Graduate school of communication
Exploring the boundaries of resistance to persuasion:
An experimental investigation of the effects of language and source on
resistance towards a promotional health message
Master thesis of Emmy Bruijstens
Student number: 10876588
Supervisor: mw. dr. M.L. Fransen
Master’s programme Persuasive Communication January 29, 2016
Abstract
How various message elements influence resistance to persuasive health messages remains a consistently pressing issue for campaign developers. Controlling language has been identified as an important cue for resistance, but combinations of language with other message elements are yet to be studied . It was expected that a source’s motives, selling versus informing, would moderate the effect of controlling language. Furthermore, no agreement has been reached on the measurement of resistance towards persuasive messages. An online experiment was conducted, combining two language and two source variations, to investigate their effects on resistance, and subsequently on attitudes and intentions. Besides measuring resistance using previously used measures, the experiment tested a new, implicit measure of resistance. Results showed that controlling language stimulates counter-arguing, which negatively impacts attitudes and intentions. The manipulation of source did not generate significant differences. Violating expectations, it was also demonstrated that when controlling language did not evoke counter-arguing, controlling language could positively influence attitudes and intentions. The experiment was unable to yield proof for the implicit measurement of resistance. The results indicated that when counter-arguing occurs, controlling language negatively affects message outcomes. However, when controlling language evokes source derogation or negative affect, or no resistance, the effect is not necessarily negative. Potential moderating variables accounting for this effect are discussed.
Exploring the boundaries of resistance to persuasion: An experimental investigation of the effects of language and source on resistance towards a promotional health message
Numerous attempts have taught health campaign developers that telling people not to engage in unhealthy behaviours is close to futile. Although most promotional health messages are created with the audience’s best interest at heart, in many cases they fail to be accepted by the ones they are addressing. In a worst case scenario, a campaign may even stimulate the undesired behaviour it is aiming to discourage (Miller, Lane, Deatrick, Young & Potts, 2007).
Even though it seems an irrational assumption that information that is meant to improve individual or public health is easily rejected, it is human nature. Persuasion research over the years has demonstrated that any attempt at influencing someone’s behaviour creates some form of reactance (Burgoon, Alvara, Grandpre & Voulodakis, 2002). The famous ‘do not press the red button’ principle illustrates this process best: when we are told not to do something, the more we want to do it. By pressing the red button, we regain the freedom that was taken away when we were told ‘not to press’. The same process is likely to occur when people are told to change their health behaviour.
Using reactance theory to explain the possible psychological reactions to health messages can assist practitioners in creating promotional health messages that avoid resistant responses, but resistance remains a consistently pressing issue in health promotion (Rains & Turner, 2007). To create a health message that is effectively persuasive, non-threatening, and appealing to the right audience, is a daunting task. Moreover, it is difficult to detect to what extent the failure of a health campaign can be attributed to the occurrence of resistance, as agreement on how to scientifically operationalise resistance has not been found so far. Only when such as measure is developed, can health campaign developers more effectively design health messages (Miller et al., 2007).
The primary goal of the present research was to extend existing research on resistance towards promotional health messages, by looking at the effects of different message elements on resistance. In addition, this study is one of the first to make an attempt at measuring resistance implicitly, as it is assumed that resistance in response to a health message is a largely unconscious process. An online experiment to test that assumption is the empirical core of this paper. The main questions that guided this research were how resistance in persuasive health messages can be minimized or avoided altogether, and how this can be operationalised most accurately.
Theoretical background Persuasion & Resistance
The knowledge that a persuasive message is likely to evoke resistance in its audience is nothing new. Some persuasion literature has considered resistance as a personality trait, where some are simply more resistant than others (Knowles & Linn, 2004). In another line of
research, resistance is seen as a state that is activated by outside influences. Brehm’s psychological reactance theory, dating back to 1966, explains how basically any persuasive appeal evokes some form of resistance in receivers. When someone feels that something is threatening his or her freedom to act according to his or her own wishes, this causes
reactance. In theory on reactance, this is described as a ‘threat to autonomy’ (Burgoon et al., 2002). Reactance theory can be used to explain why health campaigns risk creating a
boomerang effect, as reactance theory implies that an advocated option becomes less
attractive, and the threatened option becomes even more attractive than before (Brehm, 1989; Miron & Brehm, 2006). The red button-example from the introduction applies here
(Whitehead & Russell, 2004). Research studying the effects of persuasive health messages, however, is not only about the psychological state of reactance. It is about the entire process, starting with a threat to freedom and the subsequent attempt to restore this freedom, together
making up the concept of resistance, that is often researched as a variable that comes between a threat to freedom and message responses (Dillard & Shen, 2005; Jenkins & Dragojevic, 2011).
Reducing a feeling of reactance, or restoring a threatened freedom in response to a persuasive message can manifest itself in many different ways. Previous studies have outlined a wide variety of different resistance strategies people can employ to resist the influence of persuasive messages (Ahluwalia, 2000; Fransen, Verlegh, Kirmani & Smit, 2015; Zuwerink-Jacks & Cameron, 2003). In some cases, resistance may lead to simply ignoring the
persuasive message, termed ‘avoidance’ (Fransen et al., 2015). Dependent on the message and its context, people may more actively resist the message by using strategies to cope with a threat. In other cases people may want to restore their freedom by acting in the opposite direction of what was intended (Burgoon et al., 2002). Although many different strategies are found in the literature, the most commonly identified resistance strategies include: counter-arguing, source derogation, attitude bolstering, and negative affect (Ahluwalia, 2000; Jenkins & Dragojevic, 2011; Zuwerink-Jacks & Cameron, 2003).
Language and resistance
Although any persuasive message can be perceived as a threat to freedom, there is great variety in the amount of resistance different persuasive messages evoke, and certain messages can be accepted while others are rejected. Different features of a message can account for this difference. One such feature is the type of language used in a message (Dillard & Shen, 2005; Jenkins & Dragojevic, 2011; Miller et al., 2007).
Resistance research makes a distinction between controlling language and
low-controlling, autonomy-supportive language. Controlling language is identified as an important resistance cue, whereby controlling language is likely to activate reactance and subsequent resistance (Jenkins & Dragojevic, 2011; Miller et al., 2007). Controlling language can be
defined as language that uses commands and orders. This type of language promotes the desired health behaviour as something compulsory, using words like “must” and “should”. Moreover, controlling language includes demeaning elements, where anyone who does not agree with the promoted viewpoint is addressed negatively (Jenkins & Dragojevic, 2011). Low-controlling language or autonomy-supportive language, on the other hand, includes words such as “possibly” and “maybe”, and focuses on giving the receiver a choice (Miller et al., 2007).
Language use has been shown to greatly affect message responses in health
promotion. Firstly, the use of appropriate language can greatly influence the extent to which a message is perceived as threatening. When the message does not threaten the receiver’s freedom of choice, there is no need for them to restore this freedom. When a message makes clear that the receiver is free to make up his or her own mind, this will generate less negative emotions (Miller et al., 2007). Secondly, language use is shown to affect subsequent message evaluations. Jenkins and Dragojevic (2011) demonstrate that the use of controlling language leads to negative source evaluations, which in turn negatively influences receivers’ attitudes and intentions. Also, controlling language risks presenting the desired behaviour as a ‘moral good’, taking a demeaning stance towards non-compliant receivers. This is also likely to increase resistance or even create a boomerang effect (Crossley, 2002).
Overall, theory and previous findings lead to the prediction that:
H1: A health promotion message that makes use of controlling language will evoke more resistance than a message with polite/non-threatening language.
Message source and resistance
As described above, the language used in a persuasive health message can influence the amount of resistance people experience when reading the message. Studies that have found such language effects, however, did not take into account any other message elements
that could possibly have influenced resistance. What they did show is that resistance to a persuasive message as a result of controlling language is often partly directed at the source of the message. The question is whether the effect that language is expected to have on
resistance will remain when other elements, such as the source of the message, are also manipulated.
The source of a persuasive message could be an important cue factor in how people process a message and how they are influenced by a message. The Elaboration Likelihood Model provides an explanation (Petty & Cacioppo, 1984). Dependent on people’s
involvement with the topic under discussion, the source of a message can serve different purposes. Specifically, people with low elaboration likelihood can use the source of a message as a cognitive shortcut to directly accept or reject the message. When elaboration likelihood is moderate, source cues can determine the extent of thinking about the message. Under high elaboration likelihood, source does not serve as a simply cue, but can serve as a persuasive argument in favour or against the message (Petty & Cacioppo, 1984).
The way source evaluations affect message responses is a popular topic in persuasion research. Variables such as source credibility, trustworthiness, and expertise can influence audience responses to messages (Jones, Sinclair & Courneya, 2003; Paek, Hove, Ju Jeong & Kim, 2011; Pornpitakpan, 2004). For example, for some types of messages a source such as a peer is believed to be more effective in influencing attitudes or behaviours, compared to an expert (Paek et al., 2011). In other cases, experts are perceived to be more credible, because they stimulate more elaborate processing of the message (Jones et al., 2003). Also, a source combined with other message elements can moderate message responses. For example, a credible source combined with a negatively framed message is more likely to evoke resistance (Jones et al., 2003). The Persuasion Knowledge Model is useful to explain how receivers of a
persuasive message process source cues, and what this implies for further message evaluations.
The Persuasion Knowledge Model
When a promotional health message is attempting to change attitudes or behaviours on a certain issue, receivers are likely to question the motivation of the sender to promote such a change. In whose best interest is the sender operating? According to Atkin, McCardle, and Newell (2008), consumers judge a message partly by assessing the motives of the company or organisation that is sponsoring the message, and this judgement influences the way people perceive the message and the brand or organisation. For example, when the source of a message is a for-profit organisation, consumers often perceive the message as biased. Rather, information provided by a non-profit organisation can expect more positive perceptions (Smith, Bauman, McKenzie, & Thomas, 2005).
According to the Persuasion Knowledge Model by Friestad and Wright (1994), shortly PKM, receivers try to identify the perceived persuasion agent of a message: the one
responsible for the message. Over time, people gain knowledge on the tactics that campaign developers use to persuade them. As a result, persuasion tactics are more easily identified, and people can more easily resist the source and its message based on their perceived intentions. This might explain why for-profit organisations with a clear sales intent are more negatively evaluated than non-profit organisations. As there is no readily identifiable sales intent in messages from non-profit organisations, resistance towards the source becomes more difficult. In addition, the PKM predicts that once a persuasion agent is identified and
negatively evaluated, this will affect further message evaluation, and makes it more likely for receivers to resist the message altogether (Friestad & Wright, 1994). What is clear from resistance research, is that resistance is often partly directed at the source of the persuasive message. Furthermore, messages that are clearly developed for sales purposes are more likely
to be negatively evaluated, according to the PKM, compared to messages without a sales intent, which makes it more difficult to direct resistance towards the source.
As was discussed in the previous section, the use of controlling language is often directly linked to negative source evaluations (Jenkins & Dragojevic, 2011; Miller et al., 2007). Furthermore, receivers have different tools at their disposal to judge the source of a message, including perceived intentions. Sources with a clear sales intent are more easily derogated based on their intentions and tactics. It therefore seems a plausible conclusion that when controlling language is employed by an organisation with a commercial intent, this will create large amounts of resistance, as a reactance response is aroused by the language, and subsequent resistance can be easily directed towards the source. On the other hand,
controlling language used in a message by a non-profit source might weaken the effect of controlling language, as receivers are more likely to accept this from a source without a clear sales intent. Making use of polite language might aid for-profit organisations in reducing the amount of resistance that is aimed at the source.
Based on what is known about the effects of language and source on resistance, the second hypothesis expected the following interaction effect:
H2: The effect that controlling language is expected to have on resistance will be different depending on the source of the message. Specifically, the effect of controlling language on resistance will weaken when the message is from a non-profit source, and will increase when controlling language comes from a for-profit source.
It is clear from previous research that the more resistance people experience, the more negative message responses will be (Jenkins & Dragojevic, 2011; Quick & Considine, 2008). Specifically, resistance will lead to negative attitudes towards the topic that is promoted, for example through counter-arguing. Also, when receivers employ a form of resistance, they are
less likely to adopt the desired behaviour (Dillard and Shen, 2005). When a message promotes or dissuades a certain health behaviour, but receivers experience resistance, it is more likely that receivers will reject the message, and will not comply with the aims of the message. Therefore a final series of hypotheses was proposed:
H3: The more resistance that is evoked by a health message, the more negative the attitudes towards the health topic will be.
H4: The more resistance that is evoked, the more negative the behavioural intentions towards the advocated behaviour will be.
Finally, the study tested the complete moderated mediation model, that predicted that the source of a message moderates the relationship between language use and resistance, and when controlling language is matched with a for-profit source, this will create the most resistance, which will result in a negative attitude towards the topic and negative behavioural intentions. A visualisation of the conceptual model is displayed in figure 1 below.
Figure 1. Conceptual model and hypotheses.
Measuring resistance
Although Brehm’s reactance theory dates back as early as 1966, and resistance to persuasion has been a popular research topic for decades now, there is currently no agreement on how to measure resistance. Brehm’s reactance theory started off as being solely supported
by evidence from logical reasoning and common sense. At the time Brehm himself was convinced that such a phenomenon was not even measurable (Dillard & Shen, 2005;. Miller et al., 2007). Since then, studies that aimed to measure the effects of threatening messages have repeatedly used resistance as an outcome variable, but direct measurements remained ignored (Dillard & Shen, 2005).
Currently, research focuses on resistance as a variable that is influenced by a threat, and popular measurements include open-ended reports and self-report items measuring resistance, attitude change, and behaviour change (Ahluwalia, 2000; Dillard & Shen, 2005; Zuwerink-Jacks & Cameron, 2003). Also, most research currently operationalizes resistance as a combination of affect and cognition (Dillard & Shen, 2005; Zuwerink & Devine, 1996). Important to note is that current attempts at measuring resistance rely greatly on self-report data. Respondents can be asked about their opinions and attitudes before and after a
threatening message, and the amount of resistance is inferred from these responses
(Ahluwalia, 2000). In other cases, respondents are explicitly asked which resistance strategies they are most likely to use when they are being persuaded (Zuwerink-Jacks & Cameron, 2003).
There are issues concerning the accuracy of these methods, however. Specifically, the question is whether receivers resisting a message are consciously aware of doing this. Until now research has yielded scattered results about the most commonly used resistance
strategies. Even within the same study respondents said they would use certain strategies, but eventually used others. Specifically, source derogation and negative affect are found to be commonly used strategies, although people did not report they would use these strategies themselves (Zuwerink-Jacks & Cameron, 2003). A possible explanation for this inconsistency might be social desirability in responses, an issue often found in explicit measurement
irritated or attacking the source might be less socially acceptable, whereas counter-arguing and attitude bolstering are considered more legitimate strategies (Zuwerink-Jacks & Cameron, 2003). Another weakness is the uncertainty that people are able to accurately report what happened during exposure to a persuasive message. Post-rationalization is an often found weakness in explicit research, as responses reflect people’s own interpretations of how a message affected them (VandeBerg et al., 2016). The question is whether this can be generalized to a real-life situation where people use resistance. Measuring resistance implicitly might offer new insights, as this could reveal automatic reactions to threatening messages more accurately. Implicit measurements might reveal strategies people use that they are not consciously aware of, or do not want to admit to using (Petty, Fazio & Briñol, 2012). Therefore, the present study also aims to investigate the following research question:
RQ: To what extent can resistance be measured implicitly, and reveal underlying thought processes at work when resistance occurs?
To stick with the current belief that resistance consists of a cognitive and an affective element, both counter-arguing and negative affect are used to measure resistance in the present study. These strategies are most prevalent among resistance research, and both represent one of the two components of resistance. Furthermore, as this study focuses on the source of a persuasive health message, the third resistance strategy used in this study is source derogation.
Methods Research design
An online experiment with a 2 x 2 between-subjects design was conducted to test the hypotheses. In total, answers from 174 respondents were retained for the experiment (Age: M = 35.6, SD = 15.5). The experiment measured the concept of resistance explicitly, but also took an exploratory turn by including a new, implicit measure in an attempt to better explain the psychological process of resistance. Four health messages were created, in which the type of language (polite vs. controlling) and the source of the message (for-profit vs. non-profit) were manipulated. The online experiment was created with the online survey tool Qualtrics, which randomly assigned participants to one of the four conditions. See table 1 for an overview of the conditions. Participants’ results were compared across conditions to measure the effects of the message manipulations. Implicit resistance was measured before and after exposure in an attempt to reveal any changes in resistance as a result of the exposure. Explicit resistance, attitude, and intention were only measured after exposure so that participants would not be aware of the purpose of the study and what was being measured.
Table 1
Overview of experimental conditions
Controlling language Polite language
Non-profit source
(Stichting HoorMij) 1 (N = 44) 2 (N = 47)
For-profit source
(Beter Horen) 3 (N = 43) 4 (N = 40)
Sample
A non-probability sampling method was used to recruit respondents for this study. Convenience sampling was the main strategy employed. The link to the online experiment
with a request for participation was disseminated through email and Facebook, and was subsequently shared among participants through these channels. Additionally, the
experiment was placed on a website where visitors could participate voluntarily. Participants were offered an incentive for participation by allotting a 25€ gift-card among participants after data collection. Requirements for participation were an age of at least 18 years old, and to be a native Dutch speaker. A total of 268 respondents participated in the experiment, of which 187 completed the entire experiment. A final number of 13 participants was excluded based on incomplete answers within the experiment. The answers of 174 participants were retained for analysis. The respondents were aged between 18 and 70, 63.3% was female, and a large proportion had a university degree (41.1%).
Procedure
When participants clicked on the link to the experiment they first read an
introduction. The introduction shortly stated that the study was about the effects of health campaigns, but did not say anything about what was measured within the experiment. Further, the introduction provided information on anonymity and voluntary participation, and contact details of the researcher, after which each participant had to give consent in order to proceed to the actual experiment. The introduction to the experiment can be found in Appendix A.
The first task of the experiment was the implicit measurement of resistance. Next, participants completed two filler tasks, unrelated to the study, in order to avoid any priming effects from the first measure. See Appendix B for the filler tasks. After that, participants were randomly assigned to one of the four message conditions, and watched the stimulus message for at least 15 seconds. Immediately after the stimulus participants completed the same implicit resistance measure. Next, participants answered the explicit resistance measures and questions about the dependent variables. As a manipulation check,
respondents were asked who the source of the message was. Lastly, socio-demographic information was gathered. The experiment ended with thanking the participants, a short debrief about the purpose of the study, and a statement that the material was created
exclusively for research purposes and that the named organisations were not involved in any way.
Measures
The section below describes how each variable was operationalised in the experiment. For specific items and questions, please refer to Appendix C.
Independent variables.
Language. For the purpose of this study, a health message was created to serve as the stimulus in the experiment. The health topic chosen for the stimulus message was the
promotion of hearing protection. This is a topic relevant to many people, as most people are exposed to loud music every now and then, especially young adults. In the Netherlands, not many previous campaigns have targeted this topic. The created message promoted the use of hearing protection, and the type of language was manipulated to be able to find a possible difference in effects of controlling and polite language use. As controlling language consists of both controlling and demeaning language, both elements were included in the message. The polite language variation follows the description of low-controlling language by Miller et al. (2007). See Appendix D.1 for the two language variations. The manipulation of language use was pre-tested to make sure that the language in the message was perceived as either controlling or polite. This way there would be more certainty that the controlling message would evoke the predicted reactance and subsequent resistance. See Appendix D.2 for the outcomes of the pre-test.
Source. The independent variable of source was manipulated with the use of two different organisational logos, clearly visible at the bottom of the stimulus message.
Different institutions could potentially promote the topic of hearing protection. Stichting HoorMij was selected as the non-profit source in this study. This institution focuses on helping people who are suffering from hearing loss, as well as educating people about hearing loss and prevention (www.stichtinghoormij.nl). Stichting HoorMij is a collaboration of different non-governmental organisations, and the word ‘Stichting’, meaning foundation, makes the non-profit aspect clear. Beter Horen served as the for-profit source. Beter Horen is a shop specialised in hearing protection and hearing instruments, has over 250 stores or service points in The Netherlands, and is nationally well-known. The final stimulus messages can be found in Appendix D.3.
Dependent variables.
Resistance. In the experiment, the concept resistance was measured implicitly as well as explicitly. To measure resistance implicitly, a new, exploratory measure was used. Based on the word fragment and word stem completion tasks, which are often used to measure implicit memory and/or associations (Fazio & Olson, 2003), a new task was developed that could potentially uncover implicit resistance thoughts. A word search puzzle was created, which contained words representing the three most commonly used resistance strategies. The word search included 10 words representing the three resistance strategies measured in this study, such as “manipulatief”, “tegenspreken”, and “irritant”. The word search also included 10 positive/neutral words, so that participants were equally likely to find negative and positive/neutral words. See Appendix E for an overview of the words included in the word search. The goal of the word search puzzle was to measure participants’ resistance by looking at the words that are recognized most quickly. Participants were granted two minutes to find as many words as possible. Based on the words found in the puzzle, a resistance score was created by calculating the proportion of resistance words found within the first six words
found. Each participants’ first answer was excluded, as this answer was the same for almost all participants, and it was simply the easiest word to find.
The same three resistance strategies were selected to operationalise the variable of resistance explicitly. The strategies were selected to represent both the cognitive and the affective element of resistance. Also, these strategies are most commonly found in resistance literature.
Counter-arguing. This strategy was measured using a scale from previous research by Asbeek-Brusse, Fransen & Smit (2015) and was adapted to fit the current topic.
Participants rated the extent to which they agreed with three statements concerning their reactions towards the message on a 7-point likert scale (1 = totally disagree, 7 = totally agree). Means of the items were computed to a score that represented counter-arguing (M = 3.25, SD = 1.49, α = .80).
Source derogation. Perceived source expertise, likeability, and sincerity together represented source derogation as a resistance strategy. Items were adapted from previous research, and participants rated their opinion on a scale from 1 to 7 about the source of the message on 10 semantic differential items with bipolar adjectives, such as
unintelligent/intelligent, unfriendly/friendly, and insincere/sincere (Campbell, 2000; Jenkins & Dragojevic, 2011; Zuwerink-Jacks & Cameron, 2003). Items were reverse coded when necessary and averaged to create an index (M = 3.63, SD = 1.03, α = .87).
Negative affect. Four items used by Dillard and Shen (2005) were translated and used to operationalise the negative emotions often associated with resistance. Participants indicated to what extent they felt negative emotions on a 5-point scale ranging from ‘not at all’ to ‘very much’. The four items together were used to create a negative affect score (M = 3.06, SD = 1.40, α = .90).
Furthermore, respondents were asked ‘indicate what you think about wearing earplugs to protect your hearing’ using a six-item semantic differential 7-point scale by Dillard and Shen (2005), with bipolar adjectives such as undesirable/desirable, foolish/wise, and unnecessary/necessary, to form an attitude index (M = 5.72, SD = .71, α = .91). Finally, behavioural intentions were measured by asking participants how likely they were to wear earplugs next time they visit an event with loud music on a 5-point scale ranging from 1 (not likely at all) to 5 (very likely) (M = 3.10, SD = 1.27).
Results Preparatory analyses
Prior to conducting a MANOVA to test the first two hypotheses, it was tested whether the three resistance strategies as dependent variables were correlated with each other to some extent. Correlations showed that all variables were significantly correlated, and thus
appropriate for conducting a MANOVA.
Hypothesis testing
Language & resistance
An initial two-way multivariate analysis of variance (MANOVA) was conducted to test the hypothesis that there would be a direct effect of language type (controlling or polite) on resistance, and whether message source would interact with this effect. In the analysis, language and source were the independent variables and the three explicitly measured resistance strategies the dependent variables. Preceding the analysis, assumption testing was conducted to check for normality, linearity, univariate and multivariate outliers, homogeneity of variance matrices, and multicollinearity, with no important violations noted.
Firstly, a statistically significant, moderate overall effect of language on the resistance strategies was obtained, Wilks’ λ = .86, F(3,168) = 8.87, p < .001, partial η2= .14. To see
whether language had an effect on all three individual resistance strategies, results of the univariate main effects were investigated further. Results showed, using a Bonferroni adjusted alpha level of .017, that language type had a statistically significant, but small effect on all three resistance strategies, counter-arguing, F(1, 170) = 17.63, p = <.001, partial η2= .09, source derogation, F(1, 170) = 12.24, p = .001, partial η2 = .07, and on negative affect, F(1, 170) = 22.25, p <.001, partial η2= .12. Looking at the mean scores, displayed in table 2,
indicated that the direction of the relationship was as expected by the first hypothesis: counter-arguing, source derogation, and negative affect increased when the type of language was controlling.
Table 2
Means and standard deviations (SD) per language condition
Mean SD N Controlling Counter-arguing Source derogation Negative affect 3.70 3.89 2.52 1.57 1.02 .98 87 Polite Counter-arguing Source derogation Negative affect 2.82 3.38 1.85 1.27 .98 .90 87
Note. counter-arguing and source derogation were measured on a scale from 1 to 7, and negative affect from 1 to 5. A higher score indicates more resistance.
Source interaction
Next, the MANOVA also tested whether there was a main effect of source on resistance. Results showed that this effect was not significant, Wilks’ λ = .98, F(3, 168) = 1.25, p = .293, partial η2 = .02, indicating that only the manipulation of source did not produce significantly different resistance scores between conditions.
Finally, the MANOVA tested the hypothesis predicting the interaction effect of message source on the relationship between language and the use of resistance strategies was
tested. Results reveal, however, that this is not the case, and the interaction effect of source with language was not significant at the 95% confidence level, Wilks’ λ = .99, F(3, 168) = .67, p = .569, partial η2 = .01. Again, univariate main effects were observed for the interaction effect of source on each of the resistance strategies separately, and revealed that this effect was not significant for any of the forms of resistance that were measured. Although the condition where controlling language was combined with a for-profit source showed the highest scores for counter-arguing (M = 3.98, SD = 1.44) and negative affect (M = 2.59, SD = .98), source derogation was highest in the controlling language condition with a non-profit source (M = 3.92, SD = 1.02). This indicates that controlling language stimulated source derogation regardless of the source and its motives. In the polite language conditions, mean scores were very similar for all three resistance strategies, regardless of source. See table 3 below for an overview of the mean scores per condition.
Table 3
Means and standard deviations (SD) per language/source condition
Counter-arguing Source derogation Negative affect
M SD M SD M SD N Controlling + Beter Horen 3.98 1.44 3.88 1.03 2.59 .98 43 Controlling + HoorMij 3.44 1.66 3.92 1.02 2.46 1.00 44 Polite + Beter Horen 2.81 1.25 3.33 .92 1.87 .89 40 Polite + HoorMij 2.79 1.30 3.40 1.05 1.82 .93 47
Note. Counter-arguing and source derogation were measured on a scale from 1 to 7, and negative affect from 1 to 5. A higher score indicates more resistance.
As a manipulation check, respondents were asked whether they remembered who the source of the message was. A surprising outcome was that the majority of respondents
remembered the source in the for-profit condition (84.34%), but in the non-profit condition only 50.55% remembered the source correctly. To see whether this difference in recall accuracy could have influenced the nonsignificant effect of source, the source answers were recoded as either 0 (wrong) or 1 (correct). The same MANOVA was run again, this time including the source score variable as a covariate. Results show that, even after controlling for the different source memory scores, the interaction effect of source was not significant,
Wilks’ λ = .99, F(3, 167) = .73, p = .534, partial η2 = .01. Based on the nonsignificant
interaction effect of source, this variable was left out of the subsequent tests of the overall model.
Test of the complete model
In order to test hypotheses 3 and 4 that predicted the effect of resistance on attitude towards the health topic and behavioural intention, and to test the overall mediation model, the Process macro by Hayes (2013) for SPSS was used with model number 4. The process analysis was run twice, once with the mean attitude score as the outcome variable, and once with the intention score as the outcome variable. In both analyses, language was the
independent variable (X), and mean scores on the three resistance strategies served as the mediating variables (M). Source as a moderator was left out of the model test based on the nonsignificant findings from the MANOVA.
Overall, the test of the full model with three mediators was significant with both attitude as an outcome measure, F(4, 169) = 8.98, p < .001, R2 = .18, and intention, F(4, 169) = 9.96, p < .001, R2 = .19. Next to the effect of language on the three resistance strategies, which was already confirmed in the previous analysis, the process analysis tested the effect of the three resistance strategies on attitude and intention. The results demonstrate that this effect was significant only for counter-arguing on attitude, b = -.23, t(169) = -3.39, p < .001, and counter-arguing on intention, b = -.27, t(169) = -3.40, p < .001. These results indicate that
when counter-arguing scores increase with 1 unit, attitude and intention decrease, which was as expected by hypothesis 3 and 4. For source derogation, no significant effect was found on attitude, b = -.13, t(169) = -1.30, p = .195, as well as on intention, b = -.22, t(169) = - 1.83, p = .069. Lastly, negative affect did not significantly influence attitude, b = .01, t(169) = .09, p = .930, and intention b = .11, t(169) = .93, p = .356.
These results are reflected by the indirect effects with bias-corrected bootstrap confidence interval, where it is revealed that there is a significant indirect effect only for counter-arguing as a mediator, indirect effect .21, 95% BCBCI [0.10, 0.46] for attitude, and indirect effect .24, 95% BCBCI [0.08, 0.42] for intention. As there was no significant indirect effect of source derogation and negative affect as mediating variables, these mediators were dismissed. From this point, the mediation analysis was reduced to a more simple model, with only counter-arguing as a mediator.
The mediation analysis was run again, for attitude and for intention, to see whether there was a true mediating effect of counter-arguing on the relationship between language and attitude or intention. There was a significant total effect of language on attitude, F(1, 172) = 4.94, p = .028, R2 = .03, and language on intention, F(1, 172) = 7.81, p < .001, R2 = .04.
Figure 2 and 3 visualize both mediation models with the slopes of the effects. Lastly, the bootstrapping confidence intervals for the indirect effect of language through counter-arguing on attitude (indirect effect .25, 95% BCBCI [0.12, 0.44]) and on intention (indirect effect .29, 95% BCBCI [0.14, 0.48]) point out a significant mediating effect of counter-arguing in both models.
Figure 2. Mediation model with attitude
Note. Between parentheses is the total effect of language on attitude. *p < .05, ** p < .01, ***p < .001
Figure 3. Mediation model with intention
Note. Between parentheses is the total effect of language on intention. *p < .05, ** p < .01, ***p < .001
As can be seen in both models, the direction of the total effect and the direct effect are surprising, as it reveals that when language is polite, attitude and intention scores decrease, which is opposite of what was expected. On the other hand, the indirect effect, through counter-arguing, is positive. This suggests a partial mediation model, where controlling language has a negative effect on counter-arguing (more counter-arguing), which in turn has a negative effect on attitude and intention. As the direct effect and the total effect are negative, however, it can be assumed there is another factor activated by controlling language that positively influences attitudes and intention. This might explain the negative direct effect, and
why the total effect is less negative, accounting for both the negative direct effect and the positive indirect effect.
Implicit resistance
The final part of the analyses involved testing whether the word search responses could reveal any implicit uses of resistance, and whether this increased after exposure to the health message. It was also investigated whether participants in different conditions differed significantly in terms of their implicit resistance scores. The scores that were calculated for each word search represented the amount of resistance words found most quickly for each participant, for each word search.
A mixed between-within subjects analysis of variance revealed that the mean score of resistance was higher after exposure to the health message (M = 2.30 , SD = .88) than before exposure (M = 1.26 , SD = .95). Although this main effect of measurement time on the resistance score was significant, Wilks’ λ = .56, F(1, 170) = 132.43, p < .001, this effect could not be explained by the type of language used, Wilks’ λ = .99, F(1, 170) = 2.26, p = .134. The difference between the mean scores could also not be explained by the source of the message, Wilks’ λ = 1.00, F(1, 170) = .02, p = .891. Lastly, the interaction between language and source did also not account for the differences in means for word search 1 and word search 2, Wilks’ λ = .99, F(1, 170) = .83, p = .364. An alternative score was calculated based on the amount of resistance words found as a proportion of the total amount of words found. However, analyses using this score also generated similar, insignificant results. This suggests that the difference in resistance scores for the two word search puzzles was likely to be caused by another factor that was not manipulated within the message. The results also call into question to what extent resistance can be measured implicitly, and whether a higher score truly represents higher resistance.
Discussion
The present study has proven, once again, that resistance is a complex psychological process that is difficult to predict and adequately operationalise. The experiment examined the extent to which controlling versus polite language as a message element influences audience resistance, and how a for-profit or non-profit source could alter this relationship. It was also investigated what role resistance plays in the relationship between language type and the message outcomes attitude and behavioural intention. Lastly, the present study was one of the first to explore the possibility of measuring resistance implicitly, based on the assumption that resistance is a psychological, complex process that is likely to be sensitive to explicit
measurement flaws such as social desirability and post-rationalization (VandeBerg et al., 2016; Zuwerink-Jacks & Cameron, 2003). The results of the experiment have uncovered several valuable insights regarding the central questions of the study.
The analyses demonstrated a negative effect of controlling language on the use counter-arguing, source derogation, and negative, thereby confirming hypothesis 1. This finding is in line with previous studies on language use and resistance (Jenkins & Dragojevic, 2011; Miller et al., 2007; Quick & Considine, 2008). It is clear that a message that makes use of forceful and demeaning language motivates people to report more critical thoughts towards the message, the source, and to report more negative emotions. This result shows there was a considerable amount of resistance against the controlling language message, as opposed to the polite message.
Although part of the resistance was found to be directed towards the source of the message, this appeared to be based on something other than the source’s motives, contrary to what was predicted by hypothesis 2. Although source was manipulated so that there was a distinction between a sales motive (for-profit) and giving advice (non-profit), both sources generated similar scores for resistance. As was predicted based on the Persuasion Knowledge
Model, receivers of a persuasive message quickly turn to the source of a persuasive message as a point of criticism (Friestad and Wright, 1994). Here it is shown that receivers were indeed critical of the source, but this seemed regardless of the source’s motives. The PKM formulates that source evaluations can be based on different elements, such as the motive of the source, but also the tactics that the source uses to persuade (Friestad and Wright, 1994). It is likely to assume that in this particular case, it was not the source itself, but rather the way in which the source was trying to persuade, that induced resistance. Hypothesis 2 was therefore rejected.
Another possible explanation for the insignificant interaction effect of source is that the manipulation of non-profit and for-profit source was not successful. As the manipulation was not pre-tested, it cannot be validated that respondents identified the sources as clearly for-profit or non-for-profit with the corresponding motives, as was intended. The unsuccessful manipulation was potentially a result of the difference in source recall between the two sources, as discussed in the results section. Compared to Beter Horen, Stichting HoorMij is a relatively unknown institution in the Netherlands, and was therefore poorly recalled as the source of the message. Despite the fact that the interaction of source was not significant even when the difference in recall was controlled for, future studies that aim to measure the
relationship between message source and resistance should opt for either equally well-known or equally unknown sources in order to avoid this difference from influencing results.
Model test
The test of the overall mediation model produced several interesting results. First of all, counter-arguing was shown to be the only resistance strategy significantly impacting attitude and intention. Hypothesis 3 and 4, predicting that increased resistance would
negatively influence attitude and intention, were therefore only supported for counter-arguing as a resistance strategy. While respondents reported to make use of source derogation and
negative affect, this did not significantly affect further message outcomes. The nature of the three resistance strategies could have accounted for the difference in effects. Counter-arguing is identified as a message-oriented strategy, whereas source derogation and negative affect are strategies based on affective evaluations of the message (Zuwerink-Jacks & Cameron, 2003). This means counter-arguing is focused on the information provided by the message, whereby it is likely that legitimate counter-arguments will negatively influence evaluations of the message and the position it is advocating. On the other hand, source derogation and negative affect are focused more on the heuristic cues of the message, suggesting that these strategies do not necessarily influence evaluations of the advocated position, but rather evaluations of the message itself. Even though various resistance strategies are often employed
simultaneously in response to a persuasive message, cognitive resistance, and specifically counter-arguing, is found to be most effective in resisting a persuasive appeal. Source derogation is, although a common resistance strategy, found to be an ineffective means of resistance in terms of reducing persuasion (Zuwerink-Jacks & Cameron, 2003).
The alternative mediation model, testing the mediating effect of counter-arguing on the relationship between language and message outcomes, provided evidence of counter-arguing as a mediating factor between language type and attitude and intention. Until now, studies have shown that controlling language is perceived as more threatening, and perceived threat was shown to negatively influence message evaluations and message outcomes
(Jenkins & Dragojevic, 2011; Miller et al., 2007; Quick & Considine, 2008). However, this study is one of the first to demonstrate the indirect effect of controlling language on the
message outcomes attitude and intention, through the mediating effect of cognitive resistance. The alternative model also revealed a significant direct effect of language on attitude and intention, however with an unexpected direction. Compared with polite language, controlling language appeared to influence attitude and intention more positively. This
suggests that the relationship between language and attitude and intention is influenced by another factor or multiple factors that produce(s) a reverse effect compared to the mediation effect. Even though alternative explanations are only speculations, it is interesting to consider possible factors that could account for this effect.
For example, results showed that the mean scores for attitude and intention in the experiment were relatively high for both language conditions. It can be assumed the stimulus message was therefore a pro-attitudinal message for most respondents. Generally, resistance studies make use of counter-attitudinal messages to maximize the likelihood that people will experience reactance (Zuwerink & Devine, 1996). A counter-attitudinal message will evoke a negative bias, but when the message is consistent with existing attitudes, people will be biased favourably (Petty & Briñol, 2008). It is likely that message effects on resistance and on
subsequent message outcomes are different when the message does not oppose existing views. The assumption that the stimulus message was in line with respondents’ existing attitudes and intentions might explain why controlling language did not negatively influence attitudes and intentions as predicted. Only when people disagreed with the message, and were able to generate legitimate counter-arguments, did this effectively influence their attitude and intention. This unexpected effect exposes an interesting area for future research, where it should be investigated how other factors such as prior attitude or attitude importance can moderate the effect of different message elements.
So far, only one recent study has demonstrated the boundaries of the negative effect that controlling, assertive language can have on message responses (Baek, Yoon & Kim, 2015). This study identified effort investment as an important factor moderating the
relationship between language and attitudes and intentions. Effort investment, distinguishing between high-effort investment and low-effort investment, indicates to what extent receivers are already willing to make an effort towards the desired behaviour and reaching the
advocated goal. In this study, this would suggest that people who have already made an attempt at protecting their hearing, or who were already stimulated to do so, were more likely to attend to a message that uses assertive, controlling language, as they are already convinced that the advocated behaviour is worth the effort. Low-effort invested people are not yet convinced that the effort will be worthwhile, and will comply more with a polite message that emphasizes freedom of choice (Baek et al., 2015). Effort investment is one potential factor that could have accounted for the positive effect of controlling language on attitude and intention in the current study. However, the study described above focused on messages promoting environmentally friendly behaviour. It cannot be said whether these results will be similar when applied to health behaviour, leaving this an interesting gap for future
researchers.
Implicit resistance
Finally, this study attempted to add to existing resistance literature by exploring the possibility of measuring resistance implicitly, using a new, exploratory measure. Existing literature on resistance has operationalised resistance in various ways, but so far no record of an implicit measure has been found, while there is little proof that people can accurately recall the way they resisted a message. Using a measure that was based on immediate, unconscious responses aimed to uncover the unconscious process at work when people resist a persuasive message. Unfortunately, this study could not provide evidence of this unconscious process. The resistance that respondents reported to have explicitly, was not found by the implicit measure. Even though implicit resistance scores increased after exposure to the message, it cannot be confirmed this was a result of exposure to the message. Based on this study, it can be stated that the increase in resistance scores between the two word search puzzles was not caused by language type or message source. The second word search was perhaps simply easier than the first, or respondents recognized the resistance words more easily after being
primed by the first word search puzzle. It nevertheless remains of great relevance for
researchers to explore the ways in which resistance to persuasive messages can be measured, because only when resistance is effectively measured, can research provide valuable practical implications to health message designers. For example, other implicit measures, such as the Implicit Association Test (Vandeberg et al., 2016) could be adapted to measure the construct of resistance, making use of even more unconscious, immediate responses.
Limitations & directions for future research
The findings of this research are valuable in several regards, yet some limitations should be noted before any practical implications are considered. Firstly, most resistance is generally expected among young populations (Miller et al., 2007). The mean age in the present study was relatively high, making it difficult to provide practical implications for message design aimed at a more young, resistant age group. Health campaign developers still need to get a good understanding of their target group and the way they are likely to resist information before designing a message.
Secondly, as attitude and intention were measured only after exposure to the message, no inferences could be made about changes in attitude and intention as a result of the health message. A valuable addition to future research would be a test of the current model, with the inclusion of a before- and after-measure of message outcomes in order to better assess the effect of the exposure. This way the use of resistance strategies can be more efficiently related to a potential change in attitude and intention.
Lastly, this study could not establish an effect of message source on resistance, likely to be the result of an unsuccessful manipulation. The effects that different types of
organisations or institutions as message source can have on resistance nonetheless remains an understudied topic that future researchers should continue to explore, so that campaign
developers can make more informed decisions about the way the source of a message should be presented, and how source interacts with other message elements.
Conclusion
The present study aimed to provide new insights regarding resistance that could be of great value for campaign developers in the health domain, as resistance is still an issue of great concern, especially when targeting young groups. This study has shown that even when the advocated position is likely to be consistent with viewpoints of the target group, the message can still infer resistance. Language type is an important consideration in message design, and controlling language should generally be avoided. Counter-arguing as a resistance strategy should be anticipated, for example by making use of two-sided messages (Knowles & Linn, 2004), as this strategy is an effective resistance strategy for receivers of a message, that can negatively influences the objectives of the message.
References
Ahluwalia, R. (2000). Examination of psychological processes underlying resistance to persuasion. Journal of Consumer Research, 27(2), 217-232.
Asbeek Brusse, E. D., Fransen, M. L., & Smit, E. G. (2015). The More You Say, the Less They Hear: The Effect of Disclosures in Entertainment-Education. Journal of Media Psychology: Theories, Methods, and Applications, 27(4), 159-169.
Atkin, J. L., McCardle, M. & Newell, S. J. (2008). The role of advertiser motives in
consumer evaluations of ‘responsibility’ messages from the alcohol industry. Journal of Marketing Communications, 14(4), 315-335.
Baek, T. H., Yoon, S., & Kim, S. (2015). When environmental messages should be assertive: examining the moderating role of effort investment. International Journal of
Brehm, J. W. (1989). Psychological reactance: Theory and applications. Advances in consumer research, 16(1), 72-75.
Burgoon, M., Alvaro, E., Grandpre, J., & Voulodakis, M. (2002). Revisiting the theory of psychological reactance. The persuasion handbook, 213-232.
Campbell, M. C., & Kirmani, A. (2000). Consumers' use of persuasion knowledge: The effects of accessibility and cognitive capacity on perceptions of an influence agent. Journal of Consumer Research, 27(1), 69-83.
Crano, W.D. & Prislin, R. (2006). Attitudes and Persuasion. Annual Review of Psychology, 57, 345-374.
Crossley, M. L. (2002). Introduction to the Symposium'Health Resistance': The limits of contemporary health promotion. Health Education Journal, 61(2), 101-112.
Dillard, J. P. & Shen, L. (2005). On the nature of reactance and its role in persuasive health communication. Communication Monographs, 72(2), 144-168.
Donnell, A. J., Thomas, A. & Buboltz Jr, W. C. (2001). Psychological reactance: Factor structure and internal consistency of the questionnaire for the measurement of psychological reactance. The Journal of Social Psychology, 141(5), 679-687. Fazio, R. H. & Olson, M. A. (2003). Implicit measures in social cognition research: Their
meaning and use. Annual Review of Psychology, 54(1), 297-327.
Fransen, M., Verlegh, P., Kirmani, A. & Smit, E. (2015). A typology of consumer strategies for resisting advertising, and a review of mechanisms for countering them.
International Journal of Advertising: The Review of Marketing Communications, 34(1), 6-16.
Friestad, M. & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research, 21(1), 11-31.
meta-analysis of measures, Health Psychology Review, 1(2), 208-229.
Jenkins, M. & Dragojevic, M. (2011). Explaining the process of resistance to persuasion: A politeness theory-based approach, Communication Research, 1-31. doi:
10.1177/0093650211420136
Jones, L. W., Sinclair, R. C. & Courneya, K. S. (2003). The effects of source credibility and message framing on exercise intentions, behaviors, and attitudes: An integration of the elaboration likelihood model and prospect theory1. Journal of Applied Social
Psychology, 33(1), 179-196.
Knowles, E. S., & Linn, J. A. (Eds.). (2004). Resistance and persuasion. Psychology Press. Miller, C. H., Lane, L. T., Deatrick, L. M., Young, A. M. & Potts, K. A. (2007).
Psychological reactance and promotional health messages: The effects of controlling language, lexical concreteness, and the restoration of freedom. Human Communication Research, 33(2), 219-240.
Miron, A. M. & Brehm, J. W. (2006). Reactance theory-40 years later. Zeitschrift für Sozialpsychologie, 37(1), 9-18.
Paek, H. J., Hove, T., Ju Jeong, H. & Kim, M. (2011). Peer or expert? The persuasive impact of YouTube public service announcement producers. International Journal of
Advertising, 30(1), 161-188.
Petty, R. E. & Briñol, P. (2008). Psychological Processes Underlying Persuasion A Social Psychological Approach. Diogenes, 55(1), 52-67.
Petty, R. E. & Cacioppo, J. T. (1984). Source factors and the elaboration likelihood model of persuasion. Advances in Consumer Research, 11(1), 668-672.
Petty, R. E., Fazio, R. H. & Briñol, P. (Eds.). (2012). Attitudes: Insights from the new implicit measures. Psychology Press.
decades' evidence. Journal of Applied Social Psychology, 34(2), 243-281. Quick, B.L. & Considine, J.R. (2008) Examining the Use of Forceful Language When
Designing Exercise Persuasive Messages for Adults: A Test of Conceptualizing Reactance Arousal as a Two-Step Process. Health Communication, 23(5), 483-491. Rains, S. A. & Turner, M. M. (2007). Psychological reactance and persuasive health
communication: A test and extension of the intertwined model. Human Communication Research, 33(2), 241-269.
Smith, B. J., Bauman, A. E., McKenzie, J. & Thomas, M. (2005). Awareness of message source and its association with the impacts of sun protection campaigns in Australia. Health Education, 105(1), 42-52.
Vandeberg, L., Wennekers, A. M., Murre, J. M., & Smit, E. G. (2016). Implicit and Explicit Measures: What Their Dissociations Reveal About the Workings of Advertising. Advances in Advertising Research, 6, 269-279.
Whitehead, D. & Russell, G. (2004). How effective are health education programmes— resistance, reactance, rationality and risk? Recommendations for effective practice. International Journal of Nursing Studies, 41(2), 163-172.
Zuwerink-Jacks, J. & Cameron, K. A. (2003). Strategies for resisting persuasion. Basic and Applied Social Psychology, 25(2), 145-161.
Zuwerink, J. R. & Devine, P. G. (1996). Attitude importance and resistance to persuasion: It's not just the thought that counts. Journal of Personality and Social Psychology, 70(5), 931.
Appendix A
Introduction to the experiment
Beste deelnemer,
Hierbij wil ik je uitnodigen om deel te nemen aan een onderzoek dat wordt uitgevoerd onder verantwoordelijkheid van de Graduate School of Communication, onderdeel van de
Universiteit van Amsterdam.
Het onderzoek waar je aan mee gaat doen gaat over de effecten van gezondheidscampagnes. In dit online experiment zal je een voorbeeld van een campagne boodschap zien, en
verschillende soorten vragen krijgen, waaronder taakjes en puzzels. Het doel van dit
onderzoek is om meer te weten te komen over de evaluaties van gezondheidsboodschappen, en de effecten hiervan.
Het onderzoek duurt ongeveer 10 minuten.
Omdat dit onderzoek wordt uitgevoerd onder de verantwoordelijkheid van ASCoR, Universiteit van Amsterdam, heb je de garantie dat:
1) Je anonimiteit is gewaarborgd en dat je antwoorden of gegevens onder geen enkele voorwaarde aan derden worden verstrekt, tenzij je hiervoor van te voren uitdrukkelijke toestemming hebt verleend.
2) Je zonder opgaaf van redenen kunt weigeren mee te doen aan het onderzoek of je deelname voortijdig kunt afbreken. Ook kun je achteraf (binnen 24 uur na deelname) je toestemming intrekken voor het gebruik van je antwoorden of gegevens voor het onderzoek.
3) Deelname aan het onderzoek geen noemenswaardige risico’s of ongemakken met zich meebrengt, geen moedwillige misleiding plaatsvindt, en je niet met expliciet aanstootgevend materiaal zult worden geconfronteerd.
4) Je uiterlijk vijf maanden na afloop van het onderzoek de beschikking kunt krijgen over een onderzoeksrapportage waarin de algemene resultaten van het onderzoek worden toegelicht. Voor meer informatie over dit onderzoek en de uitnodiging tot deelname kun je te allen tijde contact opnemen met projectleider Emmy Bruijstens (emmy.bruijstens@student.uva.nl) of Marieke Fransen (M.L.Fransen@uva.nl). Als je graag de resultaten van het onderzoek ontvangt, neem dan ook contact op met Emmy Bruijstens.
Mochten er naar aanleiding van je deelname aan dit onderzoek klachten of opmerkingen bij je zijn, dan kun je contact opnemen met het lid van de Commissie Ethiek van de afdeling
Communicatiewetenschap, per adres: ASCoR secretariaat, Commissie Ethiek, Universiteit van Amsterdam, Postbus 15793, 1001 NG Amsterdam; 020‐525 3680; ascor‐secr‐
fmg@uva.nl. Een vertrouwelijke behandeling van je klacht of opmerking is daarbij gewaarborgd.
Ik hoop je hiermee voldoende te hebben geïnformeerd, en alvast hartelijk bedankt voor je deelname aan dit onderzoek!
Met vriendelijke groet, Emmy Bruijstens
Appendix B Filler tasks
Word completion task – neutral words 1. T_L_ _ O O _ 2. P I _ N _ 3. _ A P _ O P 4. S T R A _ _ 5. _ O O _ 6. S _ _ _ U W 7. _ C H _ T 8. _ O E _ E R 9. _ E E K 10. K _ A _ S
Welk cijfer komt op de plek van het vraagteken? 4 – 8 – 16 – 32 – ?
4 – 6 – 9 – 6 – 14 – 6 – ? 8 – 16 – 20 – 40 – ? 10 – 20 – 15 – 25 – ?
Appendix C
Items per observed variable
Counter-arguing “Tijdens het lezen van de boodschap had ik
kritiek op de inhoud hiervan”
“Tijdens het lezen van de boodschap kwamen er tegenargumenten in me op”
“Ik was sceptisch over de boodschap toen ik deze las”
Source derogation
“Geef hieronder aan wat jouw mening is over de zender van deze boodschap. Ik vind de zender van deze boodschap:”
Expertise
Onwetend/wijs
Slecht geïnformeerd/goed geïnformeerd Onintelligent/intelligent Ongeschikt/geschikt Likeability Onbetrouwbaar/betrouwbaar Onvriendelijk/vriendelijk Sincerity Oprecht/onoprecht Eerlijk/oneerlijk Niet manipulatief/manipulatief Niet opdringerig/opdringerig Negative affect
“Geef hieronder aan hoe je je voelde toen je de boodschap las.
Toen ik de boodschap las, voel de ik me:”
Boos Geïrriteerd Geërgerd Kwaad
Attitude towards the topic
“Geef hieronder aan hoe je denkt over het dragen van oordopjes ter bescherming van je gehoor:” Negatief/positief Onwenselijk/wenselijk Onnodig/nodig Ongunstig/gunstig Slecht/goed Onverstandig/verstandig
Behavioural intention “Hoe waarschijnlijk is het dat jij oordopjes
zal dragen de volgende keer dat je naar een evenement met harde muziek gaat?”
Appendix D
D.1 Stimulus text + translation
Controlling language (Dutch)
“Harde muziek beschadigt je gehoor voorgoed. Je moet nu ingrijpen, voordat het te laat is. Draag oordopjes de volgende keer dat je blootgesteld wordt aan harde muziek. Iedereen die nu geen oordopjes koopt om zichzelf te beschermen is simpelweg roekeloos. Koop nu oordopjes!”.
English translation
“Loud music damages your hearing for good. You have to act now, before it’s too late. Wear earplugs next time you are exposed to loud music. Anyone who does not buy earplugs now to protect themselves is simply reckless. Buy earplugs now!”
Polite language (Dutch)
“Harde muziek kan je gehoor voorgoed beschadigen. Zou je niet nu ingrijpen, voordat het te laat is? Probeer eens om oordopjes te dragen de volgende keer dat je wordt blootgesteld aan harde muziek. Als je jezelf wilt beschermen, is het belangrijk om oordopjes aan te schaffen. Probeer het eens!”
English translation
“Loud music can damage your hearing for good. Why not act now, before it’s too late? Try wearing earplugs next time you are exposed to loud music. If you want to protect yourself, it is important to purchase earplugs. Try it!”
D.2 Pre-test: Controlling vs. polite language
Controlling Polite
Agressief Vriendelijk/beleefd
Aanvallend Vriendelijk advies
Goed Aardig/voor je eigen bestwil/informatief
Hard/aanstootgevend Aardig
Overdreven/dwingend Informatief
Agressief/verwijt Meedenkend/aardig
Alsof je moeder tegen je praat Beleefd
Autoritair/dwingend Bijna té aardig/beleefd
Moeten/dwingend Goed advies/iets om over na te denken
Onaardig/bemoeizuchtig Goed
D.3 Stimulus material per condition
Condition 2: Non-profit source/polite language
Condition 4: For-profit source/polite language
Appendix E
Words included in word search Source
derogation
Counter-arguing Negative affect Positive Neutral
Manipulatief Reclame Commercieel Verkopen Oneens Overdreven Tegenspreken Irritant Boos Ergerlijk Betrouwbaar Eerlijk Oprecht Aardig Realistisch Zanger Water Ontbijt Stopcontact Bedorven
