‘You should have 2 pieces of fruit each day!’ ‘Ehh, hell no!’
The effects of social norm messages, reactance to norm messages,
resistance to peers and fruit consumption among adolescents
Name: Anne van Leest
Student ID: 0435473
Supervisor: Saar Mollen
Inhoudsopgave
Abstract ... 3 Introduction ... 4 Theoretical background ... 8 Adolescence and social influences on eating behaviour ... 8 Descriptive and injunctive norms ... 9 Influence of social eating norms on food consumption ... 11 Feelings of reactance ... 14 Feelings of resistance to peers ... 17 Method ... 19 Participants & design ... 19 Procedure ... 23 Materials and manipulations ... 24 Measures ... 26 Results ... 29 Randomization check ... 29 Main analyses ... 30 Moderated mediation of reactance ... 31 Mediation by reactance ... 32 Moderation by age ... 34 Moderation by resistance to peers ... 35 Discussion ... 37 Intended and actual food consumption ... 38 Reactance to social norm ... 39 Age ... 40 Resistance to peers ... 40 Limitations of the present research ... 41 Conclusion ... 42 References ... 43 Appendix 1: informed consent ... 49 Appendix 2: Parent approval ... 50 Appendix 3: Debriefing ... 51 Appendix 4: Booklet without experimental manipulation ... 52 Appendix 5: No-norm control message ... 57 Appendix 6: Descriptive norm message ... 58 Appendix 7: Injunctive norm message ... 59 Appendix 8: Fruit diary without experimental manipulation ... 60 Appendix 9: Syntax ... 65Abstract
Objectives. The aim of the study was to analyse whether descriptive and injunctive social
norm messages affected adolescents’ intentions to consume fruit and actual fruit consumption. In addition, the possible mediating role of reactance towards social norm messages and the possible moderating role of adolescents’ age and resistance to peers were investigated.
Methods. A total of 286 secondary school students were randomly assigned to one of the
three different norm messages: descriptive social norm vs. injunctive social norm vs. no-norm control. Participants filled in a questionnaire about resistance to peers, reactance to the social norm message and intentions to consume fruit. At follow-up, actual fruit consumption over three days (N = 184) was reported.
Results. The analysis showed no significant main effect. There was no effect of message type
on participants’ reactance, intentions to consume fruit, and actual fruit consumption. Above that, there was no mediating effect of adolescents’ reactance to social norm message between the relation of social norm messages of actual fruit consumption. However, reactance did affect intended and actual fruit consumption. Also no moderating effects of age or resistance to peers were found on the relationship between social norm message and reactance to social norm messages. For age, results showed a significant effect of age between social norm type and reactance to social norms. For resistance to peers, results indicated that resistance to peers significant influenced reactance to social norm.
Conclusion. Although no main effects were found, the significant results that were found
emphasize the effect that social norms can have in promoting fruit behaviour. To fully understand the relation between social norms, reactance and (intended) behaviour, future studies should investigate the effects of social norm message on actual fruit consumption and the mediating role of reactance more in-depth.
Introduction
In many European countries, people do not meet the recommendations for fruit intake (Van Rossum, & Geurts, 2013). Not only adults, but also children and adolescents do not meet the nutrition guidelines (De Wind et al., 2006). Unhealthy eating practices can lead to health issues, such as cardiovascular diseases, obesity and different types of cancer (Holt et al., 2009; Kremers, Brug, De Vries, & Engels, 2003; Stok, Verkooijen, De Ridder, De Wit, & De Vet, 2014b).
The Netherlands Nutrition Centre1 recommends the Dutch population to eat at least two pieces of fruit each day (Van Rossum, & Geurts, 2013). However, recent research indicates that Dutch children between the age of 7 and 18 years old eat only 5% of the recommended amount of fruit (Van Rossum, Fransen, Verkaik-Kloosterman,
Buurma-Rethans, & Ocke, 2011). Although the eating behaviour of Dutch adolescents is healthier than ten years ago, it is still a matter of concern (Van der Klauw, Van Keulen, & Verheijden, 2011).
Research demonstrates that adolescents know perfectly well that healthy eating is important and what healthy eating constitutes, for example eating enough fruit each day (Brown, McIlveen, & Strugnell, 2000; Croll, Neumark-Sztainer, & Story, 2001; Stevenson, Doherty, Barnett, Muldoon, & Trew, 2007). In other words, adolescents know very well that they have to eat enough fruit, but do not behave that way. To encourage adolescents to eat more fruit, it is important to understand what hinders their daily fruit consumption. Croll and
1 The Netherlands Nutrition Centre (Voedingscentrum) provides consumers information about a healthy food
choices and encourages them to make and maintain healthy eating habits. The Nutrition Centre is funded by the Ministry of Public Health, Welfare & Sport and the Ministry of Economic Affairs, Agriculture & Innovation (Van Rossum, & Geurts, 2013).
colleagues (2001) and Stok, De Vet, De Ridder and De Wit (2012) state that these hindering factors may include social norms that do not support healthy eating.
Social norms are shared, unwritten rules for behaviour in particular situations that are accepted by a majority of a group (Cialdini, 2003; Guerra, Huesmann, & Hanish, 1995). Moreover, our own behaviours and actions are guided by other people’s behaviour and their opinion about that behaviour, because it provides information on how others behave in a certain situation and what is considered as appropriate or inappropriate behaviour (Cialdini, & Goldstein, 2004). Research has distinguished two types of social norms: descriptive norms and injunctive norms (Cialdini, Reno, & Kallgren, 1990). Descriptive norms refer to what is commonly done by others in a specific situation without assigning judgments. In other words, it describes the behaviour of others and as such indicates what is the normal behaviour. Injunctive norms refer to what behaviour is commonly approved or disapproved by others, because it dictates or proscribes how someone should behave in a given situation (Kallgren, Reno, & Cialdini, 2000). It is important to distinguish between descriptive and injunctive norms, as they are hypothesized to influence behaviour through different routes (Cialdini, & Trost, 1998), respectively through informational and normative social influence. When conforming to descriptive norms, individuals want to have information about the actual behaviour of others, whereas individuals attempt to gain social approval and to avoid social disapproval when conforming to injunctive norms (Cialdini et al., 2003).
Both types of social norms can affect healthy and unhealthy eating behaviour (Robinson, Fleming, & Higgs, 2014). Moreover, perceived social norms predict regular vegetable and fruit consumption. For example, a teenager who perceives her friends to think she should eat fruit (an injunctive norm), is expected to be influenced by this and will eat more fruit (Lally, Bartle, & Wardle, 2011). However, research indicates that descriptive norms are more strongly related to eating behaviour than injunctive norms (Cruwys,
Bevelander, & Hermans, 2015; Mollen, Rimal, Ruiter, & Kok, 2013; Robinson et al., 2013; Stok, De Ridder, De Vet, & De Wit, 2014a). Cruwys and colleagues (2015) conclude in their review about the effect of social influence on food choice and consumption that the few studies that have examined the influence of different types of social norms on food consumption, support the centrality of descriptive norms and may clarify the specific normative content participants attend to. For example, Mollen and colleagues (2013) found that healthy food choices were affected by social norm messages. The health-promoting descriptive and injunctive norm message had a, respectively significant and non-significant, positive effect on food choice relative to the unhealthy descriptive norm message. However, participants who read an injunctive norm message, did not make significantly more healthy food choices than participants in the no-norm condition. Furthermore, Stok and colleagues (2014a) showed that health-promoting injunctive norms had no positive effects on fruit consumption, in fact, it caused a decrease in intentions to consume fruit, relative to health-promoting descriptive norms.
According to Stok and colleagues (2014a), this decrease in participants’ intentions to consume fruit suggests that injunctive norms may cause reactance, because injunctive norms can be perceived to restrict one’s freedom of thinking and acting. Reactance toward a
message arises when an individual feels restricted in his or her freedom of choice, i.e., when one feels like being pushed in a certain direction, especially when this direction is not in line with personal goals (Grandpre et al., 2003). To restore this freedom of choice, an individual will behave going in the opposite direction of that is desired by the social group (Rummel, Howard, Swinton, & Seymour, 2000).
Previous research indicates that healthy eating is not very high on adolescents’ list of personal goals (Croll et al., 2001; De Wind et al., 2006; Van Rossum et al., 2011). Then, an injunctive norm to eat more fruit may be received as not in line with personal goals, and
therefore, may cause reactance to the social norm. This may subsequently cause adolescents to have lower intentions to consume fruit (Stok et al., 2014a). Because there is not yet any scientific evidence to support this supposition, the goal of the present study is to investigate the effects of reactance occurring from providing descriptive and injunctive norms that promote fruit consumption on secondary school students’ intended and actual fruit
consumption. Additionally, in this study, two age groups will be compared: a group of 11- to 12-year-olds and a group of 16- to 17-year-olds. This comparison between age groups will be made, because during adolescence, adolescents become more independent from their parents and friends and, thereby, become less likely to follow their parents and peers, and will resist their influence more often, increasing with age (Collins & Laursen, 2004; Sumter, Bokhorst, Steinberg, & Westenberg, 2009; Walker, & Andrade, 1996). As a result, social peer norms may have less influence on fruit consumption in late adolescence than in early adolescence.
However, the influence of age and social peer resistance on the relation between social norm message, reactance and actual fruit intention and consumption is yet tested empirically. In order to encourage adolescents to eat more fruit, it is important to understand how social norms can hinder adolescents’ fruit consumption. Therefore, in the current study, the effects of health-promoting descriptive and injunctive social norms on adolescents’ intended and actual fruit consumption will be investigated. The research questions that will be addressed in the present study are: To what extent do descriptive and injunctive health-promoting peer fruit
consumption norms have influence on adolescents’ intended and actual fruit consumption? And to what extent is this relation mediated by adolescents’ reactance towards these social norms? And to what extent is this mediation moderated by adolescents’ age and resistance to peers?
Theoretical background
Adolescence and social influences on eating behaviour
Research has consistently shown that eating behaviour is affected by the food choices of others in our social environment (Higgs, 2015; Robinson & Higgs, 2013). This type of behaviour can be explained by a mechanism that plays an important role in eating behaviour: social norms.
Social norms are shared, unwritten rules for behaviour in particular situations that are to be considered as the standard of a social group (Cialdini, 2003; Guerra et al., 1995). Therefore, our own behaviour and actions are guided by other people’s behaviour, because it provides information on how others behave and what is considered appropriate or
inappropriate behaviour in a certain situation (Cialdini, & Goldstein, 2004). Various studies confirm the suggestion that one’s behaviour is guided by other people’s behaviour (Hermans et al., 2012, Robinson et al., 2014). For example, in a social eating context, people try to match their food consumption to that of their eating companion. Research in laboratory setting has shown that participants match their food consumption with the perceived food consumption of other participants. When participants believed that the other participants ate a small amount of junk food, they decreased their own intake of junk food. Reversely, when participants believed that the other participants ate a large amount of junk food, they
increased their own intake of junk food (Hermans et al., 2012; McFerran, Dahl, Fitzsimons, & Morales, 2010; Robinson et al., 2014; Tanner, Ferraro, Chartrand, Bettman, & Van Baaren, 2008).
For example, Hermans and colleagues (2012) found that women dyads mimicked each other’s eating behaviour. The women were more likely to eat their evening meal in
congruence with their eating partner rather than eating at their own pace. This result was found to be more salient at the start of the evening meal than at the end of it. Therefore,
Hemans and colleagues argue that mimicry may partially account for social modelling of food consumption. In line with these findings, Robinson and colleagues (2011) showed that
individuals may copy the behaviours of people around them. They also found that individuals do that in order to facilitate social interactions and to be liked by others. Robinson and
colleagues state that these findings suggest that social matching of food consumption occurs because of individuals’ desire for social acceptance. Outside laboratory setting, research has shown that habitual food consumption, including fruit and vegetable and snack food
consumption of an individual can be predicted by the food consumption of the social group that individual belongs to (Lally, Bartle, & Wardle, 2011).
The review Stok, De Vet, De Ridder, and De Wit (2016) has shown that 32 out of 33 studies, both correlational and experimental, can influence young people’s food consumption. Their findings suggest that there is a significant relation between social norms and food consumption. Above that, all experimental studies in the review suggested an effect of social norms on food consumption. In conclusion, research showed that social eating norms have an effect on individuals’ actual eating behaviour.
Descriptive and injunctive norms
Social norms result from group members’ practices (i.e., descriptive norms) or the group members’ expectations (i.e., injunctive norms; Stok et al., 2016). The distinction between these two types of social norms is proposed in the Focus Theory of Normative Conduct (FTNC; Cialdini et al., 1990). Descriptive norms refer to what is commonly done by others in a specific situation. In other words, it describes the behaviour of most others and as such it indicates what is considered normal conduct of a social group in a certain situation (Cialdini & Trost, 1998). For example, most adolescents eat at least 2 pieces of fruit each day (Stok et al., 2016). Injunctive norms on the other hand refer to what behaviour is commonly approved
or disapproved of by others in a certain social group, it dictates how someone should behave in a given situation according to most others (Kallgren et al., 2000). For example, most adolescents think their peers should eat enough fruit (Stok et al., 2016). While descriptive norms describe what the behaviour of group members actually is, injunctive norms proscribe how other members ought to behave according to most others.
It is important to distinguish between descriptive and injunctive norms, as they are hypothesized to influence behaviour through different routes (Cialdini, & Trost, 1998), respectively through informational and normative social influence (Cialdini, 2003). When conforming to descriptive norms, individuals want to gain certainty about making the right decision (Cialdini et al., 1990; Cialdini, & Trost, 1998). When other individuals make a certain decision, it provides social proof on what is the right decision to make (Cialdini, 2003; Jacobson, Mortensen, & Cialdini, 2011). The rationale here is if more others are doing it, it must be the right thing to do (Cialdini et al., 1990; Cialdini, & Trost, 1998). This line of thought is even more applicable to an ambiguous, uncertain or unfamiliar situation, because in these situations individuals are very uncertain about what to do. As a result, they are more likely to depend on how others behave (Deutsch, & Gerard, 1955). Individuals’ desire to respond appropriately to a social situation demands an accurate perception of reality. The need to interpret and react to incoming information is of importance (Cialdini, & Goldstein, 2004). In summary, individuals want to have information about the actual behaviour of others. This kind of influence on behaviour is called informational social influence (McDonald, & Crandall, 2015). However, when individuals conform to injunctive norms, their behaviour is influenced through normative social influence, because their desire is to be liked or accepted by members of their social group. This means that individuals attempt to gain social approval and to avoid social disapproval (Stok et al., 2016). As a result, injunctive norms lead to conformity, because of the desire to gain and maintain social relationships and to avoid the
risk of negative social sanctions (Cialdini, & Goldstein, 2004; Cialdini et al., 1990; Deutsch, & Gerard, 1955). The rationale here is ‘if we do what others approve of they must approve of us too’ (Cialdini et al., 1990; Cialdini, & Trost, 1998). Because adolescents highly value belonging to a social group, they tend to behave following the standards of that group (Stok et al., 2014a). The need of social approval and acceptance is called normative social influence (McDonald, & Crandall, 2015).
Influence of social eating norms on food consumption
Stok and colleagues (2016) did a systematic review to assess the role of social peer norms on food consumption. Of the 33 studies, of which 17 correlational and 16 experimental, they reviewed, all except one found significant relations between social norms and food
consumption. All experimental studies found effects of social norms on food consumption. This review showed the influence of descriptive and injunctive social eating norms on (healthy) food consumption show that both social norms can effect food intention and consumption. However, descriptive and injunctive norms do not always influence behaviour in the same direction and to same extent (Cialdini et al., 1990). Above that, various research does not include the injunctive norm message, and, therefore, only focussing on the effect of descriptive norm message (Burger et al., 2010; Stok et al., 2012; Stok et al., 2014a). For example, Burger and colleagues (2010) studied the effect of descriptive norm information on women’s healthy and unhealthy snack choices. Participants tended to choose a snack
consistent with the information of the descriptive norm, i.e., with what they believed others had chosen. Stok and colleagues (2012) also investigated the effects of descriptive norm information on fruit consumption. For their experiment, they used two types of descriptive norms: a minority (‘only 27% of Dutch university students eat sufficient fruit’, p. 960) and a majority (‘73% of Dutch university students eat sufficient fruit’, p. 960) normative norm.
Results showed that minority norms led to lower intentions to consume fruit and to lower fruit consumption when participants strongly identified with the referent group relative to the majority norm. Another experiment of Stok and colleagues (2014b) presented the same findings: participants, when strongly identified with the referent group, who read a majority descriptive norm ate more vegetables than those who read the minority norm. The three studies show that descriptive norms can have an influence on people’s food choices and consumption.
Some research included injunctive social eating norms as well, that allows researchers to compare the effects of descriptive and injunctive social norms on (intended) healthy food consumption (Lally et al., 2011; Mollen et al., 2013; Robinson et al., 2014; Stok et al., 2014a). Lally and colleagues (2011) used questionnaires to examine associations between social norms for food consumption (descriptive and injunctive) and dietary behaviour among school pupils aged 16 to 19 years old. Lally and colleagues found that descriptive norms had a strong influence on adolescents’ food consumption, including fruit and vegetables, sugar-sweetened drinks and unhealthy snacks. Furthermore, there was no indication of a relation between the injunctive norm and food consumption. Robinson and colleagues (2013) also investigated the effect of descriptive and injunctive norms on fruit consumption. They compared the effect on young women’s food choice of a health message, a descriptive norm message and an injunctive norm message, all concerning fruit and vegetable consumption. Their two laboratory studies showed that participants who read the descriptive norm message selected and consumed significantly more fruit and vegetables and less high energy snacks relative to those who read the health message. Similar to the findings of Lally and colleagues (2011), Robinson and colleagues did not find a significant effect of the injunctive norm on participants’ food choice and consumption. In sum, these studies showed no relationship between injunctive norms and behaviour
Outside laboratory settings, experimental research also showed the influence of injunctive norms (Mollen et al., 2013; Stok et al., 2014a). In a field-experiment in an on-campus food court, Mollen and colleagues (2013) investigated students’ healthy food choices by comparing three different social norms, an unhealthy descriptive norm, a healthy
descriptive norm and a healthy injunctive norm, with a no-norm condition. The experiment showed that healthy food choices were affected by social norm messages. The health-promoting descriptive and injunctive norm message had a, respectively significant and non-significant, positive effect on food choice relative to the unhealthy descriptive norm message. However, participants who read an injunctive norm message, did not make significantly more healthy food choices than participants in the no-norm condition. Stok and colleagues (2014a) also showed in an experiment that health-promoting norms had an effect on intended and actual fruit consumption. Intentions to consume fruit were similar for adolescents’ who read the descriptive norm and no-norm messages, but were significantly lower for those who read the injunctive norm message. Furthermore, participants’ who read the descriptive norm message consumed significantly more fruit than those who read either the injunctive or no-norm messages. In other words, adolescents who read the injunctive no-norm message reported lower intentions to consume fruit, but their actual fruit consumption was similar to that of the no-norm condition. Contrary to findings of Lally and colleagues (2011) and Robinson and colleagues (2013), research of Mollen and colleagues (2013) and Stok and colleagues (2014a) showed that injunctive norms can have an impact on healthy food choice and consumption. Based on these findings, the following is hypothesized:
H1: Descriptive norm messages will have more influence on adolescents’ intentions to consume fruit and actual fruit consumption than injunctive norm and no-norm messages.
Feelings of reactance
Based on the findings that injunctive social norm messages caused a decrease in adolescents’ intentions to consume fruit compared to descriptive social norm messages and no-norm messages, which had no significant effect on fruit consumption intentions, Stok and colleagues (2014a) argue that injunctive norms may cause reactance.
Brehm (1966) introduced the theory of psychological reactance. This theory explains resistance to persuasion (Grandpre et al., 2003). According to the theory of psychological reactance, reactance is a motivational reaction to threats of behavioural freedom (Brehm, 1966). For instance, when the social norm is not to consume unhealthy food, individuals may consider this as a limitation of their possible choices (Stok, De Vet, De Wit, Renner, & De Ridder, 2015). Threats to their sense of freedom will likely be perceived as undesirable and motivate individuals to restore their self-determination and freedom of choice (Grandpre et al., 2003; Stok et al., 2015). That is when individuals will have feelings of reactance.
According to Brehm (1966), individuals who enter a state of reactance search to participate in behaviours that will re-establish their freedom in order to regain control of their environment. This reactance behaviour is in another, sometimes opposite, direction of that is desired by the social group (Rummel et al., 2000). The concept of reactance is often connected to the idea of reverse psychology, which is based on the notion that telling someone not to do something would in turn make it more appealing (Brehm, 1966).
Verification of this theory has been reported in various situations, such as alcohol consumption (Allen, Sprenkel, & Vitale, 1994) and consumer behaviour (Rummel et al., 2000). For example, Allen and colleagues (1994) investigated law changes with regard to alcohol consumption had an effect on the alcohol consumption of underage students. The researchers found that the alcohol consumption of underage students was significantly higher than their legal-age counterparts, because their freedom was restricted by the law. This result
is consistent with the psychological reactance theory. Rummel and colleagues (2000) also found evidence that supports the psychological reactance theory. Their findings showed that adolescents preferred certain products, i.e., roller blades, sneakers, soda and bubble gum, more when parents disapproved those products. However, they found that younger children did confirm to their parent’s preferences. Furthermore, the results indicated that the liking of certain products, i.e., Nintendo, rollerblades and soda, significantly decreased among older children relative to younger children when friends disapproved. In studies of Stok and colleagues (Stok et al., 2014a; Stok et al., 2015; Stok et al., 2016) it is also suggested that injunctive social norm messages likely caused feelings of reactance. Because adolescents highly value belonging to a social group and attempt to gain approval from their social group members, they tend to behave following the standards of that group (McDonald, & Crandall, 2015). However, when these standards are not in line with the individual’s sense of freedom of choice, the individual will have feelings of reactance (Brehm, 1966).
An explanation of the difference between the effect of descriptive and injunctive norms with regard to reactance, which is that injunctive norm messages will cause higher feelings of reactance relative to descriptive norm messages, can be found in the meaning of both social norms. According to Deutsch, and Gerard (1955), both types of social norms refer to a different source of human motivation. Descriptive norms describe behaviour in a certain situation without mentioning restrictions (Stok et al., 2016), whereas injunctive norms proscribe or dictate how people are ought to behave through the promise of social sanctions (Kallgren et al., 2000). As a result, the language used in injunctive norms is much more forceful, controlling and freedom-threatening relative to descriptive norms (Grandpre et al., 2003).
Empirical research showed that the use of more controlling, freedom-threatening language has a great impact on whether the message is recognized as a threat to freedom of
choice (Dillard, & Shen, 2005; Grandpre et al., 2003). As explained by the psychological reactance theory, this threat to freedom of choice will cause feelings of reactance (Brehm, 1966; Grandpre et al., 2003; Stok et al., 2015). For example, Grandpre and colleagues (2003) examined the effects of explicitness of smoking messages on reactance among adolescents. They found that explicitly controlling messages have a greater influence on adolescents’ feelings of reactance than implicit or non-controlling messages. However, Grandpre and colleagues did not directly measure reactance, but only the behavioural outcomes of reactance. In addition, Miller, Lane, Deatrick, Young, and Potts (2007) investigated the effects of language use within health messages. Similar to the findings of Grandpre and colleagues (2003), the results showed that controlling language use led to higher feelings of reactance. Both studies, however, found effects on reactance towards behavioural intentions, not on actual behaviour. Effects of reactance n actual behaviour have not been tested
empirically yet.
In sum, when reading an injunctive social norm message, feelings of reactance will be higher relative to reading a descriptive norm message. This is expected to result in lower intentions to consume fruit and actual fruit consumption for those who read the
injunctive compared to the descriptive norm message. Based on this suggestion, the following is hypothesized:
H2: The relation between social norm messages (descriptive and injunctive) and intended and actual fruit consumption is mediated by adolescents’ reactance to social norm messages, whereby higher feelings of reactance will cause lower reports of intended and actual fruit consumption, especially when reading an injunctive social norm message.
Feelings of resistance to peers
The relationship between injunctive and descriptive social norm messages and reactance towards these messages, however, may depend on the age of the adolescent who reads the message. The most important developmental goal for adolescents is the successful
development of their peer relationships, whereupon adolescents may start giving more priority to maintaining positive relationships with their peers than with their parents (Brown et al., 2000). Adolescents will try hard to be accepted by others and therefore align themselves with the standards and expectations of their social group, to ensure they will not be excluded from the group. As a result, adolescents tend to conform to the social norms of their peers
(Steinberg, & Monahan, 2009).
However, this conformity is expected to be strongest in early adolescence, because the ability to resist influence from peers increases with age (Sumter et al., 2009; Steinberg, & Monahan, 2009; Walker, & Andrade, 1996). According to Steinberg, and Monahan (2009), adolescents between between 14 and 18 years old are able to resist influence from their peers, but younger adolescents have difficulties resisting the influence of their peers. In order to explain the outcomes of their study, Steinberg and Monahan refer to the onset of the adolescent period. In the onset of the adolescence period, adolescents become more independent of their parents and rely more upon their friends and peers. During the period between 10 and 14 years old, the shift towards dependence of peers is strongest, and therefore, there will be no increase in resistance to peer influence in this time period. However, later in adolescence, adolescents may feel the need to withstand the influence of their peers in order to form their own identity (Sumter et al., 2009).
The finding that during later adolescence, adolescents are less likely to follow their peers and try to resist the influence of their peers was also found in other studies (Sumter et al., 2009; Walker, & Andrade, 1996). For example, Walker and Andrade (1996) found that
none of the 15 to 17 years old conformed to the norms of their peers, while 38 percent of the children between 9 and 11 years old did, and Sumter and colleagues (2009) found in their Dutch sample that an increase in reported resistance to peer influence already happened steadily from the age of 10.
Because younger adolescents (10 to 14 years old) conform more often to the social norms of their peer group than older adolescents (14 to 18 years old), social norm messages may have a higher impact on their behaviour than on older adolescents’ behaviour. Thereby, this impact will be higher when adolescents read an injunctive norm message, because it is expected (H2) that these messages will cause feelings of reactance, because of the freedom-threatening nature of injunctive norm messages. In order to examine this suggestion, this study examines the differences between two age groups of secondary school students: a group of 11 to 14 years old and a group of 15 to 18 years old. Therefore, the following is
hypothesized:
H3: Age will moderate the mediated relation between social norm message (descriptive and injunctive) and intended and actual fruit consumption by adolescents’ reactance to the norm message, whereby increasing age will be associated with increasing levels of reactance to social norm messages., especially when reading an injunctive social norm message.
Johnson, and Buboltz (2000) studied the relationship of self and psychological reactance. Results from questionnaires showed that lower levels of peer individuation, i.e., the ability to function in a self-directed way without being controlled by peers, predict higher levels of psychological reactance. Thus, people who are less influenced by their peers will have higher feelings of reactance when they perceive threats to their freedom, for example when they read an injunctive norm message. Therefore, the following is hypothesized:
H4: Reported feelings of resistance to peers will moderate the mediated relation between social norm message (descriptive and injunctive) and intended and actual fruit consumption by adolescents’ reactance to the norm message, whereby increasing levels of reported resistance to peers will lead to increasing levels of reactance to social norm messages and lower intended and actual fruit consumption, especially when reading an injunctive social norm message.
A schematic overview of relation between the independent variable, dependent variables, mediator, including the associated hypotheses, is presented in Figure 1.
Figure 1 Schematic overview of the variables and hypothesis of the research.
Method
Participants & design
Participants were recruited in twelve classes (approximately 29 students per class) of a Dutch secondary school: Minkema College in the city of Woerden, six classes of first grade students,
two classes of fourth grade students and four classes of fifth grade students were invited to participate. All participants followed the highest level of secondary education, i.e., pre-university education or gymnasium2. A total of 291 students participated and filled in the questionnaire during school hour (Time 1). Of these 291 students, 189 students also filled in the three-day fruit diary (Time 2).
Of the 291 participants, 11 participants had fruit allergies and 2 of them were excluded from the research, because their answer on the question if they had a fruit allergy was
missing, but they did fill in that they were both allergic for 9 kinds of fruits. Besides the fact that being allergic to 9 kinds of fruit is a great restriction of the ability to consume fruit, the answers were coded as unreliable because of the ambiguous answers. The other 9 participants were allergic for 1 to 4 kinds of fruit (M = 2.22, SD = 1.20) and were not excluded from the research. Besides that, three students of the same school class filled in exactly the same answers for all the questions of the questionnaire (Time 1). Furthermore, two of them also filled in the fruit diary (Time 2) and filled in almost the similar fruit consumption as well. Their answers on the questionnaire and fruit diary seemed unreliable, and, therefore, these three participants were also excluded from the study. Furthermore, one participant consumed more fruit in three days than 3 standard deviations (SD = 1.22) above the mean (M = 1.86). This high number of actual fruit consumption indicates a deviation relative to the other data, and, therefore, actual fruit consumption of this participant was set to missing. Thus, excluded from the research were students who filled in unreliable allergy information and those who filled in exactly the same answers on all the questions of the questionnaire (N = 5). Of one
2 In the Netherlands, there is no difference in level of education when looking at secondary school graduation
diploma on levels of pre-university education or gymnasium. Therefore, in this research, level of education will not be regarded as different between the groups.
student, the data of the fruit diary was excluded from the research, because of extreme values of fruit consumption that indicated unreliability.
As shown in Table 1, the final sample consisted of 286 students, who all filled in the questionnaire, and 184 of them filled in the fruit diary as well. Two first and two fourth or fifth grade classes were randomly assigned to each condition, namely the descriptive (N = 105), injunctive (N = 98), and control condition (N = 83). Students in the same class were always assigned to the same condition to avoid discussions about the different conditions and the goal of the research. The Time 1 sample consisted of 286 participants. First grade students (N = 152) were on average 12.66 (SD = 0.68) years old, and fourth/fifth grade students (N = 134) were on average 16.26 (SD = 0.79) years old. Of the participants in Time 1, 149 students were boys and 137 students were girls. The Time 2 sample, students who completed both parts of the experiment, consisted of 184 participants. First grade students (N = 89) were between 12.67 (SD = 0.72) years old, and fourth/fifth grade students (N = 95) were 16.32 (SD = 0.80) years old. Of the participants in Time 2, 101 students were boys and 83 students were girls.
The school board gave permission for the present research. Before the questionnaire was handed out, all participants signed an informed consent form (Appendix 1). With the opt-out procedure (passive consent), parental approval was sought (Appendix 2). None of the parents made use of this option. The current procedure was approved of by the ethics committee of the University of Amsterdam.
When the students handed in both the questionnaire and the fruit diary, they received a lottery ticket for a chance to win one of five €10,- gift cards. The winning lottery tickets were picked randomly and were announced by means of an e-mail from the experimenter. In the same e-mail, all students were debriefed about the purpose of the study (Appendix 3). Table 1 Descriptive statics of final sample of the research on Time 1 and 2.
Condition Age Gender Year of education Time 1 No-norm N 83 83 83 M 14.51 1.52 2.90 SD 1.92 0.50 1.80 Descriptive N 105 105 105 M 14.25 1.57 2.61 SD 2.02 0.50 1.76 Injunctive N 98 98 98 M 14.32 1.47 2.79 SD 1.89 0.50 1.760 Total N 286 286 286 M 14.35 1.52 2.76 SD 1.94 0.50 1.77 Time 2 No-norm N 57 57 57 M 15.09 1.56 3.40 SD 1.82 0.50 1.73 Descriptive N 67 67 67 M 14.36 1.61 2.78 SD 2.09 0.49 1.80 Injunctive N 60 60 60 M 14.27 1.47 2.75 SD 1.92 0.50 1.74 Total N 184 184 184 M 14.55 1.55 2.96 SD 1.98 0.50 1.78
Procedure
Students were asked to participate in a study on daily fruit consumption during one of their classes. The experimenter visited all participating classes on the same Monday, April 18th 2016. This was done to ensure all participating classes had the same instruction about the research and started the research at the same day. Before agreeing to take part in the study the experimenter explained to the students that participation was voluntary and their answers would remain anonymous. Participants started by reading and signing an informed consent form (Appendix 1), which was subsequently collected by the experimenter.
After students signed the informed consent form, they received a booklet that
contained questionnaires that measured demographic characteristics and resistance to peers, a small informational text including the manipulation for one of the three conditions
(descriptive social norm message, injunctive social norm message or no-norm control message), and questionnaire that measured reactance to the social norm message and intentions to consume fruit. The questionnaires were for all three groups identical. The experimenter told the participants the research was about their daily expected and actual consumption of fruit. Additionally, the experimenter explained that there were no good or wrong answers, and that their answers would stay anonymous.
The booklets were then completed in class, while participants were being seated separately to ensure that they would fill in the booklet individually. Students were asked to read the questionnaire and the (social norm) message carefully. When participants had a question about the questionnaire, they could raise their hand and the experimenter would answer their question. Furthermore, they were asked to truthfully fill out the questionnaire. There was one question in the questionnaire that first graders found very difficult to
understand, namely the first question of the reactance questionnaire containing the word ‘resistance’. When the participants did not understand the meaning of this word, the
experimenter explained it: ‘when you read the text, did you have the feeling to counter the message?’ When all participants finished the experiment, they handed in their booklets.
When the participants handed in their booklets, they received a fruit diary that they had to fill in for the next three days, starting the day after filling in the booklet (on Tuesday April 19th 2016). The experimenter explained how students needed to fill in the fruit diary and asked them to return it at the end of the week (on Friday April 22nd 2016).
All participants received two reminder e-mails, reminding them that they had to fill in their fruit diary. They received this e-mail on Tuesday and Thursday. On Thursday, the e-mail also contained information reminding them that they had to hand in their diary the next day and when and where they had to do so. On Friday, the experimenter collected the booklets in 10 out of 12 classes. An English teacher collected the diaries of the remaining two classes, because of time schedule reasons. Because some participants forgot to take the fruit diary to school, they handed their fruit diary in after the may holidays on Monday May 9th 2016, that is 2,5 weeks later (N = 12). An independent samples t-test showed no differences on the dependent variables between the two groups caused he moment of handing in (all p > .05).
Materials and manipulations
Booklet. The participants of the study received a booklet, including a small instruction
text about how to fill in the booklet, a questionnaire and an informational text about fruit consumption. The informational text included the manipulation for one of the three conditions: descriptive, injunctive and no-norm control.
Manipulation. The informational texts contained the manipulation. All participants
were asked to read the text carefully. The texts the participants had to read only differed with respect to the presence or absence of a social norm (Appendices 4-7). All participants read the following main text: ‘Healthy eating can contribute to being healthy. By eating healthily, you
can maintain your weight and will not become overweight. In addition, healthy eating reduces the risk of developing several serious diseases like diabetes and coronary diseases. An
important part of healthy eating is to consume sufficient fruits. In previous studies that we conducted at high schools in the area of Woerden, we asked high school students like yourself how they think about eating sufficient fruit’.
For control group participants, the text ended with two summary points of the text: ‘1.
Healthy eating can contribute to being healthy. An important part of healthy eating is to consume sufficient fruits. 2. In previous studies that we conducted at high schools in the area of Woerden, we asked high school students like yourself how they think about eating sufficient fruit.’ In the descriptive norm condition, one additional sentence was added to the main text
about the results from these supposed previous studies: ‘This research showed that 81% of
high school students in the area of Woerden tries to eat sufficient fruit themselves.’ The
descriptive norm text also ended with two summary points: ‘1. Healthy eating can contribute
to being healthy. An important part of healthy eating is to consume sufficient fruits. 2. 81% of high school students in the area of Woerden tries to eat sufficient fruit themselves.’ In the
injunctive norm condition, one additional sentence was added to the main text about the results from these supposed previous studies: ‘This research showed that 81% of high school
students in the area of Woerden think other high school students should eat sufficient fruit.’
The injunctive norm text also ended with two summary points: ‘1. Healthy eating can
contribute to being healthy. An important part of healthy eating is to consume sufficient fruits. 2. 81% of high school students in the area of Woerden think other high school students should eat sufficient fruit.’
Fruit diary. When participants handed in their booklet, they received a fruit diary
(Appendix 8). The fruit diary consisted of a small instructional text about the fruit diary and four pages with schemes to fill in their fruit consumption, of which the last one was a backup
scheme. The fruit diary started each day with a day number (1, 2 or 3), weekday (Tuesday, Wednesday and Thursday) and date (April 19th, 20th and 21st). The fruit diary was similar for all groups, except for the two summary points at the top of each page of the fruit diary. Each student received a fruit diary that had the two summary points of the manipulation condition (descriptive, injunctive or no-norm control) they were exposed to before (in the booklet). These summary points were similar to the summary points at the end of the manipulation text in the booklet. For control group participants, the two summary points were: ‘1. Healthy
eating can contribute to being healthy. An important part of healthy eating is to consume sufficient fruits. 2. In previous studies that we conducted at high schools in the area of Woerden, we asked high school students like yourself how they think about eating sufficient fruit.’ For descriptive norm group participants, the two summary points were: ‘1. Healthy eating can contribute to being healthy. An important part of healthy eating is to consume sufficient fruits. 2. 81% of high school students in the area of Woerden tries to eat sufficient fruit themselves.’ For injunctive norm group participants, the two summary points were: ‘1. Healthy eating can contribute to being healthy. An important part of healthy eating is to consume sufficient fruits. 2. 81% of high school students in the area of Woerden think other high school students should eat sufficient fruit.’
Measures
Background variables. The first questionnaire contained five demographic questions:
age, educational level (senior general secondary school/pre-university education/gymnasium), educational grade (1 to 6), gender (girl/boy) and whether or not people had fruit allergies, and if so, which.
Moderator variable. The second questionnaire measured participants’ resistance to
Influence Scale (RPI; Sumter et al., 2009). This questionnaire consisted of five items: ‘It’s
pretty easy for my friends to get me to change my mind’, ‘Sometimes I do something that I know is wrong just to stay on my friends' good side’, ‘I will have my own opinion in front of my friends, even if I know my friends will make fun of me because of it’, ‘Sometimes I change the way I act when I am with my friends’, and ‘I say things I don’t really believe, if I think it will make my friends respect me more’. All items were assessed on a 5-point Likert scale
ranging from 1 (completely disagree) to 5 (completely agree). The third item (‘I will have my
own opinion in front of my friends, even if I know my friends will make fun of me because of it’) was turned to analyse the reliability of the items as a scale. A factor analysis indicated that
four items form a single uni-dimensional scale (EV = 2.10). These four items correlated with the total scale to a good degree (lower r = .60). The third item (‘I will have my own opinion in
front of my friends, even if I know my friends will make fun of me because of it’) was not
included in the scale, because the internal consistency of the scale is not that high (𝛼 = .66) and would be even lower when including this item (Cronbach’s alpha would decrease with .05). The scale of resistance to peers appeared to be moderately internally consistent, 𝛼 = .66 (M = 1.94; SD = 0.63). A mean resistance score was computed including the four items.
Mediator variable. Participants’ reactance to the message was assessed with an
adapted and shortened Dutch version of the Hong Psychological Reactance scale (Hong, & Faedda, 1996). This questionnaire consisted of eight items: ‘The message I just read triggered
a sense of resistance in me’, ‘It irritates me that the message points out things that are obvious to me’, ‘I find contradicting the message stimulating’, ‘The advice of the message induces me to do just the opposite’, ‘I resist the attempts of the message to influence me’, ‘I become angry, because the message restricts my freedom of choice’, ‘I consider the advice of the message as intrusion’ and ‘The message forces me to eat sufficient fruit and then I feel like doing the opposite’. The items were assessed on a 5-point Likert scale ranging from 1
(completely disagree) to 5 (completely agree). The scale of reactance to norm message has a high internal consistency, 𝛼 = .83 (M = 1.76; SD = 0.68). A mean reactance score was computed including all eight items.
Dependent variables. Participants’ intentions to consume sufficient fruit was assessed
with the questionnaire of Stok et al. (2014a). This questionnaire consisted of four items (‘I
intend/plan/expect/want to eat sufficient fruit in the coming period’) on a 5-point Likert scale
ranging from 1 (completely disagree) to 5 (completely agree). The scale of intentions to consume fruit has a high internal consistency, 𝛼 = .93 (M = 3.79; SD = 1.00). A mean intention score was computed including the four items.
Actual fruit consumption was measured by means of a fruit diary, which participants
filled out on three consecutive days. Each day started with the question ‘Did you eat fruit
today?’. When participants indicated that they did not eat fruit that specific day, they were
finished with the fruit diary for that day. When participants indicated they did eat fruit that day, they were instructed to fill in a fruit scheme. This fruit scheme was a list of 26 fruits, based on the fruit consumption list that was used by Stok and colleagues (2014a), on which participants could indicate which type of fruit they ate along with the amount of it (in pieces for the amount of larger fruits, such as bananas; in glasses for the amount of fresh pressed juices; and in handfuls for smaller fruits, such as strawberries). Three ‘other’ options were provided in case the type of fruit that a participant consumed, was not mentioned in the fruit scheme. Fruit consumption was calculated by computing the total amount of portions of fruit that participants consumed. Portions of fruits were calculated using the guideline used by Stok and colleagues (2014a), based on nutritional guidelines (Appendix 9). For very small fruits, such as strawberries, one handful was similar to one portion of fruit. For small fruits, such as apricots, two or three pieces constitutes one portion. For normal-sized fruits, such as bananas, one piece is one portion of fruit. For large fruits, such as pineapples, parts of the fruit
constitute one portion. An average daily fruit consumption score was calculated by dividing the total fruit consumption by three, because of the three days reporting fruit consumption (M = 1.86, SD = 1.22). Remarkably, there were no participants did not fill in one or two days of the fruit diary. However, not all participants handed in their fruit diary (N = 184).
Results
The analyses that are presented here were conducted on the total sample of the research, minus the five participants that were removed because of reliability issues (N = 286), except for the analyses including the students’ actual fruit consumption. In that case, only students who also filled in the fruit diary were included in the analyses. One student was removed from the fruit diary sample, because of an extremely high fruit consumption (N = 184).
Randomization check
To determine whether randomization was successful and participants were equally distributed between the three conditions, two analyses were conducted with control variables. The control variable level of secondary education was not taken into account in these analyses, because the participants all had the same level of secondary education, namely pre-university education, and therefore participants’ level of education was equal between the three conditions.
First, a one-way independent ANOVA was conducted with age and year of secondary education as dependent variables and social norm condition as independent variable. The analyses showed that there were no significant differences between the experimental conditions, regarding variables age, F (2, 283) = 0.43, p = .654, and year of secondary
gender as outcome variable and social norm condition as independent variable. The analysis showed that there were no significant differences between the experimental conditions, regarding the variable gender, 𝜒2 (2) = 2.12, p = .347. Overall, the analyses showed that
participants were equally distributed between the three conditions. This means that the randomization of this research was successful.
Main analyses
To test the first hypothesis that a descriptive norm message will have a greater influence on adolescents’ intentions to consume fruit and actual fruit consumption compared to an
injunctive norm messages or a message without a social norm, i.e., the control message, two ANOVA’s were conducted.
Intentions to consume fruit. To analyse the effects of social norm messages (i.e.,
descriptive, injunctive and control messages) on the participants’ intentions to eat sufficient fruit, a univariate ANOVA was conducted. The ANOVA was conducted with social norm message as independent variable and intentions to consume fruit as outcome variable. The results of the analysis indicated that the message that participants read had no significant influence on their intentions to consume fruit, F(2, 283) = 1.62, p = .200, 𝜂p2 = .01.
Participants who read a descriptive social norm message (M = 3.85, SD = 1.03) did not differ significantly from those who read an injunctive social norm (M = 3.86, SD = 0.97) or no-norm control message (M = 3.62, SD = 0.97) on the intentions to consume fruit.
Actual fruit consumption. To analyse the effects of social norm messages (descriptive,
injunctive and control messages) on the participants’ three-days actual fruit consumption, a univariate ANOVA was conducted. The ANOVA was conducted with social norm message as independent variable and actual fruit consumption as outcome variable. The results of the analysis indicated that the message that participants read had no significant influence on their
actual fruit consumption, F(2, 180) = 0.30, p = .744, 𝜂p2 = .00. Participants who read a
descriptive social norm message (M = 1.95, SD = 1.24) did not differ significantly from those who read an injunctive social norm (M = 1.80, SD = 1.19) or no-norm control message (M = 1.81, SD = 1.25) on actual fruit consumption.
In sum, because no main effect was found of message type on participants’ intentions to consume fruit or their actual fruit consumption, the first hypothesis is rejected. Participants who read a descriptive social norm message did not differ significantly from those who read an injunctive social norm or control message on the intention to eat sufficient fruit or their actual daily fruit consumption.
Moderated mediation of reactance
The second hypothesis states that the relation between social norm messages (descriptive and injunctive) and (intended) fruit consumption will be mediated by reactance towards social norm messages. Furthermore, the third hypothesis states that this mediation is moderated by age, the fourth hypothesis states that this mediation is moderated by participants’ resistance to peers. To test these hypotheses, Hayes’ PROCESS macro for moderated mediation models was used (Hayes, 2012). PROCESS is a computational tool for SPSS that can be used for mediation, moderation, and conditional process analysis. PROCESS uses a framework to estimate direct and indirect effects in mediator models. Additionally, the macro implements bootstrap methods for inference about indirect effects in mediation models. In the analyses for this study, 10,000 bootstrap samples were used to estimate the bias-corrected bootstrap
confidence intervals (BCBCI).
PROCESS cannot estimate a moderated mediation model with a multicategorical independent variable or moderator. Therefore, the independent variable, the social norm message, was recoded into a dichotomous variable for the moderated mediation analyses. The
new variable included the descriptive (N = 66) and injunctive social norm condition (N = 60), respectively coded as 0 and 1.
Mediation by reactance
To test the second and third hypothesis model 7 of Hayes’ PROCESS was used. The model was conducted with social norm message (descriptive or injunctive) as independent variable, intentions to consume fruit and actual fruit consumption as outcome variables, reactance to social norm message as mediator and age as moderator. The second hypothesis is that the relation between social norm messages and (intended) fruit consumption will be mediated by reactance towards social norm messages.
Intentions to consume fruit. The results in Table 2 show no significant indirect effect
of social norm on participants’ intentions to consume fruit through reactance towards the norm message, 𝛽 = -0.04, p > .05, 95% CI [-0.30, 0.21]. In other words, social norm
messages do not predict intentions to consume fruit through reactance towards the message. However, there is a significant negative effect of reactance towards the norm message on intentions to consume fruit, i.e., reactance significantly predicts intentions to consume fruit, 𝛽 = -0.59, p < .001, 95% CI [-0.77, -0.40]. This effect indicates that when reactance towards norm message increases, participants’ intentions to consume sufficient fruit declines, and vice versa.
Actual fruit consumption. The results in Table 2 show no significant indirect effect of
social norm on participants’ intentions to consume fruit through reactance towards the norm message, 𝛽 = -0.17, p > .05, 95% CI [-0.59, 0.25]. In other words, social norm messages do not predict actual fruit consumption of participants through reactance towards the message. However, there is a significant negative effect of reactance towards the norm message on actual fruit consumption, i.e., reactance significantly predicts actual fruit consumption, 𝛽 =
-0.45, p < .01, 95% CI [-0.77, -0.14]. This effect indicates that when reactance towards norm message increases, participants’ actual fruit consumption declines, and vice versa.
Table 2 Results of the moderated mediations by PROCESS (Hayes, 2012).
Moderator Dependent variable 𝛽 se t p BC Age Intention (N = 203) LL95 UL95 Norm (X) a1 -1.60 .72 -2.23 .027* -3.01 -0.19 Age (W) a2 -0.04 .03 -1.16 .249 -0.10 0.03 Interaction (XW) a3 0.11 .05 2.12 .035* 0.01 0.20 Reactance (M) b1 -0.59 .09 -6.23 .000*** -0.77 -0.40 Norm (X) c’ -0.04 .13 -0.34 .732 -0.30 0.21 Consumption (N = 126) Norm (X) a1 -1.06 .86 -1.24 .217 -2.76 0.63 Age (W) a2 -0.04 .04 -0.91 .365 -0.11 0.04 Interaction (XW) a3 0.07 .06 1.20 .233 -0.05 0.19 Reactance (M) b1 -0.45 .16 -2.80 .006** -0.77 -0.13 Norm (X) c’ -0.17 .21 -0.80 .426 -0.59 0.25 Resistance to peers Intention (N = 203) Norm (X) a1 0.06 .31 -0.21 .835 -0.54 -0.67 Resistance (W) a2 -0.34 .11 3.17 .002** 0.13 0.55 Interaction (XW) a3 -0.06 .15 -0.39 .695 -0.36 0.24 Reactance (M) b1 -0.59 .09 -6.23 .000*** -0.77 -0.40 Norm (X) c’ -0.04 .13 -0.34 .732 -0.30 0.21 Consumption (N = 126) Norm (X) a1 -0.12 .39 -0.30 .761 -0.90 0.66 Resistance (W) a2 0.19 .13 1.45 .149 -0.07 0.46 Interaction (XW) a3 0.06 .20 0.30 .762 -0.33 0.45 Reactance (M) b1 -0.45 .16 -2.80 .006** -0.77 -0.13 Norm (X) c’ -0.17 .21 -0.80 .426 -0.59 0.25 Note. PROCESS Model 7; number of bootstrap re-samples: 10.000.
In sum, the results of the analyses of H2 indicate that there were no significant indirect effects of social norm message on intended or actual fruit consumption through participants’ reactance to social norm messages. However, reactance does affect intended and actual fruit consumption: when reactance to norm message increases, both intended and actual fruit consumption declines. In other words, higher feelings of reactance will cause lower reports of intended and actual fruit consumption. The second hypothesis also suggested a difference between the type of social norm. Contrary to the expectations, there is no difference in exposure to descriptive versus injunctive norm. Therefore, the second hypothesis is partly rejected.
Moderation by age
The third hypothesis is that the mediated relation between social norm message and intended and actual fruit consumption by reactance to social norm messages is moderated by age.
Intentions to consume fruit. The results of Table 2 show that there is no significant
effect between social norm on intentions to consume fruit by reactance to norm when taking into account the moderation by age on the mediation of reactance, 𝛽 = -0.04, p > .05, 95% CI [-0.30, 0.21]. However, results indicate a significant effect of age as a moderator between the relation of social norm message on reactance towards the norm message, 𝛽 = 0.11, p < .05, 95% CI [-0.01, -0.20]. This result indicates that the effect of social norm message on reactance to norm message is influenced by age. When age is lower (M = 12.33), there is a significant positive relationship between norm and reactance, 𝛽 = 0.18, CI [0.01, 0.40]. This means that younger students have higher levels of reactance to social norms when reading an injunctive than a descriptive norm. However, when students become older, there is no
significant relation: age does not affect the relation between social norms on reactance. Thus, the positive relation between social norm and reactance only really emerge in younger
students. Additionally, as the analysis of H2 already showed, intentions to consume fruit will be predicted by reactance towards social norm message: when reactance towards norm message increases, participants’ intentions to consume sufficient fruit declines. Despite the fact that the total effect is not significant, some parts of the model are.
Actual fruit consumption. Results of Table 2 show that there is no significant effect of
age as a moderator on the mediated relation between social norm on actual fruit consumption by reactance to norm, 𝛽 = -0.17, p > .05, 95% CI [-0.59, 0.25]. There is also no significant effect of age as a moderator between the relation of social norm message on reactance towards the norm message, 𝛽 = 0.07, p > .05, 95% CI [-0.05, 0.19].
In sum, the results of the analyses of H3 indicate that there were no significant indirect effects of age as a moderator on the relation between social norm message on intended or actual fruit consumption through participants’ reactance to social norm messages. However, in the intended fruit consumption situation, younger students have higher levels of reactance to social norms when reading an injunctive than a descriptive norm. This effect indicates that injunctive norms arouse higher levels of reactance for younger age groups, but this finding was expected for older adolescents. Therefore, the third hypothesis is rejected.
Moderation by resistance to peers
To test the fourth hypothesis that the relation between social norm message and reactance to social norm messages is moderated by participants’ resistance to peers, again model 7 of Hayes’ PROCESS was used. The model was conducted with social norm message (descriptive or injunctive) as independent variable, actual fruit consumption as outcome variable, reactance to social norm message as mediator and resistance to peers as moderator. The fourth hypothesis is that the mediated relation between social norm message and intended
and actual fruit consumption by reactance to social norm messages is moderated by resistance to peers.
Intentions to consume fruit. The results of Table 2 show that there is no significant
effect of resistance to peers as a moderator on the mediated relation between social norm on intentions to consume fruit by reactance to norm, 𝛽 = -0.04, p > .05, 95% CI [-0.30, 0.21]. However, results indicate a significant influence of resistance to peers on reactance towards the norm message, 𝛽 = -0.34, p < .01, 95% CI [0.13, 0.55]. When resistance to peers declines, reactance to social norm message also declines3. Thus, when adolescents are not capable of resisting influence from their peers, they have lower feelings reactance to social norm messages compared to adolescents who can resist influence from their peers.
Actual fruit consumption. Results of Table 2 show that there is no significant effect of
resistance to peers as a moderator on the mediated relation between social norm on actual fruit consumption by reactance to norm, 𝛽 = -0.17, p > .05, 95% CI [-0.59, 0.25]. There is also no significant effect of resistance to peers as a moderator between the relation of social norm message on reactance towards the norm message, 𝛽 = 0.06, p > .05, 95% CI [-0.33, 0.45].
In sum, the results of the analyses of H4 indicate that there were no significant indirect effects of resistance to peers as a moderator on the relation between social norm message on intended or actual fruit consumption through participants’ reactance to social norm messages. However, a significant effect is found for the intended fruit consumption situation: when
3 The resistance to peers must to be seen as turned: when the variable resistance to peers decreases, the actual
resistance to peers increases, because of the questionnaire that is used. For example, the first item of the scale is ‘It’s pretty easy for my friends to get me to change my mind’ and can be answered on a 5-point Likert scale ranging from 1 (completely disagree) to 5 (completely agree). When answering 1, completely disagree, the participant shows resistance towards peers.
adolescents are not capable of resisting influence from their peers, they have lower feelings of reactance to social norm messages compared to adolescents who can resist influence from their peers. Although this finding is in line with parts of the hypothesis, the overall findings suggest no significant effects of this moderated mediation. Therefore, the fourth hypothesis is rejected.
Discussion
The aim of this study was to analyse whether descriptive and injunctive social norm messages differentially affected adolescents’ intentions to consume fruit and actual fruit consumption. The goal of this study was also to investigate the possible mediating role of reactance towards social norm messages and the possible moderating role of adolescents’ age and resistance to peer influence.
Although there is much research on the effects of social norms on (intentions of) health behaviour (Lally et al., 2011; Mollen et al., 2013; Robinson et al., 2014b; Stok et al., 2014a; Stok et al., 2016) and on reactance towards the social norms (Dillard, & Shen, 2005; Grandpre et al., 2003), research has not yet combined the two empirically. Therefore, this research will examine the mediating effect of reactance to social norms empirically.
The main expectation of this research was that there would be a main effect of descriptive social norm messages, in comparison to injunctive norm and no-norm control messages, on intentions to consume fruit and actual fruit consumption (H1). It was also hypothesized that this relation would be mediated by participants’ reactance to social norm messages (H2). Finally, it was expected that respondents’ age (H3) and resistance to peer influence (H4) would function as moderator on social norm message and actual fruit
consumption. Findings showed no significant main effects and therefore the four hypotheses were (partly) rejected.
