An experiment investigating the role of message processing in explaining the different persuasive
effects of framing on social norm messages
12971057 Master’s Thesis
Graduate School of Communication
Master’s programme Communication Science
University of Amsterdam Dr. Saar Mollen
Word count: 8,217
Date of completion: 4 February 2022
Table of contents
Abstract ... 2
Introduction ... 3
Theoretical background ... 6
Method ... 13
Results ... 20
Discussion ... 24
Conclusion ... 27
References ... 27
Appendix A – Online experiment ... 33
Appendix B – Pre-analysis plan ... 44
Appendix C - Stimulus material ... 46
Appendix D - Results main analysis ... 48
This study aims to explore whether message processing explains the differential persuasive effects of framed (i.e., positive, negative) social norm messages (i.e., injunctive, descriptive) on anticipated food choice. Based on theory, it is assumed that message frames elicit either lower or higher levels of processing and that social norms require either higher or lower levels of processing. It is hypothesized that a negative message frame will result in more message processing compared to a positive message frame. This study employed a thought- listing exercise to examine whether different levels of message processing can explain the persuasive effects of framing and whether there is a direct effect of frame on message processing. Furthermore, it is hypothesized that a correct fit between frame, eliciting high (i.e., negative) or low (i.e., positive) levels of processing, and social norm, requiring high (i.e., injunctive) or low (i.e., descriptive) levels of processing, will result in a more effective persuasive message. Thus, negatively framed injunctive social norms message and positively framed descriptive social norms message should be more persuasive than positively framed injunctive social norm messages and negatively framed descriptive social norm messages. This study employed a behavioral task to test for an interaction effect of frame and norm on food choice.
Results show neither a significant main effect of framing on message processing nor a significant interaction effect of frame and norm on anticipated food choice. However, a marginally significant interaction effect of frame and norm was found on the number of reported thoughts. The results signify that more research is necessary, specifically, towards the relationship between framing, social norm, and message processing.
According to the World Health Organization (WHO), obesity and overweight are problems of an epidemic proportion, resulting in a large array of dire health issues (WHO, 1998; WHO, 2000). Such issues include but are not limited to cardiovascular disease, diabetes, and chronic kidney disease (GBD 2015 obesity collaborators, 2017). Overweight is a rapidly increasing trend that needs to be addressed (WHO, 1997; WHO, 2000), being a problematic phenomenon in most parts of the world. Including in the Netherlands, where increased levels of overweight have manifested themselves throughout the entire Dutch population (Schokker et al., 2007), mainly forming a problem among young adults (Centraal Bureau voor de Statistiek, 2019). Including Dutch university students (Vos et al., 2015), which on average adopt less healthy eating habits and gain weight.
According to previous research, social norms can have a powerful effect on food choices (Higgs, 2015) and the amount that people consume (Herman et al., 2003; Higgs, 2015).
Considering that social norms can influence food consumption, utilizing social norms may effectively change someone's diet. For instance, social norm messages, which are meant to influence a behavioral outcome (e.g., food consumption) by presenting a social norm (i.e., injunctive, descriptive). The focus theory of normative conduct (Cialdini et al., 1990) distinguishes between both types of norms; a descriptive social norm relates to the behavior performed by most others, and an injunctive social norm relates to the perceived approval or disapproval of behavior by most others (Cialdini et al., 1990). Social norm messages concerning health behavior can be effective, especially when addressing students. Research shows that students' or young adults' health-related behavior are more likely to be predicted by social norms and social influence than the health-related behavior of older adults (McEachan et al., 2011; Rivis & Sheeran, 2003).
Furthermore, the effectiveness of social norm messages can be altered or even enhanced. According to Cialdini and colleagues (2006), phrasing messages in negative or positive terms (i.e., framing) alters the response to norm-based persuasive messages. Positively framed social norms focus on behavior that is encouraged (i.e., injunctive) or performed by others (i.e., descriptive). Negatively framed social norms focus on behavior that is discouraged (i.e., injunctive) or not performed by others (i.e., descriptive) (Cialdini et al., 2006). The findings above demonstrate the motive and relevance of using health-related social norm messages to change food consumption among students.
However, the effect of framing on social norm messages has proven to vary. It remains uncertain which combination of frame and norm is most persuasive. Evidence suggests that negatively framed, rather than positively framed, social norm messages will increase people's adherence to an advocated social norm (Cialdini et al., 2006; Hassel & Wyler, 2018). However, both studies accredit the positive effect of a negative frame to different social norms. Moreover, Cialdini et al. (2006) show that the difference in effect between an injunctive and a descriptive norm message will increase when both are framed negatively (p. 11). These findings hint at a possible interaction effect of norm and frame on behavior. Mollen and colleagues' (2016) work partially proves this assumption. Their results only show a significant interaction between a negative frame and an injunctive norm. To conclude, studies show a discrepancy in the effect of negatively framing social norms. Additionally, the assumption that frame and norm have an interaction effect on message effectiveness remains only partially proven. This study aims to show an interaction effect between a positive frame and a descriptive norm.
Underlying and explaining this interaction effect of social norm and message frame, Mollen et al. (2016) propose that differing levels of message processing might play a mediating role. That is to say that the influence of framing might depend on message processing.
However, evidence to support this claim is missing. There is no significant statistical
correlation between message processing and the increased or decreased effect of frame and norm on behavior. Even so, differing levels of message processing might still be a crucial variable in explaining the effectiveness of framing social norm messages. The work of Maheswaran and Meyers-Levy (1990) explores whether message processing accounts for the differential effects of framing on persuasion. Their results propose that the level of issue involvement, which affects message processing (Petty & Cacioppo, 1986), could determine which frame is more persuasive. They conclude that positively framed messages might be more persuasive when less detailed processing is required and that negatively framed messages might be more persuasive when more detailed processing is required.
This thesis aims to address two gaps in the social norm literature. First, the inconsistent findings of framing effects and secondly, the valence of message processing, as an underlying process that might help explain the differential persuasive effects of framing on social norm messages. Achieving these goals could help broaden the current field of persuasive communication on two levels. First, understanding a possible mediating effect of processing could help further delineate which mechanism(s) underlie(s) the effectiveness of framed social norms messages. Second, it could advance the effectiveness of framed social norms messages, set to positively influence eating habits.
This study attempts to extend the findings of Mollen et al. (2016) and aims to show whether the absence of a significant effect is methodological or theoretical by using an alternative measure to study how processing might explain the varying effects of framed social norms. The following research question aims to achieve the set goals:
What are the differential effects of framing on injunctive versus descriptive social norm messages, and can this be explained by differing levels of message processing?
Social norms are rules and standards that help guide people to behave appropriately within a group (Cialdini & Trost, 1998). The feeling of belonging to that group is a prerequisite for adhering to those norms. Social norms consist of two different types: injunctive and descriptive (Cialdini et al., 1991). According to the focus theory of normative conduct (Cialdini et al., 1990), both types are relevant for different fundamental human goals (Jacobsen et al., 2011). Whereas the descriptive norm provides information relevant for behaving correctly, the injunctive norm is relevant for determining which behavior is socially appropriate and aids in building and maintaining social relationships (Cialdini & Trost, 1998; Jacobsen et al., 2011, p.434). The goals are respectively: choosing the correct behavior and obtaining social approval.
The focus theory of normative conduct also proposes that social norms can both initiate and influence behavior, but only when made salient to those meant to perform the behavior (Cialdini et al., 1990; Cialdini et al., 1991; Cialdini et al., 2006). Hence, messages that increase the salience of a norm can effectively change behavior (Farrow et al., 2017). Considering that perceived norms have a powerful effect on food consumption (Herman et al., 2003; Higgs, 2015), normative messages can become an effective method of changing someone’s dietary intake.
Persuasion and information processing
According to the ELM (Petty & Cacioppo, 1986), there are two processing routes to persuasion, explaining the effects of persuasive communication on attitude change. These routes vary in the level at which people process information and arguments. People are persuaded by actively processing the presented information via careful consideration - i.e., the central route. Alternatively, people are persuaded without scrutinizing or deeply processing the information and let their attitudinal change be guided by simple heuristic message cues
(e.g., attractive source) – the peripheral route. A person’s motivation and ability to process or scrutinize an argument predicts via which route a message is processed (Petty & Cacioppo, 1986). Personal relevance of the message topic and a need for cognition (i.e., the predisposition to enjoy and engage in an effortful analysis of information, or not) (Cohen et al., 1955; Petty
& Cacioppo, 1986) are primary factors that motivate or enable a person to scrutinize arguments.
The higher a person’s ability and motivation to scrutinize arguments, the more likely that person processes information via the central route. Furthermore, lower ability and motivation to elaborate will likely result in information processing via the peripheral route.
Petty and Cacioppo (1986) show that the depth of message processing can vary, ultimately affecting message persuasiveness and its ability to change attitudes. Information processed via the central route can result in longer-lasting altitudinal changes and has a higher tendency to translate into behavioral change. Therefore, message processing is an essential factor to consider when trying to understand the success of persuasive communication (Shen
& Dillard, 2009; Slater, 2006).
Social norms and information processing
The differences between the underlying goals of descriptive and injunctive norms have implications for processing normative information. A review in the meta-analysis by Farrow and colleagues (2017) supports the idea that descriptive norms can function as heuristic decision-making cues that require less cognitive effort. This notion is based on Melnyk et al.’s (2011) research, whose results show that an increased cognitive load increases the influence of descriptive but decreases the influence of injunctive norms. The work of Morris and colleagues (2015) theorizes that the decision to comply with an injunctive norm will most likely require more significant cognitive effort when compared to a descriptive norm. They base their assumption on the following: (1) injunctive norms provide normative information to act in a socially acceptable manner, (2) injunctive norms influence behavior less in situations that
demand a higher cognitive load, and (3) injunctive norms can activate both an interpersonal and an intrapersonal goal (Jacobson et al., 2011; Kredentser et al., 2012). A deliberation regarding compliance to injunctive norms will involve strategic considerations concerning social status and material benefit, therefore, demand increased cognitive effort (Farrow et al., 2017).
To explain why injunctive norms might require more cognitive processing, Jacobsen et al. (2011) state that the two activated goals are motivated by dual motives and thus might conflict. They conflict because the best interest of the independent self (i.e., intrapersonal) does not necessarily align with the best interest of the social collective (i.e., interpersonal). For instance, an injunctive norm set by a mother (i.e., She wants her child to finish their vegetables) provides the child with the option to achieve the interpersonal goal of showing that they listen and gain her social approval. However, eating vegetables might conflict with their intrapersonal goal of only eating food they like; deciding which goal to achieve costs cognitive effort. Thus, an injunctive norm can result in conflicting goals and the necessity to resolve that conflict, forcing a person to strategically deliberate and choose between both goals. Hence, increasing the required amount of information processing possibly needed to respond to an injunctive norm. Taken together with the findings that descriptive norms are more influential under conditions of low elaboration and injunctive norms are more influential under conditions of high elaboration (Jacobson et al., 2011; Kredentser et al., 2012; Melnyk et al., 2011); injunctive norms are theorized to require more cognitive effort to process than descriptive norms (Farrow et al., 2017; Jacobson et al., 2011; Kredentser et al., 2012; Melnyk et al., 2011; Morris et al., 2015). Therefore, this study assumes that, on average, people will require more message processing when reading an injunctive norm message compared to reading a descriptive norms message.
9 Framing and information processing
Framing, or message frame, relates to the distinct way in which specific outcomes or goals regarding a decision are positioned in a message (Levin et al., 1998). These frames can position behavior in terms of what is discouraged or not performed by most others (i.e., negative frame) or what is encouraged or performed by most others (i.e., positive frame).
There is a difference in the depth of information processing triggered between both frames. This difference can be explained due to a negativity bias (Cialdini et al., 2006), which is the general tendency of adverse events to be more salient and to attract more attention than positive events (Rozin & Royzman, 2001). A specific aspect of this bias, called negative potency, asserts that negative events appear more potent and salient than positive events.
Strongly related to the concept of loss aversion (Rozin & Royzman, 2001, p. 298; Kahneman
& Tversky, 1979), negative potency is argued as a driving factor explaining why negative information commands more attention (Rozin & Royzman, 2001, p. 301). From an evolutionary standpoint, Dijksterhuis and Aarts (2003) suggest that people process negative stimuli faster and more in-depth than positive stimuli. Mollen and colleagues (2016) state that negative frames, because they elicit more attentional and processing resources than positive frames (p.348), will impact the effectiveness of descriptive and injunctive norms differently.
Negatively framed information will elicit more processing when compared to positively framed information (Baumeister et al., 2001; Maheswaran & Meyers-Levy, 1990; Smith &
Petty, 1996). In contrast, Jones and colleagues (2003) found a link between positively framed messages, increased levels of processing, and a stronger intention to exercise. However, these effects do not translate into a change of actual behavior. Despite previous research, it remains indefinite which message frame increases message processing (Gallagher & Updegraff, 2011;
Rothman & Updegraff, 2010; O’Keefe & Jensen, 2008). Therefore, and following the theory of negativity bias, this study assumes that negatively framed messages will elicit more message
processing when compared to positively framed messages. Subsequently, the findings of this study could help clarify which frame will increase message processing.
Social norm, framing, and processing
As a result of different effects of framing on attention, a particular combination of frame and norm might positively enhance the persuasiveness of a social norm message (Mollen et al., 2016). Combining a norm that requires higher or lower levels of cognitive processing (i.e., injunctive, descriptive, respectively) with a message frame that either elicits higher or lower levels of processing recourses (i.e., negative, positive, respectively) should eventually result in a more persuasive message. In this case, a more persuasive message would more strongly convince a recipient to eat healthier.
To test the assumptions, that the processing required by injunctive norms fits better with the processing elicited by negative frames, and that the processing required by descriptive norms fits better with the processing elicited by positive frames, Mollen et al. conducted two experiments and found a significant interaction between motivation to consume healthy food, norms, and frame. They found that the motivation to consume fruit -rather than candy- was higher when the injunctive norm message was framed negatively rather than positively.
However, this interaction was not significant. For descriptive norms messages, they found the opposite; the motivation to consume fruit -rather than candy- was stronger with positive frames than negative frames. This interaction was marginally significant.
The results of the second experiment with actual food intake as the dependent variable show a partial interaction effect of norm and frame on the consumption of healthy and unhealthy food. Expressly, negatively framed injunctive norms, compared to positively framed injunctive norms, result in participants consuming more healthy food (i.e., fruit). Moreover, positively framed injunctive norms, compared to negatively framed descriptive norms, resulted in more unhealthy food (i.e., candy) being consumed (Mollen et al., 2016, p.367). In contrast,
the results only show a partial interaction effect of norm and frame when considering descriptive norms with a positive message frame. In addition, positively framed descriptive norms do not seem to significantly affect the behavioral task (Mollen et al., 2016).
The findings support the assumption that injunctive norm messages benefit from a negative frame rather than a positive frame. However, it remains less conclusive if the reverse interaction effect exists between a positive frame and descriptive norm messages. Adding necessity to the current study, which hopes to find a significant interaction effect of both norms and frames on food choice.
Moreover, Mollen and colleagues (2016) found no evidence regarding the underlying process of message processing. They ascribe the absence of processing effects on possible methodological grounds (p. 371). The explicit self-report measures might not be a valid method to examine whether message processing or involvement with the persuasive text has changed because of framing. The outcome of a recent study has proposed that, when measuring message elaboration, self-report items might be more suitable for entertainment-related information, and though-listing, more suitable for persuasive or educational information (Shen & Seung, 2018).
Thus, measuring the message processing of framed social norms, using self-report items rather than thought-listing, might explain the absence of processing effects.
Considering persuasive messages, framing and social norms either elicit or require different levels of message processing, respectively (Baumeister et al., 2001; Farrow et al., 2017; Jacobson et al., 2011; Kredentser et al., 2012; Melnyk et al., 2011; Maheswaran &
Meyers-Levy, 1990; Morris et al., 2015; Smith & Petty, 1996). Subsequently, Mollen and colleagues (2016) show that a negative frame and an injunctive norm interact to positively affect food consumption. Finally, according to Petty and Cacioppo (1986), message processing can vary and affect message persuasion. Therefore, processing influences -in part- the success of persuasive communication (Shen & Dillard, 2009; Slater, 2006). It is theorized that social
norm messages, adhering to the correct fit of processing requirements and elicited processing recourses, will result in a more effective persuasive message than those norm messages that do not fit the elicited and required processing resources. In contrast, studies examining the effects of framing on social norms or the effects of norm and frame on processing result in conflicting or insignificant results (Cialdini et al., 2006; Hassel & Wyler, 2018; Mollen et al., 2016).
The current study attempts to replicate and extend the findings of Mollen et al. (2016), performing a conceptual replication study that uses an alternative measure for processing. The underlying factor, the amount of message processing, might help explain the varying effects of framed social norms on food choice. In the study of Mollen and colleagues, they measured message processing using six explicit self-report items. The current study employs a thought- listing exercise. Which is more suitable to measure message processing concerning persuasive information, than self-reporting (Shen & Seung, 2018). By altering the measurement of message processing, this experiment hopefully demonstrates whether the previous absence of a significant effect of message processing was due to methodological or theoretical grounds.
Based on message processing theory (Maheswaran & Meyers-Levy, 1990;; Petty &
Cacioppo, 1986; Smith & Petty, 1996) and framed social norms (Farrow et al., 2017; Jacobson et al., 2011; Kredentser et al., 2012; Melnyk et al., 2011; Morris et al., 2015), as well as preliminary empirical evidence (Mollen et al., 2016), it is hypothesized that negatively framed injunctive norm messages and positively framed descriptive norm messages, will result in a more persuasive message, compared to positively framed injunctive norm messages and negatively framed descriptive norm messages. Finally, Mollen and colleagues (2016) find that participants exposed to a negatively framed injunctive norm made healthier food choices compared to all other combinations of frame and norm. Based on these findings, the following hypotheses have been formulated.
Hypothesis 1: Negatively framed social norm messages will result in more message processing compared to positively framed social norm messages.
Hypothesis 2: A negatively framed injunctive social norm message will result in a healthier food choice when compared to a positively framed injunctive norm message. While a positively framed descriptive norm message will result in a healthier food choice, than a negatively framed descriptive social norm message.
Hypothesis 3: A negatively framed injunctive social norm message will result in a healthier food choice when compared to a negatively framed descriptive norm message. While a positively framed descriptive norm message will result in a healthier food choice, than a positively framed injunctive social norm message.
Participants and Research design
Participants were randomly assigned to one of four conditions in a 2 (social norm:
injunctive vs. descriptive) x 2 (message framing: positive vs. negative) between-subjects factorial design (See appendix C).
Participants in this experiment were recruited via two methods: (1) the research platform of the University of Amsterdam (www.lab.uva.nl) and (2) snowball sampling and convenience sampling amongst fellow UvA students (who were contacted via WhatsApp). The ethical review board approved the procedure (2021-PC-14268). Only participants that partook in the experiment via the UvA research platform were rewarded study credits. Participants contacted via WhatsApp did not receive any credits.
To ensure sufficient statistical power and to minimize the chance of Type I and Type II errors, each subsample per experimental condition should include 30 to 40 participants (Geuens
& De Pelsmacker, 2017). Accordingly, the current study aimed to include at least 50
participants per experimental condition. In total, 518 participants partook in the online experiment, and 182 completed it (35,1%).
Participants were omitted from the experiment and analysis when they did not meet the exclusion criteria (see Appendix B). Participants, fifteen years or younger, were excluded from the experiment (n = 1). Additionally, an outlier on age was found (i.e., 84 yrs old), this participant was excluded from the analysis. To benefit the experiment’s internal validity, participants who indicated not being able to read at a university level (n = 65) and did not pass the attention check (n = 225), were excluded from the experiment as well. Furthermore, for a social norm message to have a persuasive effect, the receiver must relate to the social group from which the norm stems (Cialdini et al., 1990), in this case: UvA students. To strengthen the experiment’s internal validity, participants who indicated not being UvA students were excluded from the experiment (n = 14). Also, participants who failed to complete the behavioral task by selecting more or less than three options were excluded from the analysis (n = 18).
Finally, the participants were told a cover story to obscure the experiment’s aim. They were told that the experiment aimed to understand people’s attitudes towards media messages concerning lifestyles. Subsequently, all participants were asked whether they could guess the research topic. Those participants who could explain the aim of the research study were excluded from the analysis (n = 5). After omitting all participants that did not meet the prerequisites, the eventual sample (n = 182) consisted out of 40 men (22,0%), 141 women (77,5%), and one participant who identified as a third gender, with a mean age of 21, (SD = 3,26).
The online experiment started with a general explanation of the study and a letter of informed consent (see Appendix A). Then all participants were asked to answer two demographic questions concerning gender and age. Subsequently, the participants were asked
to answer four questions related to education and occupation. The items asked them to indicate (1) their level of English reading proficiency, (2) whether they were a student at the UvA, (3) their level of education (i.e., bachelor, master, doctorate), and (4) their current occupation. The last item was an attention check. Age, reading proficiency, studying at the UvA, and the attention-check were all exclusion criteria (see Participants).
After these items, all participants were randomly assigned to one of the four conditions and exposed to a message ostensibly stemming from the student website FOILA. The participants were told that the excerpt would be a ‘media message concerning lifestyles.’ After being exposed to the message, four filler questions (see Appendix A) were presented, relating to lifestyle choices. Next, the participants were asked to select three snacks they would like to eat the following day (a behavioral measure), followed by a thought-listing exercise to measure the amount of thought they had put into the message and six items measuring their intention towards healthy and unhealthy snacks. Additionally, the participants were asked to recall the message they had just read. After this, six items measured participants’ norm perceptions (manipulation check). Finally, all participants were asked what they thought the goal of the research study was and if they had any comments or remarks. Finally, participants were debriefed and asked for permission to collect and analyze their data.
Observed variables - Independent variables
Social norm messages. This variable has two levels: injunctive and descriptive. An injunctive social norm message provides the participant with a generalized opinion of the target population regarding eating healthy or unhealthy snacks. A descriptive social norm message describes the number of people who eat healthy or unhealthy snacks within the social group.
Framed social norm messages. An injunctive norm message is framed positively when the message describes the approval related to desirable behavior. An injunctive message is framed negatively when the message relates to the disproval of undesirable behavior (Mollen et al.,
2016). A descriptive norm message is framed positively when the message describes a majority engaging in a desirable behavior. A descriptive norm message is framed negatively when the message describes that a majority is not engaging in undesirable behavior (Mollen et al., 2016).
The positive injunctive norm (see Appendix C) stated that, within the population of UvA students, the majority believes that one should eat healthily and that eating healthy items (i.e., fruit, and vegetables) is good: ‘More and more people are convinced that you should eat healthy food as a snack. (…) Try to, more often, eat fruits and vegetables as a snack in between meals, because that is good”. The negative injunctive norm (see Appendix C) stated that, within the population of UvA students, the majority believes that you should not eat unhealthy foods, and that unhealthy snacks (i.e., candy and snacks) are bad ‘More and more people are convinced that you should not eat healthy food as a snack. (…) Try to, less often, eat candy or crisps as a snack in between meals, because that is bad”. The positive descriptive norm (see Appendix C) stated that, within the population of UvA students, an increasing majority eats healthy food as a snack: ‘The number of people who eat healthy snacks is increasing. (…) More than 75% of UvA students say that they often eat fruit or vegetables when they fancy a snack in between meals.’. The negative descriptive norm (see Appendix C) stated that, within the population of UvA students, an increasing majority does not eat unhealthy food as a snack:
‘The number of people who eat unhealthy snacks is decreasing. (…) More than 75% of UvA students say that they almost never eat candy or crisps when they fancy a snack in between meals.’.
The study included three dependent variables, message processing, intention, and a behavioral measure.
Behavioral measure. The behavioral measure provided the participants a choice of food, either norm-congruent or -incongruent. The participants were asked to choose three
snacks they would want to eat tomorrow. They could choose between twelve different food items, four of which were unhealthy (i.e., crisps, donut, pink glazed cake, stroopwafel), moderately healthy (i.e., bapao, egg cake, gingerbread, yogurt with granola), or healthy options (apple, tangerine, bell pepper, tomato). Unhealthy, moderately healthy, and healthy options were coded as ‘0’, ‘1’, and ‘2’, respectively. Each participant was credited a sum score based on their choices, ranging from ‘0’ till ‘6’. Low scores would express an anticipated choice for very unhealthy food items, and high scores would express a very healthy anticipated food choice. On average, most participants chose moderately healthy food items compared to healthy and unhealthy ones (M = 3,47, SD = 1,27).
Intention measure. The intention measure included three items that measured the participant’s evaluation of statements that were congruent with the social norm, and measure intention towards buying and eating fruits and vegetables, being: “I like to eat fruit or vegetables” (M = 6,18, SD = 1,12); “I would be willing to buy fruit or vegetables, even if they cost more than candy or crisps.” (M = 5.89, SD =1,22); “I think it is good to eat fruit and vegetables.” (M = 6,75, SD = 0,66). The intention measure also included three items that measured the evaluation of statements that were incongruent with the social norm (reverse coded), measuring intention towards buying and eating candy and snacks, being: “I like to eat candy or snacks.” (M = 5,33, SD = 1,32); “I would be willing to buy candy or snacks, even if they cost more than fruits and vegetables.” (M =4,04, SD = 1,69); “I think it is good to eat candy and snacks.” (M = 2,73, SD = 1,24). All items were measured on a 7-point Likert scale (1 = completely disagree, 7 = completely agree). On average, participants showed a stronger intention towards eating and buying fruits and vegetables than candy and crisps (M = 5,12, SD
= 0,68), and indicated to agree more with statements congruent with the social norm (M = 6,27, SD = 0,83) than with statements incongruent with the social norm (M = 4,03, SD = 1,08).
Message processing. This study measures the depth of message processing by employing a thought-listing exercise (Cacioppo et al., 1997). After being exposed to a stimulus, individuals are asked to list their thoughts. The number of reported topic-related thoughts was used to assess the amount of cognitive processing provoked by the stimulus. In this experiment, participants were asked which thoughts came to mind when reading the FOILA article.
Subsequently, they were instructed to recall those thoughts and write each down in a separate textbox, with a maximum of ten thoughts. In total participants recorded 794 thoughts (M = 4.36, SD = 2.43). The outcome was coded following the procure of Shen and Seung (2018), consisting of three steps. (1) Omitting emotional responses from the data. A response was coded as emotional when it referred to the participants’ state with a word associated with emotion (Shaver et al., 1987), (n = 10, M = 0.06, SD = 0.23). (2) Excluding responses not related or relevant to the article’s topic. The responses were deemed relevant to the topic when they referred to any topic discussed in the article. Any remarks regarding the readability (e.g., syntactic, grammar, clarity) and not the content of the message were coded as not relevant.
These responses were evoked by message variables irrelevant to the experiment and not intentionally manipulated. Thus, decreasing the measurement’s validity. In total participants recorded 569 relevant thoughts (M = 3.13, SD = 1.91). (3) Coding the responses as evaluative or neutral and omitting neutral responses. This was done because evaluative thoughts better represent a greater depth of processing than non-evaluative or neutral thoughts, and such cognitive response coding is most common when measuring the depth of message processing (e.g., Jones et al., 2003; Maheswaran & Meyers-Levy, 1990; Millar & Millar, 2000; Shen &
Dillard, 2009; Shen & Seung, 2018). Responses were coded as evaluative when expressing a positive or negative sentiment regarding the message or social norm and coded as neutral when not expressing an evaluation. In total, participants recorded 507 neutral thoughts (M = 2.79, SD
= 3.38) and 108 evaluative thoughts (M = 0.59, SD = 0.84).
19 Manipulation checks
The experiment implemented a manipulation check consisting of six items to measure whether both independent variables were successfully manipulated. The manipulation check was implemented after all other variables were measured.
Two items measured the participants’ injunctive norm perception: (1) ‘Most students at the UvA will approve when I eat fruits or vegetables, when I fancy a snack.’ (M = 4.19, SD = 0.93); and (2) ‘Most students at the UvA will disapprove when I eat candy or crisps, when I fancy a snack.’ (M = 2.55, SD = 1.03). Both items were measured on a 5-point Likert scale (1
= strongly disagree, 5 = strongly agree).
Four items measured the participants’ descriptive norm perception: (1) ‘How often do you think students at the UvA eat fruits or vegetables, when they fancy a snack?’ (M = 3.01, SD = 0.75); (2) ‘How often do you think students at the UvA eat candy or crisps, when they fancy a snack?” (M = 2.90, SD = 0.81); (3) ‘According to you, what is the percentage of UvA students that eat fruits or vegetables, when they fancy a snack?’ (M = 52.61, SD = 16.67); and (4) ‘According to you, what is the percentage of UvA students that eat candy or crisps, when they fancy a snack?’ (M = 46.88, SD = 17.02). The first two items were measured on a 5-point Likert scale (1 = never, 5 = always). The last two items asked participants to express their answers numerically.
Plan of analysis
The experiment included a Chi-square test and a one-way analysis of variance to test whether the participants were randomly assigned and equally distributed among all four conditions. Accordingly, the analysis did not include age and gender as control variables. The analysis included an independent samples t-test to assess whether the independent variable (i.e., social norm) was manipulated successfully. Homogeneity of variance was tested via Levene’s F test. The results were reported only when the p-value was significant, and thus homogeneity
was not satisfied. Finally, two two-way analysis of variance tests were performed to test the hypotheses.
Concerning the data for the two-way ANOVA, all assumptions were met or tested.
Outliers were assessed by inspection of a boxplot, and normality was assessed using Shapiro- Wilk’s normality test for each cell of the design and homogeneity was assessed by Levene’s Test. There were no outliers, assessed in a boxplot as being greater than three box-lengths from the edge of the box. The Shapiro-Wilk test determined that neither of the dependent variables’
data (i.e., evaluative thoughts, anticipated food choice) was distributed normally in any of the four conditions. However, the sample size per condition was larger than 20, sampling distribution was thus assumed to be normally distributed. The results concerning evaluative thoughts and anticipated food choice were as follows, respectively: condition one W (46) = 0.63, p < .001; W (46) = 0.92, p = .004, condition two W (43) = 0.79, p < .001; W (43) = 0.94, p = .020, condition three W (44) = 0.73, p < .001; W (44) = 0.94, p = .025 and condition four W (49) = 0.62, p < .001; W (49) = 0.93, p = .004. As assessed by Levene’s test for equality of error variance, there was homogeneity of variances evaluative thoughts F (178) = 0.394, p = .758 and anticipated food choice F (178) = 2.054, p = .108.
A randomization check was performed to assess whether the demographic variables (i.e., age, gender) had to be considered control variables. First, a Chi-Square test was performed to see whether the participants were distributed equally among conditions regarding their gender. The results (χ2 (6,182) = 5.00, p = .544) showed that gender did not differ significantly between conditions. Secondly, a one-way ANOVA was conducted to see whether the participants were distributed equally among conditions regarding age. The results (F (13,168)
= 1.54, p = .108) showed no significant differences between conditions regarding participants’
age. Based on these results, participants were determined to be equally distributed (see Table E1). Accordingly, age and gender were not included as control variables.
An independent samples t-test was conducted to compare the outcome on four manipulation check items in injunctive social norm and descriptive social norms conditions and to check if the manipulation of the social norm was successful.
Injunctive norm perception. Concerning participants’ perceived approval rate among UvA students of eating healthy snacks, there was no significant difference in the scores in the injunctive social norm (M = 4.20, SD = 0.88) and descriptive social norm (M = 4.26, SD = 0.86) conditions, t (180) = -0.433, p = .666, Mdif = -0.06. Participants’ perceived disapproval rate among UvA students of eating unhealthy snacks, did also not differ significantly between the injunctive social norm (M = 2.46, SD = 1.00) and descriptive social norm (M = 2.66, SD = 1.04) conditions, t (180) = -1.29, p = .198, Mdif = -0.20.
Descriptive norm perception. Concerning participants’ perception of how often UvA students ate healthy snacks, homogeneity of variance was tested but not satisfied via Levene’s F test, F (178.44) = 5.244, p = .023. There was a significant difference in the scores for injunctive social norm (M = 2.88, SD = 0.65) and descriptive social norm (M = 3.18, SD = 0.75) conditions, t (178.44) = -2.94, p = .004, Mdif = -0.31. Concerning participants’ perception of how often UvA students ate unhealthy snacks, there was a significant difference in the scores for injunctive social norm (M = 3.07, SD = 0.77) and descriptive social norm (M = 2.75, SD = 0.79) conditions, t (180) = 2.73, p = .007, Mdif = 0.12.
Regarding the participants’ perception of the percentage of UvA students eating healthy snacks, there was a significant difference in the scores for injunctive norms (M = 48.52, SD = 14.39) and descriptive norms (M = 56.53, SD = 17.79) conditions, t (180) = -3.33, p = .001, Mdif = -8.01. The participants’ perception of the percentage of UvA students eating unhealthy
snacks, homogeneity of variance was tested but not satisfied via Levene’s F test, F (179.93) = 1.500, p = .222. There was a significant difference in the scores for injunctive norms (M = 50.44, SD = 16.49) and descriptive norms (M = 43.47, SD = 16.90), t (197.93) = 2.81, p = .005, Mdif = 6.97. Based on these results the manipulation of the descriptive and injunctive norm was deemed successful.
Message processing. To test H1, a two-way analysis of variance was run with social norm (i.e., injunctive, descriptive) and framing (i.e., positive, negative) as independent variables and number of evaluative thoughts as the dependent variable. It was expected that negatively framed social norm messages would result in more thoughts than positively framed social norm messages. The two-way ANOVA did not show a statistically significant main effect of framing on the number of thoughts, F (1, 178) = 0.14, p = .712, η2 < .001. Participants who read a negatively framed article reported the same number of thoughts (M = 0.57, SD = 0.79) as those participants who read a positively framed article (M = 0.62, SD = 0.89). In addition, the results did not show a significant main effect of social norm (F (1,178) = 1.13, p = .290, η2
= .006) on the number of thoughts. However, unexpectedly, the results did show a marginally significant interaction effect of framing and social norm (F (1,178) = 4.01, p = .047, η2 = .022) on the reported number of evaluative thoughts. Approximately two percent of the variation in the reported evaluative thoughts could be attributed to this effect; this is a small effect size.
Participants in the negatively framed injunctive social norm condition (M = 0.77, SD = 0.87) and in the positively framed descriptive social norm condition (M = 0.68, SD = 0.93) reported on average more evaluative thoughts, compared to participants in the positively framed injunctive social norm condition (M = 0.57, SD = 0.86) and the negatively framed descriptive social norm condition (M = 0.39, SD = 0.67). Based on the abovementioned findings (see Table E2 & E3), H1 was not supported.
Behavioral measure – Anticipated food choice. To test H2 and H3, a two-way analysis of variance was run with social norm (i.e., injunctive, descriptive) and framing (i.e., positive, negative) as independent variables and participants’ anticipated food choice as the dependent variable. Negatively framed injunctive social norm messages and positively framed descriptive social norm messages were expected to result in healthier anticipated food choices, compared to positively framed injunctive social norms and negatively framed descriptive social norms, respectively. The two-way ANOVA showed no statistically significant interaction effect of social norm and framing on anticipated food choice, F (1,178) = 1.04, p = .310, η2 = .006.
Participants who read a negatively framed injunctive social norm did not report a healthier anticipated food choice (M = 3.33, SD = 1.25) than participants exposed to a positively framed injunctive social norm message (M = 3.63, SD = 1.36). Likewise, participants who read a positively framed descriptive social norm message did not report a healthier anticipated food choice (M = 3.39, SD = 1.28) than participants exposed to a negatively framed descriptive social norm message (M = 3.47, SD = 1.24). In addition, there was neither a main effect of framing F (1,178) = 0.34, p = .561, η2 < .005, nor of social norm F (1,78) = 0.07, p = .793, η2
< .001. On average, participants who read a positively framed norm message (M = 3.51, SD = 1.32) or a negatively framed norm message (M = 3.40, SD = 1.24) reported a moderately healthy anticipated food choice. Participants who read an injunctive norm message (M = 3.48, SD = 1.30) or a descriptive norm message (M = 3.43, SD = 1.25), reported a moderately healthy anticipated food choice. Based on the abovementioned findings (see Table E4 & E5), H2 and H3 were not supported.
Intention measure – Relevant thoughts. Additionally, a two-way analysis of variance was run with social norm (i.e., injunctive, descriptive) and framing (i.e., positive, negative) as independent variables and all topic-related thoughts as the dependent variable. Meaning
evaluative and neutral thoughts (i.e., non-evaluative) were grouped as the dependent variable.
The two-way ANOVA did not show a statistically significant main effect of framing on the number of topic-related thoughts, F (1, 178) = 0.12, p =.729, η2 < .005. In addition, the results did not show a significant main effect of social norm (F (1,178) = 0.05, p = .821, η2 < .001), nor a significant interaction effect of framing and social norm (F (1,178) < 0.01, p = .996, η2
< .001) on the reported number of topic-related thoughts.
Intention measure. Additionally, a two-way analysis of variance was run with social norm (i.e., injunctive, descriptive) and framing (i.e., positive, negative) as independent variables and intention as dependent variable. The two-way ANOVA did not show a statistically significant main effect of social norm on intention, F (1, 178) = 2.59, p =.109, η2 = .014. In addition, the results did not show a significant main effect of framing (F (1,178) = 1.09, p = .299, η2 = .006), nor a significant interaction effect of framing and social norm (F (1,178) = 1.64, p = .202, η2 = .009) on the reported number of topic related thoughts.
The study aimed to document the differing persuasive effects of framing two types of social norm messages and understand whether differing levels of message processing could explain this persuasive effect. Specifically, the study tested whether a positively framed descriptive social norm message and a negatively framed injunctive social norm message showed superior effects over a positively framed injunctive social norm message and a negatively framed descriptive social norm message on a behavioral task (i.e., food choice).
Subsequently, the effects of framed social norm messages on a thought listing exercise were compared.
The results did not show a significant main effect of framing on message processing.
Nor did the results show direct evidence to corroborate whether different levels of processing explained the persuasive effects of framing social norms. This outcome was comparable to
Mollen and colleagues’ (2016) results and might have a similar explanation: the absence of framing effects was due to methodological grounds. Although more implicit than self-report items, the thought-listing exercise could be lacking precision in its current state. Specifically, the coding of the thoughts in this experiment might have had two errors. (1) When using thought-listing data to infer cognitive representations and processes, it might prove insightful to go beyond just scoring and comparing the number of relevant thoughts and distinguish evaluative thoughts (Cacioppo et al., 1997). However, when trying to measure and compare the depth or amount of processing, excluding non-evaluative thoughts from analysis resulted in a less valid measurement because excluded non-evaluative thoughts still express message processing. Secondly, treating both types of evaluative thoughts (i.e., positive, negative) as equivalent and interchangeable, rather than separating them and comparing their ratio, could have misrepresented the amount of message processing (Cacioppo et al., 1997). Therefore, the reported evaluative thoughts might have been unsuitable to represent the possibly different levels of message processing, as elicited by framing.
However, there might be an alternative explanation for the lack of a significant main effect of framing. Namely, framing did not directly influence message processing. Analysis showed a (marginally) significant interaction effect of norm and frame on evaluative thoughts.
These results infer that the type of norm and how this norm is presented interact and affect the depth of message processing. The interaction seems to operate in the same pattern as the proposed interaction of Mollen and colleagues (2016). Being that negatively framed injunctive and positively framed descriptive norms, outperformed positively framed injunctive and negatively framed descriptive norms. This seems to indicate that norm and frame can indeed be combined to enhance a message outcome (i.e., processing). Furthermore, this inferred interaction effect does not wholly stand in line with the assumption that framing alone elicits differing levels of processing (Maheswaran & Meyers-Levy, 1990; Smith & Petty, 1996).
These results do not necessarily break with the assumption that negative framing, compared to positive framing, elicits more processing (Baumeister et al., 2001; Maheswaran & Meyers- Levy, 1990; Smith & Petty, 1996). However, if message processing was influenced by an interaction effect of norm and frame, rather than by framing alone, it could clarify why previous research was not able to definitively conclude which message frame increased message processing (Gallagher & Updegraff, 2011; Rothman & Updegraff, 2010; O’Keefe & Jensen, 2008). Still, there was no significant framing effect on message processing; thus, hypothesis one was rejected. The results also did not show a significant interaction effect of norm and frame on participants’ anticipated food choice. Based on this, hypotheses two and three are also rejected.
Concerning the research question, based on these results, differing persuasive effects cannot be conclusively attributed to differing levels of message processing. Nor can the differing persuasive effects of framing descriptive and injunctive social norms be described.
Implications for future research
As mentioned before, the thought-listing exercise, measuring the number of reported thoughts, was used to indicate message processing. Even though less explicit than the measurement used by Mollen et al. (2016), the current measurement and coding showed room for improvement. First, when coding the responses, all relevant thoughts should have been included for analysis. Secondly, the topic-relevant thoughts (i.e., positive, negative, and non- evaluative), should have been analyzed as three separate factors.
Current findings raise questions concerning whether message processing is directly influenced by frame. The results indicate an interaction effect of norm and frame on message processing. This is an interesting finding because it was expected that processing would mediate the effects of framing on social norms, rather than processing being affected by frame and norm. It might be a fruitful avenue of future research to test whether norm and frame have
an interaction effect on message processing. The outcome of such research could contribute to a better understanding of how message processing might be an underlying factor explaining the varying persuasive effects of framing on social norms.
The study could not differentiate between the supposedly different effects of framing on injunctive and descriptive norm messages. Likewise, the experiment could not indicate which role message processing played in explaining these framing effects. However, findings suggest an interaction effect of social norms and message framing on message processing. A possible explanation for this could be that message processing does play a role in explaining the varying persuasive effects of framed social norm messages, only not as expected. The results might also indicate that social norms do more than require processing recourses. Rather than processing being elicited by message frame, it might be a combination of frame and norm that facilitate processing. It remains unclear if enhanced levels of processing result in a stronger adherence to norms advocated in the message. More research is required to understand the role that processing plays in understanding the effects of framed social norms messages on food choice.
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5(4), 323-370. doi:10.1037/1089- 2622.214.171.1243
Cacioppo, J. T., Von Hippel, W., & Ernst, J. M. (1997). Mapping cognitive structures and processes through verbal content: The thought-listing technique. Journal of
Consulting and Clinical Psychology, 65(6), 928-940. doi:10.1037/0022- 006x.65.6.928
Centraal Bureau voor de Statistiek. (2019, April 10). Kwart 18- tot 25-jarigen te zwaar.
Retrieved January 16, 2022, from https://www.cbs.nl/nl-nl/nieuws/2019/15/kwart-18- tot-25-jarigen-te-zwaar
Cialdini, R. B., Demaine, L. J., Sagarin, B. J., Barrett, D. W., Rhoads, K., & Winter, P. L.
(2006). Managing social norms for persuasive impact. Social Influence, 1(1), 3-15.
Cialdini, R. B., Kallgren, C. A., & Reno, R. R. (1991). A focus theory of normative conduct:
A theoretical refinement and reevaluation of the role of norms in human behavior.
Advances in Experimental Social Psychology, 24, 201-234. doi:10.1016/s0065- 2601(08)60330-5
Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct:
Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58(6), 1015-1026. doi:10.1037/0022- 35126.96.36.1995
Cialdini, R. B., & Trost, M. R. (1998). Social influence: Social norms, conformity, and compliance. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (pp. 151-192). New York, NY: McGraw-Hill.
Cohen, A. R., Stotland, E., & Wolfe, D. M. (1955). An experimental investigation of need for cognition. The Journal of Abnormal and Social Psychology, 51(2), 291-294.
De Vos, P., Hanck, C., Neisingh, M., Prak, D., Groen, H., & Faas, M. M. (2015). Weight gain in freshman college students and perceived health. Preventive Medicine Reports, 2, 229-234. doi:10.1016/j.pmedr.2015.03.008
Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influences upon individual judgment. The Journal of Abnormal and Social Psychology, 51(3), 629-636. doi:10.1037/h0046408
Dijksterhuis, A., & Aarts, H. (2003). On wildebeests and humans. Psychological Science, 14(1), 14-18. doi:10.1111/1467-9280.t01-1-01412
Farrow, K., Grolleau, G., & Ibanez, L. (2017). Social norms and pro-environmental behavior:
A review of the evidence. Ecological Economics, 140, 1-13.
Gallagher, K. M., & Updegraff, J. A. (2011). 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
GBD 2015 Obesity Collaborators. (2017). Health effects of overweight and obesity in 195 countries over 25 years. New England Journal of Medicine, 377(1), 13-27.
Geuens, M., & De Pelsmacker, P. (2017). Planning and conducting experimental advertising research and questionnaire design. Journal of Advertising, 46(1), 83-100.
Hassell, H. J., & Wyler, E. E. (2018). Negative descriptive social norms and political action:
People aren’t acting, so you should. Political Behavior, 41(1), 231-256.
Herman, C. P., Roth, D. A., & Polivy, J. (2003). Effects of the presence of others on food intake: A normative interpretation. Psychological Bulletin, 129(6), 873-886.
Higgs, S. (2015). Social norms and their influence on eating behaviours. Appetite, 86, 38-44.
Jacobson, R. P., Mortensen, C. R., & Cialdini, R. B. (2011). Bodies obliged and unbound:
Differentiated response tendencies for injunctive and descriptive social norms.
Journal of Personality and Social Psychology, 100(3), 433-448.
Jones, L. W., Sinclair, R. C., & Courneya, K. S. (2003). undefined. Journal of Applied Social Psychology, 33(1), 179-196. doi:10.1111/j.1559-1816.2003.tb02078.x
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk.
Econometrica, 47(2), 263. doi:10.2307/1914185
Kredentser, M. S., Fabrigar, L. R., Smith, S. M., & Fulton, K. (2012). Following what people think we should do versus what people actually do. Social Psychological and
Personality Science, 3(3), 341-347. doi:10.1177/1948550611420481
Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76(2), 149-188. doi:10.1006/obhd.1998.2804
Maheswaran, D., & Meyers-Levy, J. (1990). The influence of message framing and issue involvement. Journal of Marketing Research, 27(3), 361. doi:10.2307/3172593 McEachan, R. R., Conner, M., Taylor, N. J., & Lawton, R. J. (2011). Prospective prediction
of health-related behaviours with the theory of planned behaviour: A meta-analysis.
Health Psychology Review, 5(2), 97-144. doi:10.1080/17437199.2010.521684 Melnyk, V., Herpen, E. V., Fischer, A. R., & Van Trijp, H. C. (2011). To think or not to
think: The effect of cognitive deliberation on the influence of injunctive versus descriptive social norms. Psychology and Marketing, 28(7), 709-729.
Millar, M. G., & Millar, K. U. (2000). Promoting safe driving behaviors: The influence of message framing and issue involvement. Journal of Applied Social Psychology, 30(4), 853-866. doi:10.1111/j.1559-1816.2000.tb02827.x
Morris, M. W., Hong, Y., Chiu, C., & Liu, Z. (2015). Normology: Integrating insights about social norms to understand cultural dynamics. Organizational Behavior and Human Decision Processes, 129, 1-13. doi:10.1016/j.obhdp.2015.03.001
O'Keefe, D. J., & Jensen, J. D. (2008). Do loss-framed persuasive messages engender greater message processing than do gain-framed messages? A meta-analytic review.
Communication Studies, 59(1), 51-67. doi:10.1080/10510970701849388 Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion.
Communication and Persuasion, 1-24. doi:10.1007/978-1-4612-4964-1_1
Rivis, A., & Sheeran, P. (2003). Descriptive norms as an additional predictor in the theory of planned behaviour: A meta-analysis. Current Psychology, 22(3), 218-233.
Rothman, A. J., & Updegraff, J. A. (2010). Specifying when and how gain-and loss-framed messages motivate healthy behavior: An integrated approach. In G. Keren (Ed.), Perspectives on framing (pp. 257-278). Psychology Press.
Rozin, P., & Royzman, E. B. (2001). Negativity bias, negativity dominance, and contagion.
Personality and Social Psychology Review, 5(4), 296-320.
Schokker, D. F., Visscher, T. L., Nooyens, A. C., Van Baak, M. A., & Seidell, J. C. (2007).
Prevalence of overweight and obesity in The Netherlands. Obesity Reviews, 8(2), 101- 107. doi:10.1111/j.1467-789x.2006.00273.x
Shaver, P., Schwartz, J., Kirson, D., & O'Connor, C. (1987). Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology, 52(6), 1061-1086. doi:10.1037/0022-35188.8.131.521
Shen, L., & Dillard, J. P. (2009). Message frames interact with motivational systems to determine depth of message processing. Health Communication, 24(6), 504-514.
Shen, L., & Seung, S. (2018). On measures of message elaboration in narrative communication. Communication Quarterly, 66(1), 79-95.
Slater, M. D. (2006). Specification and Misspecification of theoretical foundations and logic models for health communication campaigns. Health Communication, 20(2), 149-157.
Smith, S. M., & Petty, R. E. (1996). Message framing and persuasion: A message processing analysis. Personality and Social Psychology Bulletin, 22(3), 257-268.
World Health Organization. (1998). Obesity: Preventing and managing the global epidemic;
Report of a WHO consultation on obesity.
World Health Organization. (2000). Conclusion and recommendation. In Obesity: Preventing and managing the global epidemic (pp. 240-247).
Appendix A – Online experiment
Thank you for taking the time to participate in this research study. The research study will be conducted online. This research study is aimed specifically at Students at the University of Amsterdam (UvA), and who are able to read English at an University level.
Before you start this research study, it is important that you are aware of the procedure of this research. Please read the following instructions carefully. If any questions or concerns arise, please feel free to ask any questions or for clarification via firstname.lastname@example.org.
To benefit the validity of the research data, it is very important that you as a participant pay attention when participating in this research study. Therefore, a so-called attention check is included as one of the questions. If you do not pass this attention check your participation will be terminated and you cannot participate anymore.
1) Aim of the research
The goal of this research study is to understand people’s attitudes towards certain aspects of media messages concerning lifestyles.
During the research study you will be presented with a short text, after which several questions will be asked. The results and data of this research study will be completely anonymous, and your privacy is guaranteed. Participation in this survey will take about 10 minutes. Participants via the LAB, will be rewarded with credits.
3) Information regarding research
This research is performed under the responsibility of the Amsterdam school of
Communication research (ASCoR), which is a part of the university of Amsterdam (UvA).
Additionally, you reserve the following rights:
A) The data and results of this research study shall be used to answer the research question central to this research study as (briefly) described above. Your personal data and results might be subject to further analysis, only if this suits or serves the research study’s aim. Any and all research data or results published in scientific journals will be made completely anonymous and cannot be traced back to you as an individual. To conclude: research data that is completely anonymized, may be – for the purpose of scientific ends- be shared with other researchers. Research data that is completely anonymized might be published to the public.
B) If requested, a summary of the research data will be made available to you. In case you want to receive this summary please indicate so, via email and within three months after completion of the research study.
C) Terminating your participation with this research study, will not have any consequences for you personally. Ending your participation is possible at any given moment, without any form of consequences and without any statement of reasons.
D) If you have any questions or concerns, prior or post participating with, and regarding to this research study, please direct them at the Master thesis supervisor Saar Mollen
(email@example.com). In the event that you have any complaints regarding the research study or its researchers, please direct them at the Ethics commission of the ASCoR, via ascor-secr- firstname.lastname@example.org. Your remark and or complaint will be treated confidentially.
I hope to have fully and completely informed you.
- I hereby declare to by adequately and sufficiently informed as to the nature and method of upcoming research study, as described in the previous section.
- I agree to participate on a pure voluntarily basis. In doing so I remain the right to withdraw my consent at any given moment and without any given reason.
- Whilst participating in this research study the safeguarding of my privacy will be
guaranteed. All use of my personal data or answers shall be done in total anonymity. No third parties shall be able to access my data without my explicit permission.
- I fully understand the information presented above, and hereby give permission that my personal data may be used for this research study.
To download the Informed consent as a file, please click the link.
• I do not accept
• I accept
Please fill in the following demographic information 1. Gender
• Non-binary / third gender
• Prefer not to say
2. Please fill in your age in number
Participants that were not at least 16 years old, received the following message before being excluded from the study: