The impacts of character identification on attainment of HIV prevention goals: an entertainment education programme in Botswana.
One central strategy in Makgabenang's modeling is the use of positive, negative, and transitional character development in the programme's storylines (Pappas -DeLuca, Kraft, Galavotti, et al, and 2008). Narratives include positive role models who consistently make responsible health decisions and negative characters that consistently engage in risky and irresponsible behaviour. Transitional characters are also included; they may start out a storyline by making risky decisions, but evolve over time into empowered, knowledgeable, and positive characters. These dynamic transitional characters are designed to serve as the most identifiable role models for audience members (Galavotti, Pappas-DeLuca and Lansky, 2001). Moreover, Makganbenangs's producers believe that individuals struggling with decision-making may identify with transitional characters because they are also making decisions about how to live their lives. Little research, though, has attempted to see if this strategy of promoting identification with transitional characters is effective in attaining Makganbenangs's goals. The purpose of this paper, then, is to assess the effectiveness of Makganbenangs's strategy in promoting pro-social knowledge, attitudes, and behaviours among the target audience of Batswana aged between 10 and 49 years of age in the greater Gabarone region of Botswana.
Identification and Makgabenang 1s Strategy
The MARCH strategy suggests that audiences will most likely identify with the dynamic and transitional characters in E-E storylines. Character identification in MARCH is comprised of feelings of affinity, friendship, similarity and liking of media characters (Kuhlmann, Kraft, Galavotti, et al., 2008). This identification with characters is strongly linked to the way media influences behavior in audiences (Basil, 1996), consistent with the founding principles of Bandura's Social Learning Theory (2004).
Some support for this strategy exists. Lovell et al. (2007) examined Makgabaneng to see how audiences identify with characters in the drama. This study found that transitional storylines were most influential in the behaviour change process. They suggested that producers begin to use more than one transitional character for each behavioural change objective to improve audience identification with agents of behaviour change and provide several models to which the audience can relate. Following this suggestion, Kuhlmann et al. (2008) were interested in how character identification influenced Makgabaneng's objective of preventing mother-to-child-transmission. Their research found that audience members who reported having a favourite Makgabaneng character were more likely to get tested than those who did not have a favourite character. This finding reaffirms the notion that entertainment-education provides audiences a model for individual self-efficacy by showing individuals how to deal effectively with a situation (Singhal, Cody, Rogers and Sabido, 2004). The more identification audiences feel towards characters, the more influential the storyline may prove. However, this research does not illustrate how alternative character identification may influence audience actions; indeed Kuhlmann and her research team did not examine whether the favourite character was a positive, negative, or transitional role model. This study also conflated identification with liking in its measures, even though the authors correctly state that liking is only one part of identification.
In addition to liking, Jose and Brewer (1984) suggest that identification increases with greater perceived similarity between character and reader, and that the overall liking of a storyline increases with greater character identification. Based on these relationships, it makes sense that individuals who have a favourite Makgabaneng character would be more influenced by the Makgabaneng storyline. However, it also suggests that this attachment may not always lead to positive behaviour change. A listener may identify with a negative role model or transitional character that makes bad choices or with a positive role model or transitional character that makes positive choices. Indeed, one could predict that individuals living more positive lives would more strongly identify with positive characters, because that is the character that holds the most similarity to their everyday life, but also that individuals living more negative lives may in turn feel strongly attached towards negative characters, because they recognise elements of their everyday life in those negative storylines. If audiences model the behaviour of the characters with whom they most identify (Jose and Brewer, 1984), and if Slater and Rounder (2002) are correct that the way audiences respond to characters in a narrative will enhance consequential persuasive effects, it is possible that EE efforts may sometimes result in more negative behaviour change.
It is clear that the audience identification process proves much more complex than the automatic identification with transitional characters. Singhal and Rogers (2003) explain that, in EE efforts, audiences selectively expose themselves to a media text, selectively perceive the message, selectively recall later on and finally selectively utilise the message for their own purposes. However, formative research shows that the MARCH strategy holds promise as a vehicle for behaviour change (Pappas-DeLuca et al., 2008; and Galavotti, Pappas-Deluca and Lansky, 2001). Currently, Makgabaneng storylines are focused on five public health objectives based on the recommendations of their funders, PEPFAR. The five themes that the organisation has chosen to focus on include abstinence, alcohol reduction, partner reduction and fidelity, prevention with positives, and intergenerational sex. Makgabaneng has been successful in reaching and sustaining listenership among hard-to-reach younger demographics, aged 15-24; listeners of Makgabaneng hold higher levels of HIV knowledge and less stigmatising attitudes towards those affected by HIV; and, listeners report greater HIV-preventive behaviours than non-listeners (Republic of Botswana, 2005). Based on these findings, and on Kuhlmann and colleagues' exploratory work regarding liking, we hypothesise that:
H1: People who have a character in Makgabenang with whom they identify most will hold knowledge and beliefs and enact behaviours that are more consistent with PEPFAR goals than will people who do not have a character with whom they identify most.
However, one must question what happens to the outliers of the identification process who do not identify most with the transitional characters? What if they identify with a negative role model; will their behaviours be worse? Based on existing literature, it is hypothesised that different role models (positive, negative and transitional) should have different effects on audience members' knowledge, beliefs and actions. Specifically, this study hypothesises that:
H2: Makgabaneng listeners who most strongly identify with a positive role model in the storyline will hold knowledge and beliefs and enact behaviours most consistent with the PEPFAR goals; Makgabaneng listeners who most strongly identify with a negative role model in the storyline will hold knowledge and beliefs and enact behaviours least consistent with the PEPFAR goals; and, Makgabaneng listeners who most strongly identify with a transitional role model in the storyline will fall somewhere in between.
Survey research is an appropriate methodology to use when a researcher is interested in descriptive, explanatory and exploratory data (Babbi, 2007). Survey research is also a useful approach in evaluating whether exposure to EE leads to pro-social change. For example, Vaughan, Regis and Catherine (2000) surveyed 1,238 citizens living in St. Lucia. They found that 35 percent of St. Lucians had listened to the radio drama Apwe Plezi and that listeners demonstrated knew more about contraception options and held stronger views on the importance of fidelity than non-listeners. Similarly, Orme, Salmon and Mages (2006) surveyed 121 young people aged 13-16 years old to determine the success of Project Jump. They found that Project Jump was beneficial in dealing with problems relating to school exclusion, poor educational achievement, risk of criminal activity and family violence. Based on these EE survey analyses, we developed a similar set of measures and procedures.
Residents of Gabarone, Botswana and neighbouring villages participated in this study (N = 227). The participants were 80 males (35.2%) and 131 females (58.8%); sixteen (7.0%) chose not to disclose their sex. Due to Tswana cultural norms, most participants refused to disclose their age, level of household income, or size of family.
Participants were recruited through intercept methods. Participants were verbally consented, as Tswana cultural norms made signed consent impracticable. Each participant, after providing consent, was asked to complete a paper-and-pencil survey measuring the variables of interest. To examine the dependent variables, a 10-item set of questions related to PEPFAR goals and HIV/AIDS knowledge was created to measure the general outcomes toward target knowledge, attitudes and actions emphasised by the producers of Makgabenang (see Table 1). Participants responded to statements such as "I waited to have sex until I was married" to measure behavioural outcomes, "HIV can be transmitted through blood transfusion" to measure knowledge outcomes, and "Alcohol helps people unwind after a stressful day" to measure attitudinal outcomes. Each statement was measured on a five-point Likert-type scale with 1 indicating strong disagreement and 5 indicating strong agreement. Items were reverse-coded as appropriate. Because these items represent a diverse set of outcomes, and because there is little theoretical or practical reason to expect that all outcomes are affected equally by exposure, each item was treated separately.
Participants were asked which character in Makgabenang they identified with most. To investigate the first hypothesis we conducted a series of independent samples t-tests to compare the PEPFAR outcomes between individuals who named a character with whom they identified most (n = 59) to those who did not name a character (n = 168). To test the second hypothesis, if an individual did not name a character, their responses were excluded from the next analysis. We conceptualised the three character types as points along a continuous spectrum. We then coded all those who identified with positive role models as a "1," negative role models as " -1," and transitional role models as "0," allowing us to transform nominal data into continuous data. This procedure allowed us to use ANOVA to test, for each dependent variable, whether identification with a different character type had a main effect. Because these groups were approximately of equal size with equal variances
in means, we also employed the Tukey HSD procedure to compare differences between pairs of character types. These ANOVAs used character type as the fixed factor and the PEPFAR and knowledge measures as the dependent variables.
The first hypothesis predicted that individuals who had a character with whom they identified most would attain more positive PEPFAR outcomes than individuals who did not. This hypothesis was not supported for most outcomes; it was supported only for knowledge about whether HIV could be transmitted through saliva. As Table 1 indicates, independent samples t-tests show that there was not a significant difference between the two groups. Simply identifying with a character is not associated with more positive PEPFAR outcomes.
The second hypothesis predicted that identification with positive role models would be linked to the greatest attainment of PEPFAR message goals, negative role models the least, and transitional role models in between. Identification with a positive, negative, or transitional character did not have a consistent effect on whether a listener was more or less likely to hold knowledge, attitudes, or beliefs in line with PEPFAR goals or to have acted in line with PEPFAR goals. The hypothesis was not supported. On specific PEPFAR goals, however, some interesting differences emerged (See Table 2). Individuals who identified most with a positive role model were more likely to have delayed sexual debut until after marriage, and those who identified most with a transitional character least likely to have delayed; people who identified most with negative role models did not differ from the other two groups. Individuals who identified with transitional characters were more likely to believe that alcohol helped relieve stress than the other two groups. These individuals were more likely to use birth control than individuals who identified with negative characters; they did not differ from those who identified with positive role models. In other areas related to prior behaviour, individuals who identified most with different types of characters were no more or less likely to have been unfaithful in a relationship or to have engaged in intergenerational sexual relations. Similarly, identification with different characters had no effect on a person's intention to use condoms or ability to recognise stigma. Effects on knowledge were mixed; identification with different types of characters did not reflect different levels of knowledge about HIV transmission through blood transfusion, but people who identified most with positive role models were more likely to know that saliva was not a transmission route for HIV than were people who identified with other character types. Finally, and perhaps most useful, is that people who identified most with transitional characters were more likely to believe that their choices affected their life course. This may help explain the choices made by these individuals in relation to abstinence, birth control, and alcohol (and, were the trend data statistically significant, in fidelity and intergenerational sex); that is, because these individuals had made life choices similar to the transitional characters they recognized consubstantiality with those characters and, in turn, as they recognised that the transitional characters' choices determined the characters' life courses, the listeners also recognised their own life choices determined their own life courses.
Discussion and Implications
This research asked whether audience identification with characters in an entertainment-education product affected prosocial outcomes on a variety of PEPFAR goals. The hypothesis that simply having a character with whom one identified would lead to prosocial outcomes was not supported. Although Kuhlmann and colleagues (2008) found that simply having a favourite character in Makgabenang was sufficient to bring about prosocial change in this EE product, we found that having a character with whom one identifies is not enough. This may be because we asked about a broader array of outcomes than Kuhlmann et al. did, and found only one difference in knowledge and no effect on the other PEPFAR outcomes. We conclude, therefore, that it is not enough to promote identification, but that one must promote identification with particular characters if one wants to effect prosocial change.
This promotion of prosocial change, however, is not as simple as getting people to identify with one kind of character or shaping that character so that individuals identify with him or her. If one goes beyond liking as a measure of identification, it is intuitive that individuals who identified with negative role models most would have worse PEPFAR outcomes than individuals who identified with positive role models, even if their absolute level of identification were the same. This intuition would be incorrect. In fact, we found that people who identify most with transitional characters--those who are neither positive nor negative role models--had some of the worst PEPFAR outcomes. People who identify with transitional characters are more likely to drink, to have had early sexual debut, and to have had intergenerational sexual relations. Their actual (mainly negative) behaviours may be what allows some listeners to identify with the transitional characters. Since these characters have also made poor decisions in the past and make model making more positive life choices because of these poor decisions, these transitional characters may be strong pathways to model individual level behaviour change. Indeed, because transitional characters are making choices that make meaningful differences, they model that one's own choices shape the outcome of a life course. And, since individuals who identify with transitional characters also believe in this internal locus of control, the modeling is likely to be particularly influential. Makgabenang's writers can capitalise on this relationship by showing the transitional role model who make prosocial choices as being rewarded and those who make poor choices as being punished. At the same time as we would encourage Makgabenang to take great advantage of these transitional role models, we would discourage them from focusing on identification with positive or negative role models. Negative role models present little advantage; modeling their behaviour does not lead to better results on any of the PEPFAR outcomes as compared to these other character types. And, if individuals already identify with positive role models, it is likely because they are already living prosocial lives. In short, the best opportunity for promoting prosocial behaviour change is to encourage identification with transitional role models who make good choices are rewarded for doing so.
Bandura, A. (2004). Health Promotion by Social Cognitive Means. Health Education & Behavior, 31, 143-164.
Basil, M. (1996). Identification as a Mediator of Celebrity Effects. Journal of Broadcasting and Electronic Media, 40, 478-495.
Galavotti, C., Pappas-DeLuca, M., & Lansky, A. (2001). Modeling and Reinforcement to Combat HIV: The MARCH Approach to Behavior Change. American Journal of Public Health, 91 (10), 1602-1607.
Jose, P., & Brewer, W. (1984). Development of Story Liking: Character Identification, Suspense and Outcome Resolution. Developmental Psychology, 20 (5), 911-924.
Kuhlmann, A., Kraft, J., Galavotti, C., Creek, T., Mooki, M., & Ntumy, R. (2008). Radio role models for the prevention of mother-to-child transmission of HIV and HIV testing among pregnant women in Botswana. Health Promotion International, 23(3), 260-268.
Lovell, Pappas-DeLuca, Kuhlmann, et al, (2007). "One day I might find myself HIV-positive like her": Audience Involvement and Identification with Role Models in an Entertain ment-Education Radio Drama in Botswana. International Quarterly of Community Health Education, 28 (3), 181-200.
Pappas-DeLuca, K., Kraft, J., Galavotti, C., et al. (2008). Entertainment-education Radio Serial Drama and Outcomes Related to HIV Testing in Botswana. Global Health Sciences Literature Digest, 20 (6), 486-503.
Republic of Botswana. (2005). Botswana AIDS Impact Survey III. Gabarone, Botswana: NACA.
Singhal, A., Cody, M., Rogers, E., & Sabido, M. (2004). Entertainment-education and Social Change. Mahwah, NJ: Lawrence Erlbaum Associates.
Singhal, A., & Rogers, E. (2003). Combating AIDS: Communication Strategies in Action. Thousand Oaks, CA: SAGE.
Slater, M., & Rouner, D. (2002). Entertainment-education and Elaboration Likelihood: Understanding the Processing of Narrative Persuasion. Communication Theory, 12 (2), 173-191.
Vidmar, N., & Rokeach, M. (1974). Archie Bunker's Bigotry: A Study in Selective Perception and Exposure. Journal of Communication, 24 (1), 36-47.
L. Meghan Peirce is an Assistant Professor in the Department of Communication Studies at West Chester University of Pennsylvania. She received her PhD from Ohio University, and her research focuses on new media for social change.
Benjamin Bates is an Associate Professor in the School of Communication Studies at Ohio University. He received his PhD from the University of Georgia and has published more than 50 competitively-selected articles.
Table 1 T-Test Results of PEPFAR Outcomes and Identification Identification Present Absent t df Abstinence: I waited to have sex 3.13 2.81 1.21 163 until I was married. Alcohol reduction: Alcohol helps 3.85 3.65 0.84 163 people unwind after a stressful day (R). Birth control: I use birth control 3.60 3.12 1.83 163 every time I have sex. Empowerment: My life is determined 4.21 4.12 0.53 163 by my own actions. Fidelity: I have never cheated on my 3.75 3.49 1.18 163 partner. Intergenerational sex: I have been 3.62 3.71 -0.38 163 involved in a relationship with someone seven years older or younger than me (R). Knowledge: HIV can be transmitted 4.52 4.48 0.26 163 through blood transfusion. Knowledge: HIV can be transmitted 3.75 3.13 3.01 * 163 through saliva (R). Prevention: I will use a condom 3.42 3.27 0.78 163 every time I have sex in the next 6 months. Stigma: Most people will isolate 2.56 2.63 -.036 163 someone if they know they are infected with HIV (R). NOTE: Higher means indicate greater attainment of PEPFAR goal; (R) indicates item was reverse coded; * = p < .05. Table 2 Analysis of Variance in PEPFAR Outcomes by Character Type Participant Identifies Most With Negative Transitional Positive N in condition. 19 11 29 Abstinence: I waited to have sex 3.00ab 1.82a 3.55b until I was married. Alcohol reduction: Alcohol helps 3.89a 2.36b 3.78a people unwind after a stressful day (R). Birth control: I use birth 2.88a 4.27b 3.35ab control every time I have sex. Empowerment: My life is 3.50a 4.55b 4.07ab determined by my own actions. Fidelity: I have never cheated 3.95 3.09 3.68 on my partner. Intergenerational sex: I have 3.37 2.82 3.71 been involved in a relationship with someone seven years older or younger than me (R). Knowledge: HIV can be 4.63 4.55 3.86 transmitted through blood transfusion. Knowledge: HIV can be 3.16a 3.00a 4.15b transmitted through saliva (R). Prevention: I will use a condom 3.37 3.27 2.79 every time I have sex in the next 6 months. Stigma: Most people will isolate 2.95 2.55 2.79 Participant Identifies Most With F (d.f.) sig. [H.sup.2] N in condition. Abstinence: I waited to have sex 5.97 (2,56) <.005 .18 until I was married. Alcohol reduction: Alcohol helps 6.23 (2,54) <.005 .19 people unwind after a stressful day (R). Birth control: I use birth 3.133 (2,51) =.05 .11 control every time I have sex. Empowerment: My life is 3.12 (2,51) <.05 .11 determined by my own actions. Fidelity: I have never cheated 1.87 (2,52) n.s. .07 on my partner. Intergenerational sex: I have 2,55 (1.51) n.s. .05 been involved in a relationship with someone seven years older or younger than me (R). Knowledge: HIV can be 2,56 (2.95) n.s. .10 transmitted through blood transfusion. Knowledge: HIV can be 2,54 (6.88) <.005 .20 transmitted through saliva (R). Prevention: I will use a condom 2,56 (1.58) n.s. .05 every time I have sex in the next 6 months. Stigma: Most people will isolate 2,55 (0.33) n.s. .01 someone if they know they are infected with HIV (R). NOTE: Higher means indicate greater attainment of PEPFAR goal; (R) indicates item was reverse coded; means with different subscripts indicates significance at the p < 0.05 level