engagement with entertainment narratives, featuring characters with mental illnesses of varying levels of public stigma, impacts stigmatizing attitudes and intentions to interact with individuals with mental illness generally. Participants (n = 234) were randomized to one of the three conditions: (1) a more stigmatized mental illness (schizophrenia), (2) a less stigmatized mental illness (depression), or (3) a non-mental illness control (cancer). Participants in the more stigmatized condition reported significantly less identification with characters than those in the less stigmatized condition, and greater identification with the characters were associated with more positive attitudes and behavioral intentions. Narrative counterarguing was associated with less positive attitudes and intentions toward people with mental illness. Implications based on these findings include identifying ways to increase engagement with less familiar mental illnesses to optimize the positive outcomes associated with narrative engagement.

Mental illness, broadly defined, includes personality (e.g., depression and schizophrenia) and substance abuse disorders included in the International Classification of Diseases (ICD) and Diagnostic and Statistical Manual of Mental Disorders (DSM) (Kirschstein, 2000). Although 1 in 5 Americans are diagnosed with a mental health disorder each year (National Alliance on Mental Illness [NIMH], 2021), systematic reviews highlight the importance of reducing public stigma toward mental illness using media messages (Pescosolido et al., 2021; Yu et al., 2023). Stigma is a deeply discrediting attribute that is inherently rooted in social interaction (Goffman, 1963). Stigma toward mental illness in particular refers to negative attitudes and behaviors toward individuals diagnosed with a mental illness (Smith & Applegate, 2018).

Although data suggest that stigmatizing attitudes toward mental illness have decreased over the last decade, stigma associated with more serious mental illnesses (e.g., schizophrenia) remains high (Pescosolido et al., 2021). The public tends to hold more stigmatizing attitudes toward mental illnesses associated with danger (Link & Phelan, 2001; Smith, 2007). Stigma is a public health problem because it can be an obstacle to education, job opportunities, and health services for those with mental illness (Jetten, Haslam, Cruwys, & Branscombe, 2018), which may prevent them from receiving treatment and exacerbate conditions (Corrigan, Markowitz, & Watson, 2004; Yu et al., 2023). To help address this problem, we examine how entertainment narratives may be used to influence and change stigmatizing attitudes held by the public toward those with a mental illness.

Representation of mental illness in the media has been noted to influence public perceptions if mental illness (Smith, 2007; Smith & Applegate, 2018), as well as how individuals with mental illness cope with perceived stigma (Meisenbach, 2010).Although narrative portrayals of mental illness are shown to strengthen stigmatizing attitudes (Gerbner, 1998; Smith, 2007), entertainment narrative, in particular, have been promoted as a means to attenuate stigmatizing attitudes toward mental illness (Ma, 2017; Parcesepe & Cabassa, 2013). Thus, identifying features in narratives that may reduce public stigma is critical to improving the social environment for those with mental illness. Narratives presented on television provide an opportunity for mediated contact (i.e., simulated interactions) with characters who present as having a mental illness. Much in the same way that face-to-face personal contact with someone with a mental illness is associated with less stigmatizing attitudes (Corrigan et al., 2001a), positive interactions with characters with mental illness can also reduce stigmatizing attitudes toward the character (Caputo & Rouner, 2011; Ma, 2017). However, the level of public stigma for a mental illness portrayed in a narrative could impact the degree to which individuals engage with a character experiencing it.

In this study we use mediated contact theory (Ortiz & Harwood, 2007; Schiappa, Gregg, & Hewes, 2005), which draws on contact theory (Allport, 1954) and social cognitive theory (Bandura, 1986, 2001), as well as the extended elaboration likelihood model (E-ELM; Slater & Rouner, 2002) to investigate whether narrative engagement differs between individuals who view a narrative featuring a character with a more stigmatized (schizophrenia) versus a less stigmatized (depression) mental illness. We focus on these two illnesses since they are at differing points along the stigma continuum, with schizophrenia is reported as being a more stigmatized mental illness than depression (Pescosolido et al., 2021). We also seek to identify cognitive factors (e.g., narrative engagement and counterarguing) that may be associated with attitudes and intentions to interact with and support policies for individuals with mental illness.

Theoretical Frameworks

Viewing a narrative featuring a character with a stigmatized illness may allow viewers to experience a sense of personal interaction with the character (Horton & Wohl, 1956; Schiappa, Gregg, & Hewes, 2005). Mediated contact theorizing argues that a narrative featuring a character with a stigmatized illness may leave audiences with a positive or negative perspective of the illness, which results in reduced or increased negative attitudes toward the social group represented by the characters (Park, 2012). Negative perspectives on an illness may occur when messages incite negative cognitive and affective responses (e.g., danger, fear or disgust) among the receivers of the message that, in turn, increase negative attitudes and behaviors toward a stigmatized group (Smith, 2007). Conversely, positive portrayals of mental illness could subdue negative attitudes when audiences are able to empathize or take the perspective of the stigmatized character (Chung & Slater, 2013). For example, reading a vignette about a person with a mental illness that included (versus not included) information about the treatment and healing of that mental illness, was shown to reduce stigmatizing attitudes held by the participant (McGinty, Goldman, Pescosolido, & Barry, 2015).

The E-ELM (Slater & Rouner, 2002) further suggests that the entertaining features of narratives may help to promote audience engagement with a storyline and its characters that can help to overcome message resistance and impact persuasive outcomes. The subtle message of modeling prosocial behavior toward a stigmatized group member within a narrative may also encourage the audience to unknowingly accept the prosocial argument embedded in the narrative (Moyer-Gusé, 2008; Slater & Rouner, 2002). Studies exploring mediated contact have examined a number of character-focused variables, such as parasocial interaction (Schiappa, Gregg, & Hewes, 2005) or character typicality (Joyce, Harwood, & Springer, 2020). However, because the processes of narrative transportation, identification, and counterarguing are most critical to persuasive outcomes in the E-ELM (Slater & Rouner, 2002), they are the focus of this study.

As an individual becomes absorbed or transported into a narrative, the messages embedded within the narrative may impact their beliefs, attitudes, and behaviors (Green & Brock, 2000). Audiences may be more willing to accept narrative messages about individuals with mental illness, as compared to non-narrative messages (Ritterfeld & Jin, 2006). Even if the beliefs of the individual differ from that of the narrative (e.g., holding stigmatizing attitudes toward mental illness), the entertaining features of the narrative could still promote transportation (Chung & Slater, 2013).

In contrast, when viewers experience identification, they feel immersed in the narrative and may have moments where they temporarily take on the perspective of the main character (Cohen, 2001). Narrative features such as story quality and entertainment value induce the identification process, which has been observed to promote empathy with characters representing stigmatized groups (Chung & Slater, 2013; Moyer-Gusé, Dale, & Ortiz, 2019). However, Chung and Slater (2013) note that perspective-taking, “significantly depends upon the audience member’s ability to understand the thoughts and emotions of the character” (p. 906). Thus, it may be more difficult for audiences to understand and identify with characters experiencing rarer and more stigmatized mental health conditions (e.g., schizophrenia versus depression).

Schizophrenia is described as a more stigmatized mental illness than depression; cross-sectional data have consistently demonstrated that public perceives individuals with schizophrenia as more dangerous (Corrigan et al., 2001a; Hazell, Berry, Bogen-Johnston, & Banerjee, 2022; Pescosolido et al., 2020). Survey data also suggest that more stigma, defined as negative attitudes and behaviors, toward individuals with schizophrenia also may be due to less familiarity with the illness (Corrigan et al., 2001a). Considering the prevalence rate of schizophrenia is less than 1% among U.S. adults, compared to 7.8% expressing massive depressive episodes (NAMI, 2023), individuals may indeed be less familiar with schizophrenia than depression. Because dramatized shows are designed to pull audiences into the story, we propose that narrative transportation will be comparable between narratives featuring a character with schizophrenia and a narrative with a character with depression (H1). However, we expect that character identification will be lower among those viewing the narrative featuring the more stigmatized mental illness (schizophrenia) than the less stigmatized mental illness (depression) (H2).

At the heart of the E-ELM (Slater & Rouner, 2002) is the idea that narratives influence individuals via transportation and identification, which, in turn, decrease message resistance and counterarguing. Counterarguing can be described as the extent to which an audience formulates disagreements with a narrative presented (Slater & Rouner, 2002). Less counterarguing is associated with more narrative-consistent attitudes, as the audience is theorized to be too engaged with the narrative to formulate arguments against the persuasive information in the narrative (Moyer-Gusé, 2008). Counterarguing was observed to be lower among audiences exposed to narratives (versus non-narratives) about mental illness (Ma & Nan, 2018). Thus, we propose that transportation (H3) and identification (H4) will be negatively associated with counterarguing, such that those who are more transported and report more identification with characters will engage in less message counterarguing.

Outcomes of Mediated Contact with Characters with Mental Illness

Although stigma operates on different social levels (Link & Phelan, 2001), this study focuses on stigmatized attitudes. Stigmatizing attitudes are negative attributes associated with a particular social group that normalize disease avoidance and discrimination (Phelan, Link, & Dovidio, 2008). Studies have shown that counterarguing is associated with more favorable attitudes toward mental illness (Ma & Nan, 2018). Therefore, we anticipate that exposure to a narrative featuring a character with mental illness will influence counterarguing, and in turn, attitudes (Slater & Rouner, 2002). Specifically, we propose that counterarguing will be positively associated with stigmatizing attitudes (H5) such that those who counterargue with the narrative more will report more stigmatizing attitudes.

Attitudes are a key theoretical predictor of behavioral intentions (Fishbein & Ajzen, 1975). Intentions to avoid or social distance from stigmatized groups is a key measure of public stigma toward mental illness (Corrigan, Edwards, Green, Diwan, & Penn, 2001b). If people perceive individuals with mental illness to be dangerous, then they are more likely to avoid interactions with them or disapprove of them being part of their social circle (Link, Cullen, Struening, Shrout, & Dohrenwend, 1989). These perceptions may even impact whether individuals support mental health policies and services, as attitudes have been observed to be associated with voting intentions regarding stigmatized groups (Slater, Rouner, & Long, 2006). Changes in policies can remove barriers to education, careers, and health services, thus improving the general well-being of those with mental illness (Corrigan, Markowitz, & Watson, 2004). We anticipate that narrative exposure will influence attitudes and behavioral intentions (Figure 1). Specifically, we predict that stigmatizing attitudes will be positively associated with social distancing (H6) and negatively associated with support for policy in favor of those with mental illness (H7), such that those holding more stigmatizing attitudes will be more likely to social distance and less likely to support policies favoring people with mental illness.

Figure 1. Study conceptual Model and direct predictions based on the extended- elaboration likelihood Model and mediated contact theory.

Note. * As compared to a narrative featuring a less stigmatized mental illness or non-mental illness control.

Method

Procedure

After consenting to participate, participants completed a baseline survey (Time 1 (T1)) assessing demographic characteristics, level of familiarity with the depicted illnesses, and baseline attitudes toward cancer and mental illness generally. One week later, participants were randomized to one of the three narrative message conditions, including (1) a more stigmatized mental illness (schizophrenia; n = 78), (2) a less stigmatized mental illness (depression; n = 83), or (3) a non-mental illness control (cancer; n = 73). Cancer was chosen as the non-mental illness control because the disease carries little stigma; cancer patients are perceived as not dangerous (i.e., low avoidance) and not responsible for their disease, apart from lung cancer (Vrinten, Gallagher, Waller, & Marlow, 2019). Immediately after viewing the narrative, participants completed a follow-up survey (Time 2 (T2)) that included questions about narrative engagement, counterarguing, stigmatizing attitudes, and behavioral intentions.

Sample

English-proficient, adult participants were recruited from a student research panel at a large university. They were told that the study aimed to understand how televised narratives impact attitudes toward health conditions. Two hundred and fifty-five participants completed both surveys, but 21 participants were removed due to failing to answer most questions (n = 1), failing multiple attention checks and basic plot questions (n = 3), completing the surveys in less than 5 minutes or more than 24 hours (n = 8), or for more than one of the previous justifications (n = 9). Thus, the effective sample size was 234.

Stimuli

Three episodes of the popular Grey’s Anatomy television show were used because it depicts narratives with depression (season 16, episode 5), schizophrenia (season 6 episode 3), and cancer (season 15, episode 7). The episodes were edited so that all three narratives were a similar length (i.e., about 25 minutes). Content was removed from each episode so the narratives focused only on main characters (Jo Karev, Tom Crawley, and Catherine Fox) and their decision-making regarding treatment and recovery for their respective health conditions. However, the integrity of each episode of Grey’s Anatomy was maintained, including traditional intro and outro voice-overs. Considering participants were watching these episodes on a computer of their choosing, maintaining the entertainment value was important for ecological validity.

Measures

All variables were measured on 1 to 7 Likert scales unless specified otherwise. Confirmatory factor analyses (CFA) was conducted in Mplus (Muthén & Muthén, 2007) to assess scale dimensionality for specific conditions, when appropriate. Model cutoff values for small sample sizes (i.e., n < 1000) described by Hu and Bentler (1999) and Kenny et al. (2015) were used here: Root Mean Square Error of Approximation (RMSEA) < .10, Comparative Fit Index (CFI) > .95, and standardized root mean squared residual (SRMR) < .08.

Narrative Engagement

An 11-item transportation scale (Green & Brock, 2000) was used to measure the extent to which participants felt absorbed into the narrative (e.g., “The narrative affected me emotionally”). The reverse-coded items loaded weakly on a factor and were removed, as in previous studies (Murphy, Frank, Chatterjee, & Baezconde-Garbanati, 2013). In addition, items less relevant to the video narrative (e.g., “While I was watching the show, I could easily picture the events in it taking place”) were dropped from the model because the video provided pictures of the narrative. With five remaining items, the one-factor structure of transportation was confirmed and showed a moderate fit to the data (RMSEA = .08, CFI = .96, SRMR = .03, χ2 (5) = 13.15, p > .01)) with standardized item loadings from .49 to .76. For analyses, the mean of the items was used (Cronbach’s α = .76, M = 4.70, SD = 1.09).

Identification was assessed using Cohen’s (2001) 6-item perspective-taking sub-scale (e.g., “I felt I knew exactly what [character] was going through”). Cohen’s measure also includes an additional four items assessing transportation, which were measured separately as described above, and goal-sharing. We focused only on perspective-taking since it is more theoretically aligned with mediated contact compared to transportation and goal-sharing (Chung & Slater, 2013). The 6-item perspective-taking subscale demonstrated marginal fit to the data (RMSEA = .12, (CI: 0.08–0.16) CFI = .93, SRMR = .04, χ2 (9) = 40.28 (p < .01), with standardized item loadings from .64 to .82. Although the CFI was out of range, the SRMR demonstrates acceptable fit and the confidence interval for the RMSEA includes acceptable cut-off value (RMSEA < .01; Kenny et al., 2015). Cronbach’s alpha for this previously validated scale was also acceptable (α = .87); thus, the mean of the items was used in the analyses (M = 4.33, SD = 1.14).

Counterarguing

A 4-item scale was used to assess counterarguing (Moyer-Gusé, Chung, & Jain, 2011). The measure included items like “the video tried to pressure me to think a certain way” and “the video tried to force its opinions on me.” An initial CFA showed a poor fit to the data (RMSEA = .28, CFI = .82, SRMR = .09, χ2 (2) = 39.56 (p < .01); upon further evaluation, two items (i.e., “I sometimes found myself thinking of ways I disagreed with what was being depicted in the video” and “I found myself looking for flaws in the way information was presented in the video”) were shown to load more weakly onto the factor. These two items were focused on the viewer (finding flaws or disagreements with the narrative) rather than message itself; thus, given our focus on narrative effects, we maintained just the two items focused on the message. The mean of the two items was used in the analysis (Pearson’s r = .75, M = 2.81, SD = 1.28).

Stigmatizing Attitudes

Stigmatizing attitudes were measured at T1 and T2 using a 12-item devaluation-discrimination scale (Link, Mirotznik, & Cullen, 1991). Results of two-factor CFA showed that two items (“Most people feel that entering a hospital for [mental illness/cancer] is a sign of personal failure” and “Most people would not hire a former [mental illness/cancer] patient to take care of their children, even if the person had been well for some time”) loaded weakly onto a single factor a both time points and were dropped; both items suggest that the person with the illness had undergone successful treatment, which has been noted to be less stigmatizing (McGinty, Goldman, Pescosolido, & Barry, 2015). Once removed, the retested two-factor model demonstrated a good fit to the data (RMSEA = .10 [.09–.11], CFI = .83, SRMR = .07, χ2 (169) = 565.84 (p < .01), standardized item loadings from .40 to .91 The mean score of the remaining 10 items were used for analyses (T1: Cronbach’s α = .92, M = 3.33, SD = 1.20; T2: Cronbach’s α = .93, M = 3.34, SD = 1.19).

Behavioral Intent

Behavioral intentions were measured using two variables: policy support and social distancing. A 3-item scale was used to measure support for policies to support those with mental illness, including the likelihood a participant would “sign a petition” or “vote for a political candidate” that promoted resources for individuals with the respective illnesses portrayed in the video (Slater, Rouner, & Long, 2006). Social distancing was measured using a 14-item scale (Arkar & Eker, 1992; Eker & Arkar, 1991). The items asked about the likelihood of the participant interacting in various social settings with the character (e.g., “I would be uncomfortable by [character] becoming my next-door neighbor”). For social distancing, the five reverse-coded items (e.g., “I would play cards, etc. with [character] if I saw her/him in a social gathering”) loaded weakly onto the factor (values in the range of .39 to .60), so those items were removed. After dropping the items, the two-factor CFA for behavioral intentions demonstrated moderate fit to the data (RMSEA = .08, CFI = .92, SRMR = .05, χ2 (53) = 139.00 (p < .01) with standardized loadings between .66 to .93. Therefore, means for policy support (Cronbach’s α = .92, M = 5.63, SD = .24) and social distancing (Cronbach’s α = .93, M = 3.18, SD = 1.08) were calculated for use in the analyses.

Data Analysis

Correlations, means, and standard deviations for model variables are reported in Table 1. After examining descriptive data and correlations between model variables, independent samples t-tests and analysis of variance (ANOVA) were used to assess differences across conditions in IBM SPSS Statistics (version 27) (IBM Corp, 2020). Main study hypotheses were tested using Analysis of Covariance (ANCOVA) due to the inclusion of categorical (condition) and continuous variables. All models controlled for baseline stigmatizing attitudes (T1), previous illness diagnosis and disease familiarity because they may influence the outcomes of interest (Caputo & Rouner, 2011; Corrigan, Edwards, Green, Diwan, & Penn, 2001b). Disease familiarity was measured using a 12-item illness contact report, which inquired if participants (yes or no) “have seen someone with [illness] in passing” or “live with a person who has [illness]” (Holmes et al., 1999). Scores were aggregated, such that a value of one was assigned to each “yes” response and a zero for each “no” response and summed. Indirect effects of identification and transportation on outcomes, serially mediated by counterarguing and stigmatizing attitudes, were also assessed using the PROCESS macro Model 6) for SPSS (Hayes, 2012).

Table 1. Descriptive statistics and correlations between Model variables

Predictors	1.	2.	3.	4.	5.	6.	7.	8.	9.
1. Contact with Illness^a	-
2. Diagnosed with Illness^b	.23**	-
3. T1 Stigmatizing Attitudes	−.05	.15*	-
4. Identification	.01	.05	−.02	-
5. Transportation	.04	−.01	−.17**	.45**	-
6. Counterarguing	−.11	.07	.10	.08	−.07	-
7. T2 Stigmatizing Attitudes	−.04	.26**	.59**	−.15*	−.17*	.17**	-
8. Social Distancing	−.03	.02	.19**	−.26**	−.17*	.28**	.31**	-
9. Policy Support	.08	.12	.01	.05	.22**	−.26**	.01	−.15*	-
Mean	7.81	0.19	3.33	4.33	4.70	2.81	3.34	3.18	5.63
SD	3.33	0.39	1.20	1.14	1.09	1.28	1.20	1.08	1.24

Note. ^aSum used for analysis, ^bNo = 0, Yes = 1, T1 signifies baseline measures, T2 signifies posttest measures, * p < .05, ** p < .01.

Results

Demographics and Individual Characteristics

Participants in this study were relatively young (M = 19.71; SD = 2.07) and the majority were non-Hispanic White (61.1%; n = 143) and female (72.6%; n = 170). One participant had been previously diagnosed with cancer (0.4%; n = 1), and 29.1% (n = 68) of the sample had been diagnosed with depression or anxiety. Overall, 18.4% (n = 43) of participants had been previously diagnosed with the illness depicted in the narrative to which they were exposed.

Manipulation Check

We first assessed the prediction that stigmatizing attitudes toward mental illness generally were higher than for cancer at T1. Results of an independent samples t-test showed that, at T1, stigmatizing attitudes toward cancer (M = 2.62; SD = 0.90) were lower than stigmatizing attitudes toward mental illness (M = 3.65; SD = 1.18), t (231) = 6.62 (p < .01), which suggests the cancer narrative was an appropriate control. In addition, paired sample t-tests showed no significant differences in stigmatizing attitudes toward people with the respective illnesses (i.e., mental illness and cancer) between T1 and T2 (p > .05).

Direct Relationships Between Model Variables

As shown in Table 2, no significant differences were detected across conditions on transportation (p > .05; H1 supported). This finding was not likely the result of ceiling effects given the variable skewness (−0.46) and kurtosis (0.26) did not have an absolute value higher than one (George & Mallery, 2010). However, significant differences were detected across conditions on identification (F(2) = 15.72; CI = 95%; p < .05). As expected, those in the schizophrenia condition (M = 3.86; SD = .13) reported less identification than those in the depression condition (M = 4.84; SD = .13; Mean Diff. = −.98; SE = .18; p < .05; H2 supported). Participants in the cancer condition also reported less identification than those in the depression condition (M = 4.24; SD = .14; Mean Diff. = −.60; SE = .19; p < .05).

Table 2. Analysis of covariance predicting narrative engagement and counterarguing

	Model 1 n = 226	Model 2 n = 226	Model 3 n = 226
	Identification	Transportation	Counterarguing
	F(df)	F(df)	F(df)
Condition	15.72(2)**	0.24(2)	0.47(2)
Contact with Illness	0.00(1)	0.21(1)	3.30(1) ^a
Diagnosis with Illness	0.13(1)	0.03(1)	1.13(1)
T1 Stigmatizing Attitudes	.24(1)	5.16(1)*	.71(1)
Identification			5.10(1)*
Transportation			2.54(1)
Counterarguing
T2 Stigmatizing Attitudes
Error	220	220	218
R^2	0.13	0.03	0.05

Note. ^ap < .10, *p < .05, **p < .01.

In terms of the effects of narrative engagement on counterarguing, as shown in Table 2, transportation was not (p > .05) associated with counterarguing, and identification was positively rather than negatively associated with counterarguing (b = .20; SE = .09; p < .05) (H3 and H4 not supported). However, when controlling for T1 (baseline) stigmatizing attitudes, identification and counterarguing were associated with T2 stigmatizing attitudes toward the illnesses depicted in the narrative; greater identification was associated with lower stigmatizing attitudes toward individuals with the depicted illness (b = −.18; SE = .07, p < .05), whereas more counterarguing was associated with more stigmatizing attitudes toward individuals with the depicted illness (b = .11, SE = .05; p < .05; H5 supported).

Finally, as shown in Table 3, counterarguing was positively associated with intentions to social distance (b = .22; SE = .05, p < .05) and negatively associated with intentions to support mental illness-related policies (b = −.24, SE = .06, p < .05). T2 stigmatizing attitudes toward individuals with the depicted illness were also positively associated with social distancing (b = .13; SE = .06; p < .05; H6 supported), but unassociated policy support (p > .05; H7 not supported). Additionally, the effects of narrative engagement (identification and transportation) on behavioral intentions via the serial mediators counterarguing and stigmatizing attitudes were assessed in four mediation models (model 6; Hayes, 2012), but no indirect effects were detected.

Table 3. Analysis of covariance predicting attitude and behavioral intentions

	Model 4 n = 226	Model 5 n = 226	Model 6 n = 226
	Stigmatizing Attitudes	Intent to Social Distance	Intent to Support Policy
	F(df)	F(df)	F(df)
Condition	0.80(2)	28.96(2)**	1.46(2)
Illness Contact	1.24(1)	0.43(1)	0.01(1)
Previous Illness Diagnosis	9.84(1)**	1.56(1)	4.11(1)*
T1 Stigmatizing Attitudes	84.11(1)**	0.30(1)	0.39(1)
Identification	7.41(1)**	3.99(1)*	0.01(1)
Transportation	0.07(1)	.28(1)	8.08(1)**
Counterarguing	5.21(1)*	21.93(1)**	13.95(1)**
T2 Stigmatizing Attitudes		4.50(1)*	0.03(1)
Error	217	216	216
R^2	0.43	0.38	0.13

Note. ^ap < .10, *p < .05, **p < .01.

Discussion

This study examined whether narrative engagement (transportation and identification) differed among viewers watching narratives featuring characters with more stigmatized (schizophrenia) versus less stigmatized (depression) mental illness. Cognitions were found to influence attitudes and intentions to interact with individuals with a mental illness. Findings also suggest that baseline stigmatizing attitudes toward mental illness may limit one’s ability to experience successful mediated contact with characters who have schizophrenia, a more stigmatized mental illness than depression. Additionally, results suggest that counterarguing may be particularly influential on attitudes and behavioral intentions following exposure to the narrative, warranting interventions to reduce counterarguing with narratives about mental illness.

Narrative Engagement

No differences were detected across conditions on transportation, which suggests that participants were equally transported, regardless of the type of mental illness portrayed in the narrative. This finding aligns with E-ELM, which suggests that transportation could help to overcome counter-attitudinal information about a stigmatized character (Slater & Rouner, 2002). However, in the present study, transportation was negatively associated with stigmatizing attitudes at baseline, suggesting that initial stigmatizing attitudes impeded the extent to which individuals were able to be transported. Smith (2007) argues that stigma cues that label a social group in a message could lead viewers to have negative reactions to the message. In this case, perhaps those with more negative attitudes about a stigmatized group found it difficult to engage with narrative. Thus, narrative message designers must be careful to understand baseline stigmatizing attitudes, as they may have implications for narrative engagement, including the ability to engage with and be persuaded by it.

As expected, participants reported lower identification with characters with more stigmatized mental illness (schizophrenia). However, the character with depression was found to be more identifiable than the character with cancer. Although cancer was found to be less stigmatized than mental illness, the sample reported more personal familiarity with depression in terms of being previously diagnosed with the condition. This reflects current prevalence rates as 7.8% of the US population experiences a massive depressive episode each year (NAMI, 2019), while less than 1% of the US population is newly diagnosed with cancer each year (National Institutes of Health [NIH], 2020). This level of familiarity may have made the depression narrative more relevant to the viewers than the cancer narrative. However, presenting narratives about familiar mental illnesses could exacerbate negative attitudes held toward more unfamiliar mental illnesses (e.g., schizophrenia). Therefore, future research should continue to identify ways to destigmatize mental illnesses that carry more stigma, such efforts have noted the use of narratives about healing and treatment (McGinty, Goldman, Pescosolido, & Barry, 2015) or depicting mental illness along a continuum (Peter et al., 2021) can increase public understanding and support for those with mental illness.

Although theory suggests that those who identify more with characters (and are more transported) will engage in less counterarguing (Slater & Rouner, 2002), these relationships were not supported in this study. Rather, identification was directly associated with less stigmatizing attitudes and intentions to social distance, circumventing counterarguing. Theory provides some explanation for these findings. Social cognitive theory (SCT) argues that observational learning can influence behaviors. Mediated contact theorizing, which pulls from SCT, also suggests identification can have a direct effect on attitudes and behaviors when characters model positive interactions with a stigmatized group (Ortiz & Harwood, 2007; Wong, Massey, Barbarti, Bessarabova, & Banas, 2022). Indeed, meta-analysis of mediated contact theorizing showed that identification with characters in a narrative about stigmatized groups directly impacted attitudes (Banas, Bessarbova, & Massey, 2020). Developing identifiable characters therefore seems imperative to reducing stigma toward mental illness.

Counterarguing

Results showed that counterarguing, however, was positively associated with stigmatizing attitudes and intent to social distance and negatively associated with policy support. Although counterarguing was low, this finding follows narrative persuasion theorizing, which suggests that counterarguing may minimize prosocial effects (Moyer-Gusé, 2008; Slater & Rouner, 2002). Therefore, interventions that hope to use entertainment television shows to reduce stigmatizing attitudes toward mental illness should seek to minimize the potential for counterarguing. To do so, future research should seek to identify aspects of the narrative the audience may counterargue (e.g., story accuracy, character choices, persuasive message) as the identification of these factors may lead to a more persuasive message design.

Implications

Theoretical and practical implications for these findings are twofold. First, findings show that familiarity (measured here by illness contact) may play a role in narrative engagement. Personal familiarity with mental illness has been shown to increase story relevance (Caputo & Rouner, 2011). Considering a third of the sample had been previously diagnosed with the depicted illness, future research on this topic may also need to consider self-referencing. It may seem counterintuitive for a viewer to reference a sense of self when identification is often identified as a “loss of self” (Cohen, 2001), but self-referencing can be conceptualized as a psychological process that is activated to comprehend messages in a way that connects information to concepts of the self (i.e., identity) or firsthand experiences (de Graaf, 2014; Escalas, 2007). Considering messages about a stigmatized identity may impact how people with that identity cope with stigma (Meisenbach, 2010), future research should explore this relationship and effects of self-referencing on narrative engagement and outcomes regarding self-stigma.

Limitations

Although this study used ecologically valid stimuli to understand the impact of narratives on stigmatizing attitudes toward mental illness, there are some limitations. First, three single (partial) episodes of Grey’s Anatomy were compared. Although the general narrative stories were held constant, several other content-based variables may also impact these findings, such as the perceived valence of the interactions toward the character (Park, 2012), character typicality (Joyce, Harwood, & Springer, 2020), and parasocial interaction (Schiappa, Gregg, & Hewes, 2005) with the character with the stigmatized illness. Future research may benefit from finding multiple examples of each disease portrayed in other medical dramas and conducting multi-level analysis to control the differences between narratives (Slater, Peter, & Valkenberg, 2015).

In addition, we removed several items from previously-validated scales following CFA. While this measurement issue may be an indication of low attention by the participants, only three participants were dropped for failing the attention check questions, thus it suggests that scholars may need to further refine validated scales related to narrative engagement as they often theoretically and empirically overlap (e.g., identification and transportation; Tal-Or & Cohen, 2016) or are theoretically unclear (Counterarguing; Moyer-Gusé & Nabi, 2010). Finally, this study used cross-sectional data from a convenient sample of predominantly young and female college students during a time of unique national circumstances. Considering a third of our sample had been previously diagnosed with depression or anxiety, these demographic characteristics may impact how these narratives were processed and the generalizability of these results.

Conclusion

Mediated contact with stigmatized illnesses can reduce stigmatizing attitudes. This study advances our understanding of how this process may occur by investigating the extent to which narrative engagement differs for narratives with characters with mental illnesses that carry varying levels of stigma. We found that characters with more familiar and less stigmatized mental illness (i.e., depression) promote narrative engagement and more prosocial behavioral intentions toward individuals with mental illness. With theoretical and practical implications for narrative persuasion theories, researchers and writers of entertainment media should work together to promote inclusive prosocial attitudes and behaviors.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.