The effects of homophily moderated by message framing on attitude and
self-efficacy towards smoking cessation in a digital interactive intervention
Xiangjun Ji
Master’s Thesis
Graduate School of Communication Master’s Program: Communication Science
Persuasive Communication
Supervisor: Bas van den Putte Word Count: 7907 Date: 30 January 2020
Abstract
Objective: To explore the persuasion effects of homophily moderated by message framing
theory towards smoking cessation in a digital interactive intervention. Design: A survey about smoking cessation among 153 Chinese smokers was conducted after participating in an interactive digital game. Main outcome measure: Attitude, self-efficacy, behavioral beliefs and control beliefs towards smoking cessation. The pre-measure about all variables were also tested. Results: Homophily has a persuasion effect on attitude, but does not influence the self-efficacy, behavioral beliefs and control beliefs. No significant interactive effects of homophily and message framing are found. No indirect effects are confirmed from the homophily towards attitude and self-efficacy via behavioral beliefs and control beliefs.
Conclusion: The high perceived similarity in the entertainment-education intervention has
more significant effects compared to the low perceived similarity on attitude towards smoking cessation.
Keywords: smoking cessation, perceived similarity, message framing, health behavior
Introduction
Digital games, also called video games, refers to an interactive and narrative digital text (Bosman, 2016). The adventure game, such as an interactive fiction game, is one type of digital games, where players are designed to solve problems in a narrative structure game storyline by collecting several hints (Dickey, 2006). Due to its problem-solving gene, adventure games were used for educational purposes in the early times. For instance, Quinn (1994) proved that playing narrative games is compelling to increase learning abilities. Digital games are frequently applied to educate and change behaviors in health interventions
when the electronic devices are prevalent in daily use and the broad coverage of the Internet technologies (Ritterfeld, Cody, & Vorderer, 2009). Previous studies, such as Portnoy, Scott-Sheldon, Johnson and et al. (2008) and Krebs, Prochaska & Rossi (2010), evidence that digital games are sufficient to overcome motivational issues on adopting healthy behaviors. From several years ago, a lot of entertainment-education interventions started to be designed by digital games, which has been applied into targeting smoking cessation, condom use, and other social issues (Singhal & Rogers, 2001). The digital game intervention can be seen as a practice of the entertainment-education programs by incorporating entertainment factors to persuade people covertly in adopting a healthy lifestyle (Kaiser Family Foundation, 2004). As following with Moyer-Guse (2008), the education-entertainment in my study is referred to “prosocial messages that are embedded into popular entertainment media content”
(Moyer-Guse, 2008).
Scholars believe that entertainment-education offers a more powerful way on persuading attitude and behavior change than traditional persuasive messages because the entertaining content produces more enjoyment and arises less resistance to the persuasive messages (Murphy, Frank, Chatterjee, & Garbanati, 2013; Slater & Rouner, 2002). According to the entertainment overcoming resistance model proposed by Moyer-Guse (2008), narrative structure, transportation, identification, enjoyment, parasocial interaction, liking and perceived similarity (also called homophily) are the leading entertainment features. The narrative feature makes viewers follow the events as storytelling that is named involvement. Generally, involvement in the entertainment-education includes engagement, immersion and transportation (Slater & Rouner, 2002). By engaging with storylines and characters, people are less in criticizing but more in engaging with the story, as explained by the extended elaboration likelihood model (E-ELM ) (Shrum, 2004).
However, the inconsistent research results often happen to the theoretical development in empirical research. As the theory growing, not all of the features in the entertainment-education have a persuasion effect, such as homophily. For instance, a study conducted amongst 82 American college students by Lee and Bichard (2006) reported that no persuasion effects on the actual behavior change were found towards drinking in a narrative storytelling intervention when participants perceived similarity (gender) with the protagonist. According to a review of exploring the characteristics of narrative interventions, carried out by de Graaf, Sanders, & Hoeken (2016), only two studies out of eleven announced a marginally significant persuasion effect on intention when the perceived similarity was manipulated, and none study reported a persuasion effect from homophily on attitude and self-efficacy. However, a limitation of the systematic review conducted by de Graaf, Sanders, & Hoeken is that only the non-interactive interventions were included in analysis. Thus, it is still vague that the persuasion effect of homophily on attitude and self-efficacy in the interactive intervention. As attitude and self-efficacy are important predictors for actual behavior change (Fishbein & Cappella, 2006), it is worthwhile to measure the attitude and self-efficacy then predict an actual behavior change. An online interactive intervention proves that the interacting activity can increase viewers’ engagement so that improves the persuasion
effect in performing a desired behavior (Meischke, Lozano, Zhou, and et al., 2011). Regarding the E-ELM and the entertainment overcoming resistance model (Moyer-Guse, 2008), the enhanced engagement increases the homophily then leads to a more powerful effect on attitude, beliefs, self-efficacy, and eventually for the actual behavior change. Thus, it is plausible that the perceived similarity in an interactive intervention is effective for behavior determinants and behavior change, compared with the non-interactive interventions. Therefore, to fill the gap of the previous study, it is essential to explore the persuasion effect of perceived similarity in an interactive intervention. This study will adopt an interactive
digital game intervention to investigate the persuasion effects of perceived similarity on attitude and self-efficacy towards an epidemic unhealthy behavior, smoking. Tobacco has been recognized as a threat to health by resulting chronic diseases (i.e. bad breath or sore throat) and fatal diseases (i.e. lung cancer or oral cancer) (Wald & Hackshaw, 1996). The result may provide theoretical proofs to design an interacting smoking cessation intervention in the future to help more smokers get rid of tobacco addiction in an interesting but painless way.
Smoking cessation is seen as a prevention behavior commonly. Prevention behaviors (i.e. smoking cessation) are functional to prevent the potential illness and maintain one’s health status, whereas the detection behaviors (i.e. screening) emphasize taking risks to find a potential threat to one’s health. Gain-framing will be more persuasive than the loss-framing
in promoting prevention behavior interventions based on the prospect theory (Rothman, Bartels, Wlaschin, & Salovey, 2006). However, in the smoking cessation interventions, the framing is not always more persuasive than the loss-framing. Sometimes the gain-framing works better than the loss-gain-framing in the smoking cessation intervention when smokers perceived less risks (Toll, Salovey, O’Malley, and et al., 2006), whereas the loss-framing produces more persuasive effects than gain loss-framing when people recognized a higher level of risks about smoking (Moorman & van den Putte, 2008). Hence, it is worth to investigate the unclear effect of message framing on smoking cessation. Besides, in line with the entertainment-education, the gain-framing stresses the potential gain produced by a behavior, which is a pleasant feeling for viewers and increase the engagement. Thus, a moderated influence of message framing on the persuasion effect of perceived similarity on attitude and self-efficacy can be expected towards smoking cessation.
Combined with the integrated model, proposed by Fishbein and Cappella (2006) and the entertainment overcoming resistance model, two predictable factors for attitude and
self-efficacy are also included in the model as mediators. They are behavioral beliefs and control beliefs.
In all, the research questions are in the following:
RQ1: Does the perceived similarity of characters have significant effects on self-efficacy or attitude?
RQ2: If so, are such the effects mediated by beliefs and moderated by message framing strategy?
Theoretical Framework
The behavior determinants change and the Entertainment-education
According to the integrated model proposed by Fishbein and Cappella (2006), attitude and self-efficacy are significant determinants to influence the probability of taking a desired behavior. Thus, it is important to affect attitude and self-efficacy in health interventions in order to influence people to perform a health behavior. Behavioral beliefs and control beliefs are two predictors of attitude and self-efficacy based on the integrated model (Fishbein & Cappella, 2006). Behavioral beliefs include two kinds, one is the instrumental beliefs that describes the costs of performing the behavior, the other one is affective beliefs indicating the feelings derived from the behavior (Ajzen & Driver, 1991). Control belief refers to the perceived difficulty of performing the desired behavior. As mentioned by Ajzen, the perceived difficulty comes from various sources, such as personal or friends’ experiences, or the second piece of information (Ajzen & Driver, 1991). Hence, if people realize a positive feeling can be derived by performing a behavior (i.e. the benefits brought by the behavior) and perceive fewer obstacles in acting the behavior, the beliefs towards attitude and self-efficacy will be influenced and attitude as well self-self-efficacy will change as the result. People are expected to perform health behavior eventually. Therefore, it is essential for the
integrated model working with the other intervention theories to influence beliefs towards attitude and self-efficacy. The entertainment-education (E-E) is a choice because it has the function to influence people’s beliefs.
Entertainment-education has been proven to be effective in health interventions by overcoming resistance to persuasion. Generally, entertainment-education refers to an educational message embedded into the entertainment content aiming to bring about social and behavioral change, transmitted by the mass media (Singhal & Rogers, 2002). In most cases, entertainment-education content influences viewers’ perception of the real world. The research found that people incorporate the information from the entertainment-education media content with their own understanding then form a new perception of a given behavior (Brodie, Foehr, Rideout, and et al., 2001). There is another theory approach, the social cognitive theory, to demonstrate the influence of the entertainment-education on behavior determinants proposed by Bandura. The social cognitive theory indicates that one is able to learn by imitating and modeling others’ behaviors when he/she is motivated (Bandura, 2009).
If the desired behavior is successfully performed by a model in a story and brings about benefits, audiences will increase their confidence and self-efficacy on acting in the same way and change the attitude by learning about the benefits. Thus, the entertainment-education serves a role in influencing viewers’ attitude and self-efficacy and presumably affect the
actual behavior. According to the entertainment overcoming resistance model proposed by Moyer-Guse (2008), entertainment-education includes two features that significantly influence attitude and self-efficacy, one is the narrative structure, the other one is the protagonists. The two features will be discussed in detail as follows.
The effects of perceived similarity on attitude, self-efficacy, and beliefs
The narrative is that a protagonist goes through a series of events, which are fictional or real (Moyer-Guse, 2008). Normally, a narrative has a story result after the experience and
makes audiences feel entertained. Thus, the narrative structure and story results compose a completed storyline and make sure the engagement because of the entertaining story.
With viewers engaging in the entertainment story, two involvements occur: one is the narrative involvement, another one is the characters’ involvement. Involvement here refers to
audiences are followed with story events written in the entertainment media content (Moyer-Guse, 2008). Once audiences involved with the story, they mainly focus on the narrative storylines and characters, concentrate less on the implicit persuasion messages that viewers believe them as the risk to threaten their freedom on individually thinking. Thus, viewers produce less counterarguing towards the persuasion messages so that beliefs and attitudes will be implicitly influenced. The second involvement is made by identification, wishful identification, similarity, parasocial interaction and liking. Though these terms are similar and sometimes are mixed together, they tend to be effective in interventions supported by different theory constructions. With the indirect learning process proposed by Bandura in the social cognitive theory, one of the antecedences to model a behavior is that people should be motivated to act in the same manner (Moyer-Guse, 2008). The motivation here can be seen as the positive outcome expectancies and self-efficacy to perform the given behavior. The homophily with character is a powerful factor to motivate people to perform a health behavior. People are easier to perceive positive effects and strengthen confidence as well as improve self-efficacy by provided with a useful solution for a problem if they observed a model who is similar to themselves successfully act in this manner (Murphy, Frank, Chatterjee, & Garbanati, 2013). The perceived similarity also increases the engagement to the storyline because people pay more attention to the information that is relevant to them and thus produce less counterarguing (Moyer-Guse, 2008). The engagement can be enlarged by an interactive intervention because it requires more interactive activities from audiences with the intervention as proposed by Webster, Michie, Estcourt, and et al. (2015). Besides, the
advantages of interactive digital interventions are increasing engagement by personalizing and tailoring materials and advice individually (Webster, Michie, Estcourt, and et al., 2015). Tough as mentioned forehead, part of earlier studies denied the effect of perceived similarity in the non-interactive interventions (de Graaf, Sanders, & Hoeken 2016; Lee & Bichard, 2006). With the high-attractive nature of the interactive web-delivered intervention, the interactive intervention is expecting to significantly increase participants’ engagement towards the program, compared with the non-interactive digital interventions (Vandelanotte, Spathonis, Ekin, and et al., 2007). Therefore, the perceived similarity in an interactive intervention is likely to have stronger persuasion effects on attitude and self-efficacy.
By working with the integrated model and the entertainment-education to construct the theoretical framework of this study, the perceived similarity is picked as the independent variable, attitude and self-efficacy are the dependent variables. Due to the homophily with the character that can decrease the obstacles and increase positive feelings on performing a health behavior, the behavioral beliefs and control beliefs are included as mediators for attitude and self-efficacy.
The hypothesis of the effects of homophily on attitude and self-efficacy are as following:
H1a: The high perceived similarity (vs. low perceived similarity) has more influences on
attitude and behavioral beliefs towards smoking cessation.
H1b: The high perceived similarity (vs. low perceived similarity) has more influences on
self-efficacy and control beliefs towards smoking cessation.
H2a: The effect of high perceived similarity of characters (vs. low) on attitude is mediated by
behavioral beliefs.
H2b: The effect of perceived similarity of characters on self-efficacy is mediated by control
The effects of the message framing in the E-E program on attitude, self-efficacy, and beliefs
Message framing theory is originated from the prospect theory, proposed by Tversky and Kahneman in 1981. People normally respond differently to prevention behavior and detection behavior. Individuals are more likely to avoid risks when considering a potential gain in order to keep the gain but prefer to take risks when considering a potential loss hoping to avoid the loss (Rothman, Bartels, Wlaschin, & Salovey, 2006). Thus, in the health communication domain, people are more responsive to preventive behaviors (i.e. smoking cessation) that stress a potential benefit (gain-framing), whereas are reactive for detective behaviors (such as the cancer screening) that emphasize a potential loss (loss-framing). Gallagher and Updegraff (2011) confirmed the main effect of the gain-framed message on attitude and intention towards prevention behavior by conducting a meta-analysis of empirical studies towards the message framing theory. Hence, there are some expected effects from message framing towards smoking cessation on attitude, behavioral beliefs, self-efficacy and control beliefs.
With the explicitly positive nature of the gain-framing messages, an assumption about the interactive effects composed by perceived similarity and message framing towards smoking cessation is formed, that is the high perceived similarity with a gain-framing message has more persuasion effects on attitude, self-efficacy, behavioral beliefs and control beliefs than the low perceived similarity with a gain-framing message. And for the low perceived similarity conditions, the loss-framing messages are more persuasive than the gain-framing messages. The gain-gain-framing messages stress a potential gain of a healthy behavior, which is more enjoyable for receivers than the loss-framing describing a potential loss (Brusse, Fransen, & Smit, 2017). In line with the feature of entertainment-education, enjoyment can decrease the counterarguing of recipients and also increase engagement and
perceived similarity eventually (Moyer-Guse, 2008; Murphy, Frank, Chatterjee, & Garbanati, 2013). Thus, a gain-framing message is expected to add more enjoyment than the loss-framing message in the E-E program so that has a stronger persuasion effect. By contrast, in the low perceived similarity condition, audiences perceive less similarity with the protagonist, therefore, they raise less fear to the potential loss and thus increase engagement even though the main character is failed to quit smoking in the story. Gain-framing message should be useless in the low perceived similarity condition due to people may not realize the benefits of stop smoking from a virtual character that is not similar to themselves and refuse to quit smoking. Thus, a loss-framing message should promise a more persuasive effect in the low perceived similarity condition than the high perceived similarity condition. As self-efficacy is an important factor of actual stop smoking behavior (Stecher, DeVellis, Becker, & Rosenstock, 1986), control beliefs and self-efficacy will be influenced by the interactive effects of homophily and message framing. Although attitude is not as significant as the self-efficacy towards smoking cessation behavior, there are some studies that confirms the main effects of perceived similarity and message framing to attitude change (Gallagher & Updegraff, 2012; Moyer-Guse, 2008) but no interactive effect in smoking cessation has been explored yet. It is also interesting to investigate the interactive effects on attitude towards smoking cessation. Thus, the interactive effects on attitude and behavioral beliefs will also be measured in this study.
Therefore, the hypothesis about the interactive effects are as following:
H3a: The effect of perceived character similarity on attitude is moderated by framing, that is,
in a gain frame intervention high perceived similarity has more effect than low perceived similarity, whereas in a loss frame intervention high perceived similarity has less effect than low perceived similarity.
H3b:The effect of perceived character similarity on self-efficacy is moderated by framing,
that is a gain framing intervention high perceived similarity has more effect than low perceived similarity, whereas in a loss frame intervention high perceived similarity has less effect than low perceived similarity.
H4a: The effect of perceived character similarity on behavioral beliefs is moderated by
framing, that is a gain framing intervention high perceived similarity has more effect than low perceived similarity, whereas in a loss frame intervention high perceived similarity has less effect than low perceived similarity.
H4b: The effect of perceived character similarity on control beliefs is moderated by framing,
that is a gain framing intervention high perceived similarity has more effect than low perceived similarity, whereas in a loss frame intervention high perceived similarity has less effect than low perceived similarity.
In sum, all the hypothesis are summarized in statistical diagram 1:
Statistical diagram 1 The conceptual model.
Methods Design
To test the hypotheses, a 2 ( perceived similarity: high vs. low) × 2 (message framing: gain framing on personalized advice vs. loss framing on personalized advice) between-subjects randomized factorial design has utilized attitudes and self-efficacy as dependent variables, behavioral beliefs and control beliefs as mediators.
Table 1.
The experimental design
Framing
Gain framing Loss framing
high perceived similarity of characters
Group1: high perceived similarity of characters ×gain
framing
Group2: high perceived similarity of characters ×
loss framing
low perceived similarity of characters
Group3: low perceived similarity of characters ×gain
framing
Group4: low perceived similarity of characters
×loss framing
Perceived similarity of characters (high vs. low) in an interactive game were the independent variables, and message framing (gain vs. loss) of health advice from doctors were served as moderators. In order to qualify as a digital interactive intervention, an interactive fiction game was specially created. Interactive fiction games are referred as a genre of adventure games that players explore some locations or achieve goals controlled by text commands (Sharma, Ontanón, Mehta, & Ram, 2007). PowerPoints was used to create the interactive fiction game with images and text commands, which had four characters personally designed by participants’ demographic factors. Gender is reported that has
significant effects on the perceived similarity of characters between fictional characters and audiences, both in male or female groups (Austin, Roberts, & Nass, 1990). Appiah in 2001 also found that people report a high level of perceived similarity with the character if they were at the same age (as cited in Hoffner & Buchanan, 2005). Therefore, in the current study, age and gender were used to manipulate the perceived similarity of characters.
Perceived similarity of characters
In the high perceived similarity conditions, participants created the main character before the game started based on their demographic factors, gender and age. Because the experiment was designed to spread via weblink over social media, WeChat and Sina Weibo, which are two popular social media platforms in China to ensure the coverage of sampling. According to the official report of user images by Sina Weibo and WeChat in 2018, most of the users were between 23 to 30 years in 2018. Thus, there were three age groups, based on the users’ age range, they were 18-25, 26-30 and above 30. The personalized characters were
clustered into four groups (See Appendix A): Chinese male college student Xiaoming (18-25 year group), Chinese male adult Ming (26-30 and above 30-year groups), Chinese female college student Xiaolan (18-25 year group), and Chinese female adult Lan (26-30 and above 30-year groups). According to the Report of the Chinese Adult Smoking Behavior Survey,
regulated by the Chinese Center for Disease Control and Prevention in 2015, 52.1% of males in China were smokers, while only 2.7% of females were a smoker in 2015 (CCDC, 2016). Adult here means that people who smoke in China are above 15 years old. Each of the two groups in high perceived similarity included the personalized characters in males and females. Due to the most of smokers in China are male, in order to have a significant effect on manipulation, in the two low perceived similarity conditions (gain-framing and loss-framing), participants employed the same Chinese female college student character Xiaolan regardless of their gender and age. Participants were automatically assigned to the four experimental conditions by Qualtrics.
In all, participants who were assigned to the high perceived similarity groups can create the main character based on their gender and age, whereas participants who were assigned to the low perceived similarity groups employed the same female character.
All participants in each condition experienced the same storylines. The story described an experience about quitting smoking and had two different endings, one successfully stopped smoking (gain-framing), and another one failed to stop smoking (loss-framing). The storyline started with the main character who had a regular smoking habit felt the difficulty in breathing and then decided to quit smoking when he/she received the physical examination report, which indicated that the uncomfortable feeling was caused by smoking. Then he/she received health advice from the doctor on smoking quitting, that part was created by different framing strategies (gain. vs. loss) and colored in red to stress.
Message framing strategy
According to the research conducted by Delucchi, Tajima, and Guydish (2009), the barrier that smokers frequently occur when they intend to quit smoking is a ‘lack of training’
on withdrawal symptoms. Practitioners and smokers reported that experience the negative effects of withdrawal symptoms on attitude and self-efficacy towards smoking cessation
(Delucchi, Tajima, and Guydish, 2009). Smokers who try to get rid of tobacco have a relapse because of the torture of withdrawal symptoms, such as anxiety, depression and insomnia. Therefore, the given health advice was explicitly designed for how to cope with withdrawal symptoms aiming to positively influence smokers’ attitude and self-efficacy towards smoking
cessation. In this study, the withdrawal symptoms were described as negative feelings (anxiety or depression) and insomnia (Hughes & Hatsukami, 1986). The advice was created in the gain-framing (i.e. “you can cope with nicotine withdrawal symptoms and will not have insomnia or anxiety if you continue with the nicotine replacement therapy”), and in the loss-framing (i.e. “you cannot handle with nicotine withdrawal symptoms and will be troubled by insomnia or anxiety if you fail to do the nicotine replacement therapy ”). Then the character found another barrier when he/she followed/or not followed the doctor’s advice on coping
with withdrawal symptoms. Then participants can choose the character turned help from an expert or a friend during the quitting. The given solution about the barrier was described in gain framing (“You can restrain cigarette cravings if you keep a quit journal regularly” ) and in loss framing (“You cannot restrain cigarette cravings if you fail to keep a quit journal regularly”), too. No matter the high or low perceived similarity groups, participants in the gain framing condition experienced the positive game result that successfully stopped smoking, whereas in the loss framing had a negative result that failed to quit smoking. Due to the participants were all from China, the game and survey questionnaire were designed in Chinese.
Manipulation check
Participants were asked to indicate the level that they perceived similarity (PSC) between the character and themselves by the question: “I think the character is similar to me
in terms of the outlook, age, and gender,” and measured from“1 = strongly disagree” to “5 = strongly agree” (Ensher & Murphy, 1997) (α = .95, M = 3.20, SD = .95).
Control variables
To ensure that other differences did not cause the effects of perceived similarity of characters during the experiment, nicotine dependence and intention to quit smoking of participants were assessed.
Smoking behavior was assessed by the Heaviness of Smoking Index (HSI). HSI has two parts, one is cigarettes smoked per day (CPD), another part is time to first smoke (TTFC) (Borland, Yong, Balmford, and et al., 2010). CPD was coded by 1: 0–10 cigarettes/day, 2: 11–20 cigarettes/day, 3: 21–30, and 4: 31+ cigarettes/day. TTFC was coded by 1: 61 min or more, 2: 31–60 min, 3: 6–30 min, and 4: 5 min or less. The HSI was computed by CPD and TTFC then recoded into three categories of dependence: 1 = low: 2–3, 2 = moderate: 4–5, and 3 = high: 6–8. Intention to quit smoking was assessed by a question: “Have you ever intended to quit smoking?”, 1 = “yes”, 2 = “no”.
Measures
Dependent variables
Attitudes. Participants were asked three questions about attitude on quitting smoke
(i.e., “If you quit smoking within the next six months, this would be…”) and measured by three continua “1= unwise” to “5 = wise”, “1 = unpleasant” to “5= pleasant”, and “1 =
negative” to“5= positive” (Hummel, Candle, Nagelhout, and et al., 2018) (α = .78, M = 4.18, SD = 3.71).
Self-efficacy. Participants were requested to indicate their reaction about giving up
smoking in the next six months on the three 5-point scales. The statements are as follows: “Do you feel confident about quitting smoking in the next six months?” (1= Unconfident, 5 =
confident); “How possible you can be sure on successfully stop smoking” (1= impossible, 5 = possible); “For me, to quit smoking in the next six months is” (1= extremely difficult, 5 = extremely easy) (Maurer & Pierce, 1998) (α = .81, M = 3.97, SD = 4.37).
Mediator
Behavioral beliefs. To measure the behavioral beliefs, participants were asked a few
questions about smoking: “If I quit smoking, this will prevent that I will get oral or
pulmonary diseases; If I quit smoking that will keep me from bad breath and difficulty in breathing; If I quit smoking, I will experience withdrawal symptoms (e.g., insomnia or anxiety)” and measured by a 5-scale semantic differential scales from 1 = Extremely unlikely to 5 = Extremely likely (Ajzen, 2006) (α = .78, M = 3.94, SD = 3.92).
Control beliefs. To measure the control beliefs, participants were asked to indicate
their reactions towards situations that people might be tempted to smoke on the 5-scale semantic differential scales. The statements are as follows: “The following are some situations in which certain people might be tempted to smoke. Please indicate whether you are sure you could refrain from smoking in each situation: when I feel very anxious, when I am angry, and when I have the urge to smoke” (1 = not very sure, 5 = absolutely sure) (Etter, Bergman, Humair, & Perneger, 2000). “I feel like I have control over my feelings of withdrawal from cigarettes”; “I believe that I am capable of dealing adequately with withdrawal symptoms from smoking”; “I am able to manage the feelings of withdrawal from smoking that I experience”) (1= absolutely disagree, 5 = absolutely agree) (Schnoll, Martinez, Tatum, and et al., 2011) (α = .89, M = 3.74, SD = 8.26).
All dependent variables and mediators were divided the sum by the number of items. So the scale of attitude, self-efficacy, behavioral beliefs were from 0 to 15, and for the control beliefs were from 0 to 28, in which 0 means negative.
Participants
All of the 153 adults (age>18) participants (in both gender) came from China. 39 participants are between 18-25 years old, 69 people are 26-30, and 45 participants are above
30. The high-gain condition had 40 people, and the high-loss had 34 people, the low-gain had 38 people, and the low-gain group had 38 people.
People who did not smoke daily were excluded from the experiment. All of participants have a daily smoking habit thus no one was excluded from the experiment that 137 participants were male and 16 were female. For the intention to give up smoking, 150 people intended to stop smoking, while only 3 people did not plan to. Besides, the majority of participants have low to moderate levels of HSI, 57 people adopt with the low HSI, and 89 people have moderate HSI. Only 7 participants are heavily addicted to tobaccos.
Procedures
Participants attended a pretest before the experiment and a posttest after the experiment to evaluate the change on attitude, self-efficacy, behavioral beliefs and control beliefs towards smoking cessation. Gender, age, intention to quit smoking and heaviness to tobaccos were also collected before the pretest. The whole experiment consists of four parts: the first part is to collect demographic factors, intention and heaviness to smoking; the second part is the pretest; the third part is the game, and the last part is the posttest. The suggested completion duration is 15 minutes. The average completion time in the real experiment is 16.8 minutes.
Results Randomization check & Background variables check
To test the randomization, a one-way ANOVA was conducted. The experimental conditions did not differ with respect to age, F (3,149) = .39, p = .75; gender, F (3,149) = 1.41, p = .24; HSI, F (3,149) = .89, p = .44; intention, F (3,149) = 1.22, p = .30. The randomization in the current study was successful.
However, as the Table 2 shows, gender had a moderated negative correlation with attitude (r = -.37), and self-efficacy (r = -.31), but weakly negatively associated with
behavioral beliefs (r = -.25) and control beliefs (r = -.17). Results in Table 2 indicates that age had the marginally weak associations with all variables. Hence, to avoid any confusing results, only gender was included as the control variable in all analyses. Also, the pre-test results for dependent variables and mediators were controlled for all the analyses.
Table2.
Background variables check
gender Age behavioral beliefs-post r =-.25, p =.02 r =.13, p =.08 control beliefs-post r =-.17, p =.04 r =.11, p = .14 attitude-post r =-.37, p <.001 r =.13, p =.13 self-efficacy-post r =-.31, p <.001 r =.11, p =.16 Manipulation check
To test if participants in different experimental conditions recognized a different level of perceived similarity of characters (PSC), a one-way ANOVA was conducted. Results are showed in Table 3 indicate that participants in the high perceived similarity Groups 1 (M = 3.80, SD = 0.51) and 2 (M = 3.74, SD = 0.56) recognized a higher level of PSC than in the low perceived similarity Groups 3 (M = 2.76, SD = 1.05) and 4 (M = 2.56, SD = 0.80), F (3,149) = 27.17, p < .001. Thus, the manipulation was successful.
Table3.
Descriptive statistics of PSC by the different experimental conditions
High-gain (Group1) 3.80a 0.51 High-loss (Group2) 3.74a 0.56 Low-gain (Group3) 2.76b 1.05 Low-loss (Group4) 2.56b 0.80 F (3,149) 27.17 p < .001 Descriptive Statistics
Table 4 shows the Mean and Standard Deviation of all dependent variables and mediators after the experiment for all experimental groups. In general, behavioral beliefs, control beliefs, attitude and self-efficacy were marginally higher in the high perceived similarity conditions than in the low perceived similarity conditions. But in all, surprisingly, the variables in question scored high in all experimental groups.
Table4.
Mean and Standard Deviation for the experimental groups (the postcondition) High-gain (n = 40) High-loss (n = 34) Low-gain (n = 38) Low-loss (n = 41) behavioral beliefs 12.32 (1.85) 12.02 (1.89) 11.94 (2.22) 11.14 (2.72) control beliefs 23.50 (3.84) 22.35 (4.24) 22.45 (4.82) 22.27 (4.80) attitude 13.15 (1.14) 12.73 (1.98) 12.50 (1.81) 12.17 (2.45) self-efficacy 12.47 (2.13) 12.17 (2.24) 11.97 (2.48) 12.04 (2.34) Hypothesis test
Attitude and behavioral beliefs
To test the effects of different levels (high vs. low) in the perceived similarity of characters (PSC) on post-attitude and post-behavioral beliefs (H1a), the two-way ANCOVA analyses were conducted. For the attitude test, PSC and pre-attitude were used as the independent variables. As for the behavioral beliefs test, PSC and pre-behavioral beliefs were independent variables. Gender was included as the covariate for two tests. The results are summarized in Table 7.
Result indicated that PSC had a significant effect on attitude, F (1, 152) = 8.09, p = .005, but the effect was not significant on behavioral beliefs (p = .12). Participants who were in the high perceived similarity conditions increased attitude (M = 12.95, SD = 1.59) more than participants who were in the low perceived similarity conditions (M = 12.32, SD = 2.16). Thus, the H1a was partially confirmed.
H2a
In order to test the mediation effect of post-behavioral beliefs towards PSC (high vs. low) on post-attitude (H2a), Hayes’ PROCESS macro model 4 (Hayes, 2013) was used. Gender and pre-attitude were included as covariates. The results are summarized in Table 5.
There were no significant direct effect (p = .19) or indirect effect (BCBCI = [-.27, .28]) of PSC on attitude towards smoking. However, the results from PROCESS macro model 4 indicates that behavioral beliefs (b = .23, se = .06, t = 1.33, p < 0.01) was a significant predictor of attitude towards smoking cessation. Tough behavioral beliefs had an effect on attitude, PSC failed to influence the behavioral beliefs. Thus, PSC did not significantly affect attitude, neither directly nor indirectly. Hence, the H2a was rejected.
Table5.
Mediation effect for attitude
direct effect = .27 indirect effect = .11 behavioral beliefs on attitude
se = .21, p = .19
boot SE = .08, BCBCI = [-.27, .28]
b = .23, se = .06, t = 1.33, p < 0.01
H3a & H4a
To test the interactive effects made by PSC (high vs. low) and message framing (gain vs. loss) on post-attitude (H3a) and post-behavioral beliefs (H4a), two Univariate ANCOVA analyses were performed. PSC, message framing and pre-attitude were the independent variables for the attitude test. For the behavioral beliefs test, PSC, message framing and pre-behavioral beliefs were the independent variables. Gender was included as covariate in two tests. Results are summarized in Table 7.
Results showed that no significant main effects of message framing on attitude (p = 0.19) or on behavioral beliefs (p = .08). And also the interactive effects were not significant on attitude (p = .67), nor behavioral beliefs (p = .52). Thus, the H4a and H5a were both rejected. All the test results are presented in Statistical diagram 2.
Statistical diagram 2
Results of attitude and behavioral beliefs
Self-efficacy and control beliefs
H1b
To test the effects of perceived similarity of characters (PSC) on post-self-efficacy and post-control beliefs (H1b), two Univariate ANCOVA analyses were conducted. For the self-efficacy test, PSC and pre-self-efficacy were the independent variables. As for the
control beliefs test, PSC and pre-control beliefs were the independent variables. Gender was included as the covariate for both tests. The results are summarized in Table 7.
The result indicated that PSC in any condition had no significant effects on self-efficacy (p = .68), and effect also was not significant on control beliefs (p = .72). Thus, the H1b was rejected.
H2b
In order to test the mediation effect of post-control beliefs towards PSC (high vs. low) on post-self-efficacy (H2b), Hayes’ PROCESS macro model 4 (Hayes, 2013) was used. Gender and the pre-measure of self-efficacy were included as covariates. The results are summarized in Table 6.
There were no significant direct effect (p = .47) or indirect effect (BCBCI = [-.13, .50]) of PSC on self-efficacy towards smoking. However, the results indicates that control beliefs was a significant predictor of self-efficacy towards smoking cessation (b = .24, se = .27, t = .74, p < 0.01). Tough control beliefs had an effect on self-efficacy, PSC did not significantly influence the control beliefs. Thus, PSC refused to have a significant effect on self-efficacy, neither direct or indirect. Hence, the H2b was rejected.
Table6.
Mediation effect for self-efficacy
direct effect = .17 indirect effect = .17 control beliefs on self-efficacy
se = .22, p = .47 boot SE = .16, BCBCI = [-.13, .50] b = .24, se = .27, t = .74, p < 0.01 H3b & H4b
To test the interactive effects composed by PSC (high vs. low) and message framing (gain vs. loss) on post-self-efficacy (H3b) and post-control beliefs (H4b), two Univariate ANCOVA analyses were performed. For the self-efficacy test, PSC, message framing and pre
self-efficacy were the independent variables. For the control beliefs test, PSC, message framing and pre-control beliefs were the independent variables. Gender was included as covariate for both tests. Results are summarized in the Table 7.
Surprisingly, by differing with the attitude test, results shows that a significant main effect of message framing on self-efficacy, F (1,152) = 4.61, p = .03. However, for the control beliefs, no significant effect was found from message framing (p = .89). And also the interactive effects of homophily and message framing were not significant on self-efficacy (p = .43), nor control beliefs (p = .07). Hence, the H3b and H4b were rejected. All the test results are presented in Statistical diagram 3.
Statistical diagram 3
Result of self-efficacy and control beliefs
Table 7.
ANCOVA test of homophily and message framing
homophily effect
attitude F (1, 152) = 8.09, p = .005
behavioral beliefs F (1, 152) = 2.48, p = .12
control beliefs F (1, 152) =.13, p = .72 message framing effect
attitude F (1,152) = 1.77, p = .19
behavioral beliefs F (1, 152) = 3.08, p = .08
self-efficacy F (1,152) = 4.61, p = .03
control beliefs F (1, 152) = .21, p = .89 homophily * message framing effect
attitude F (1, 152) = .18, p = .67
behavioral beliefs F (1, 152) = .83, p = .52
self-efficacy F (1, 152) = .63, p = .43
control beliefs F (1, 152) = 3.48, p = .07
Conclusion
The present study has mixed results. The effects of homophily on attitude was confirmed (H1a), but to self-efficacy was denied (H1b). For the beliefs, homophily neither showed a significant effect on behavioral beliefs (H1a) nor on control beliefs (H1b). Thus, the H1a was partially confirmed but H1b was totally refused. The high perceived similarity in the entertainment-education intervention has more significant effects compared to the low perceived similarity on attitude towards smoking cessation. As for the mediation test, homophily did not present a significant effect on attitude via behavioral beliefs (H2a) and effect was also insignificant for self-efficacy via control beliefs (H2b). Thus, the H2a and H2b were also rejected. Nonetheless, behavioral beliefs had an influence on attitude and control beliefs significantly influenced self-efficacy, which are aligned with the integrated model. For the interactive effects, the results confirmed the significant main effect of message framing on self-efficacy, but defined the main effect from message framing on
by homophily and message framing towards attitude (H3a), behavioral beliefs (H4a), self-efficacy (H3b) and control beliefs (H4b) were all rejected.
Discussion & Limitation
This study primarily aims to explore the main effects of perceived similarity of characters and the moderation effects of message framing on attitude and self-efficacy towards smoking cessation in a digital interactive invention, based on the entertainment overcoming resistance model (Moyer-Guse, 2008). To thoroughly investigate the probability, I formed a model for homophily and message framing for attitude and self-efficacy change towards smoking cessation, including behavioral beliefs and control beliefs as mediators. A bunch of studies confirmed the persuasion effects of perceived similarity of characters on smoking cessation, no matter in the interactive or non-interactive interventions (Kim, Shi, & Cappella, 2016; Song, Kim, Kwon, & Jung, 2013). Besides, as message framing is confirmed to have the persuasion effects on smoking cessation in the non-interactive interventions (Gallagher & Updegraff, 2011), it is worth to explore the interactive effects of homophily and message framing towards smoking cessation in an interactive digital intervention. Hence, there are two goals of my study: first, does the perceived similarity of characters have significant effects on self-efficacy or attitude? Second, are such the effects mediated by beliefs and moderated by message framing strategy?
The result confirm the effect of perceived similarity on attitude is in line with the study from Moyer-Guse (2008), who proposed that attitude can be influenced by the perceived similarity in a narrative structure demonstrated in the entertainment overcoming resistance model.
However, by conflicting with the previous study, the results fail to find the effects from the perceived similarity of characters towards self-efficacy, behavioral beliefs as well as control beliefs on smoking cessation. A possible explanation is that an environmental
constraint about culture context hinders the effects. Assessment of environmental constraint is a significant prerequisite of self-efficacy on performing a desired behavior (Gist & Mitchell, 1992). The behavior is likely to take into action when people who master with the relevant skills are motivated and do not perceive the environmental constraints at the same time. Environmental constraint refers to the main factors that are around people limit the performance of a certain behavior. In this case, cultural context is served as an environmental constraint in my study due to all of the participants came from China that a nation is in the collectivist culture context. According to the integrated model, culture is an explainable factor for behavioral and control beliefs (Fishbein & Cappella, 2006). In China, there is a special norm that smoking is related to the gift tradition and social cohesion, caused by Chinese cultures (Zhang, Chan, Fong, and et al., 2012). Most of the smokers in China perceive social pressure as the constraint to quit smoking, such as gift-giving on social occasions (Ceraso, McElroy, Kuang, and et al., 2009; Zhang, Chan, Fong, and et al., 2012). Thus, for the most of Chinese smokers, culture and social context presumably limit their self-efficacy towards smoking cessation. Due to the culture and social context were not added into the experiment, it is still vague that how the Chinese culture influences the effects from perceived similarity and message framing on attitude, self-efficacy and beliefs. In the future, to better explore the smoking cessation behavior in the collectivist culture nations, such as in China or North Korea, it is essential to include the social context and culture traditions as background variables to avoid the confusing results.
In addition, the result also refuses the interactive effects combined by the perceived similarity and message framing on attitude, behavioral beliefs, self-efficacy and control beliefs. People who were in the high perceived similarity conditions did not change their beliefs, attitude and self-efficacy when they read the gain-framing messages more than those who read the loss-framing messages. And people in the low perceived similarity conditions
were not significantly influenced when they read the loss-framing messages than those who read the gain-framing messages.
This result declines my original assumption that in a narrative storyline, a gain-framing message in a high perceived similarity condition brings about more enjoyment to viewers than a loss-framing message, which increases the engagement to the perceived similarity with characters then promote to change attitude, self-efficacy and beliefs. And people who were in the low perceived similarity conditions may perceive less fear towards the loss-framing messages should be significantly influenced, compared to those who read a gain-framing message. A possible reasons for the failure of the gain-framing is that, compared to the gain-framing messages, loss-framing message equally brought the enjoyment to participants as the gain-framing message in the stimuli. Although at the end of the story, the protagonist was failed to stop smoking, which is a negative result compared to the successfully quitting result. The story did not demonstrate any negative consequences (i.e. lung cancer or a shorter life expectancy) caused by failed to quit smoking, then participants did not perceive the fear at all, which also explains for the defeat of the loss-framing in the low similarity condition. By adding with the virtual storyline and protagonists presenting in the enjoyable cartoon images, the storyline in the loss-framing message was as pleasurable as the gain-framing messages, so the engagement in both framing conditions was not significantly different. In the future study, it is reasonable to measure the difference in people’s engagement in both the gain-framing messages and loss-framing messages, to avoid
the unclear result.
Interestingly, the result confirms that behavioral beliefs have an influence on attitude and control beliefs can affect self-efficacy, which is aligned with the integrated model. It is another empirical finding to support that beliefs are the stronger predictors for attitude change and self-efficacy increase in a smoking cessation topic in particular. And also, an effect of
message framing towards self-efficacy is confirmed, which is consistent with Roathman (2006). A gain-framing message is significantly more likely than a loss-framing message to promote prevention behaviors, as self-efficacy is an important predictor of smoking cessation.
The main criticism of my study is the sample bias that male smokers (137) are dominantly more than female smokers (16). Although the majority of Chinese smokers are male, it is worth to explore the effect of gender on homophily and message framing towards smoking cessation. According to the previous research, female is more responsive to gain-framing messages than male towards prevention behaviors (e.g. applying the sun cream), and female are more engaged into messages than male during the intervention (Rothman, Salovey, Antone and et al., 1992), thus, the intervention might be more persuasive for female than male. Accordingly, it is essential to consider the effect of gender towards smoking cessation in a digital E-E intervention.
Additionally, as mentioned earlier, another limitation in my study is the lack of distinguish between the gain-framing and loss-framing messages. The negative consequences caused by failed to stop smoking should be addressed at the end of the story to reduce the enjoyment and increase more fear, which reminds participants that a severe result may happen if you fail to stop smoking.
Implications
The current study explores the influence of perceived similarity of characters and message framing theory on attitude, self-efficacy, behavioral beliefs and control beliefs towards smoking cessation within a digital interactive intervention, but fail to add proofs to affirm the most effects of homophily on self-efficacy, behavioral beliefs and control beliefs in the entertainment-education health intervention. However, it confirms the persuasion effect to attitude. Results found that participants are significantly influenced by the digital E-E intervention on attitude but not on self-efficacy and beliefs. Thus, it is applicable to apply the
E-E program to change Chinese smokers’ attitudes towards smoking in the future. Results also show that participants are more responding to a gain-framing message than a loss-framing message to increase self-efficacy towards smoking cessation. Besides, the study result proves the predictable effects of behavioral beliefs and control beliefs to attitude and self-efficacy, which is another empirical evidence for the integrated model under the smoking cessation topic. For future interventions, it is applicable to include beliefs in order to change people’s attitudes and self-efficacy.
Appendix A
The character images Chinese male student image
Chinese male adult image
Chinese female student image
Chinese female adult image
References
Ajzen, I., & Driver, B. L. (1991). Prediction of leisure participation from behavioral, normative, and control beliefs: An application of the theory of planned behavior. Leisure sciences, 13(3), 185-204.
Ajzen, I. (2006). Constructing a theory of planned behavior questionnaire.
Austin, E. W., Roberts, D. F., & Nass, C. I. (1990). Influences of family communication on children's television-interpretation processes. Communication research, 17(4), 545-564.
Bandura, A. (2009). Social cognitive theory of mass communication. In Media effects (pp. 110-140). Routledge. Borland, R., Yong, H. H., Balmford, J., Cooper, J., Cummings, K. M., O'Connor, R. J., ... & Fong, G. T. (2010).
International Tobacco Control Four country project. Nicotine & Tobacco Research, 12(suppl_1), S4-S11. doi:10.1093/ntr/ntq050
Brodie, M., Foehr, U., Rideout, V., Baer, N., Miller, C., Flournoy, R., & Altman, D. (2001). Communicating health information through the entertainment media. Health affairs, 20(1), 192-199.
Brusse, E. D. A., Fransen, M. L., & Smit, E. G. (2017). Framing in entertainment-education: Effects on processes of narrative persuasion. Health communication, 32(12), 1501-1509.
doi:10.1080/10410236.2016.1234536
Ceraso, M., McElroy, J. A., Kuang, X., Vila, P. M., Jorenby, D. E., Fiore, M. C., ... & Ren, H. (2009). Peer Reviewed: Smoking, Barriers to Quitting, and Smoking-Related Knowledge, Attitudes, and Patient Practices Among Male Physicians in China. Preventing chronic disease, 6(1).
Delucchi, K. L., Tajima, B., & Guydish, J. (2009). Development of the smoking knowledge, attitudes, and practices (S-KAP) instrument. Journal of drug issues, 39(2), 347-363.
Dickey, M. D. (2007). Game design and learning: A conjectural analysis of how massively multiple online role-playing games (MMORPGs) foster intrinsic motivation. Educational Technology Research and Development, 55(3), 253-273. doi:10.1007/s11423-006-9004-7
Ensher, E. A., & Murphy, S. E. (1997). Effects of race, gender, perceived similarity, and contact on mentor relationships. Journal of Vocational Behavior, 50(3), 460-481.
Etter, J. F., Bergman, M. M., Humair, J. P., & Perneger, T. V. (2000). Development and validation of a scale measuring self‐efficacy of current and former smokers. Addiction, 95(6), 901-913.
Fishbein, M., & Cappella, J. N. (2006). The role of theory in developing effective health communications. Journal of communication, 56, S1-S17. doi:10.1111/j.1460-2466.2006.00280.x
Gallagher, K. M., & Updegraff, J. A. (2012). Health message framing effects on attitudes, intentions, and behavior: a meta-analytic review. Annals of behavioral medicine, 43(1), 101-116.
doi:10.1007/s12160-011-9308-7
Graaf, A. D., Sanders, J., & Hoeken, H. (2016). Characteristics of narrative interventions and health effects: A review of the content, form, and context of narratives in health-related narrative persuasion research. Review of Communication Research, 4, 88-131. doi:10.12840/issn.2255-4165.2016.04.01.011
Gist, M. E., & Mitchell, T. R. (1992). Self-efficacy: A theoretical analysis of its determinants and malleability. Academy of Management review, 17(2), 183-211.
regression-based approach. Guilford publications.
Hoffner, C., & Buchanan, M. (2005). Young adults' wishful identification with television characters: The role of perceived similarity and character attributes. Media psychology, 7(4), 325-351.
Hughes, J. R., & Hatsukami, D. (1986). Signs and symptoms of tobacco withdrawal. Archives of general psychiatry, 43(3), 289-294.
Hummel, K., Candel, M. J., Nagelhout, G. E., Brown, J., van den Putte, B., Kotz, D., ... & de Vries, H. (2018). Construct and predictive validity of three measures of intention to quit smoking: Findings from the International Tobacco Control (ITC) Netherlands Survey. Nicotine and Tobacco Research, 20(9), 1101- 1108. doi:10.1093/ntr/ntx092
Kim, M., Shi, R., & Cappella, J. N. (2016). Effect of character–audience similarity on the perceived effectiveness of antismoking PSAs via engagement. Health Communication, 31(10), 1193-1204. doi:10.1080/10410236.2015.1048421
Krebs, P., Prochaska, J. O., & Rossi, J. S. (2010). A meta-analysis of computer-tailored interventions for health behavior change. Preventive medicine, 51(3-4), 214-221.for health promotion and behavioral risk reduction: a meta-analysis of 75 randomized controlled trials, 1988–2007. Preventive medicine, 47(1), 3-16.
doi:10.1016/j.ypmed.2008.02.014
Lee, M. J., & Bichard, S. L. (2006). Effective message design targeting college students for the prevention of binge-drinking: Basing design on rebellious risk-taking tendency. Health communication, 20(3), 299-308. doi:10.1207/s15327027hc2003_9
Maurer, T. J., & Pierce, H. R. (1998). A comparison of Likert scale and traditional measures of self-efficacy. Journal of applied psychology, 83(2), 324.
Meischke, H., Lozano, P., Zhou, C., Garrison, M. M., & Christakis, D. (2011). Engagement in “my child's asthma”, an interactive web-based pediatric asthma management intervention. International journal of medical informatics, 80(11), 765-774. doi:10.1016/j.ijmedinf.2011.08.002
Moorman, M., & van den Putte, B. (2008). The influence of message framing, intention to quit smoking, and nicotine dependence on the persuasiveness of smoking cessation messages. Addictive behaviors, 33(10), 1267-1275. doi:10.1016/j.addbeh.2008.05.010
Moyer-Gusé, E. (2008). Toward a theory of entertainment persuasion: Explaining the persuasive effects of entertainment-education messages. Communication theory, 18(3), 407-425.
Murphy, S. T., Frank, L. B., Chatterjee, J. S., & Baezconde-Garbanati, L. (2013). Narrative versus nonnarrative: The role of identification, transportation, and emotion in reducing health disparities. Journal of
Communication, 63(1), 116-137. doi:10.111/jcom.12007
Portnoy, D. B., Scott-Sheldon, L. A., Johnson, B. T., & Carey, M. P. (2008). Computer-delivered interventions Quinn, C. N. (1994, July). Designing educational computer games. In Proceedings of the IFIP TC3/WG3. 2 Working Conference on the Seign, Implementation and Evaluation of Interactive Multimedia in University Settings: Designing for Change in Teaching and Learning (pp. 45-57).
Rich, Z. C., & Xiao, S. (2012). Tobacco as a social currency: cigarette gifting and sharing in China. Nicotine & Tobacco Research, 14(3), 258-263. doi:10.1093/ntr/ntr156
Ritterfeld, U., Cody, M., & Vorderer, P. (Eds.). (2009). Serious games: Mechanisms and effects. Routledge. Rothman, A. J., Bartels, R. D., Wlaschin, J., & Salovey, P. (2006). The strategic use of gain-and loss-framed messages to promote healthy behavior: How theory can inform practice. Journal of communication, 56(suppl_1), S202-S220. doi:10.1111/j.1460-2466.2006.00290.x
Sharma, M., Mehta, M., Ontanón, S., & Ram, A. (2007, June). Player modeling evaluation for interactive fiction. In Proceedings of the AIIDE 2007 Workshop on Optimizing Player Satisfaction (pp. 19-24).
Singhal, A., & Rogers, E. M. (2001). The entertainment-education strategy in communication campaigns. Public communication campaigns, 3, 343-356.
Song, H., Kim, J., Kwon, R. J., & Jung, Y. (2013). Anti-smoking educational game using avatars as visualized possible selves. Computers in Human Behavior, 29(5), 2029-2036. doi:10.1016/j.chb.2013.04.008
Shrum, L. J. (2004). The cognitive processes underlying cultivation effects are a function of whether the judgments are on-line or memory-based. Communications-Sankt Augustin Then Berlin-, 29, 327-344. doi:10.1515/comm.2004.021
Toll, B. A., O’Malley, S. S., Katulak, N. A., Wu, R., Dubin, J., George, T. P., ... & Salovey, P. (2006). Message framing for smoking cessation with bupropion: A randomized controlled trial. In annual meeting of the Society for Research on Nicotine and Tobacco, Orlando, Florida.
Vandelanotte, C., Spathonis, K. M., Eakin, E. G., & Owen, N. (2007). Website-delivered physical activity interventions: A review of the literature. American journal of preventive medicine, 33(1), 54-64. Wald, N. J., & Hackshaw, A. K. (1996). Cigarette smoking: an epidemiological overview. British medical bulletin, 52(1), 3-11.
condom use in heterosexual men: development of a theory-based interactive digital intervention. Translational behavioral medicine, 6(3), 418-427. doi:10.1007/s13142-015-0338-8
