How the persuasiveness of statistical evidence compared to personal testimonials depends on the recipient's distance from the message issue

Wieluch, Marina I.; Praxmarer‐Carus, Sandra Article — Published Version How the persuasiveness of statistical evidence compared to personal testimonials depends on the recipient's distance from the message issue Journal of Consumer Behaviour Provided in Cooperation with: John Wiley & Sons Suggested Citation: Wieluch, Marina I.; Praxmarer‐Carus, Sandra (2024) : How the persuasiveness of statistical evidence compared to personal testimonials depends on the recipient's distance from the message issue, Journal of Consumer Behaviour, ISSN 1479-1838, Wiley Periodicals, Inc., Hoboken, NJ, Vol. 24, Iss. 1, pp. 75-91, https://doi.org/10.1002/cb.2404 This Version is available at: https://hdl.handle.net/10419/313749 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ RESEARCH ARTICLE How the persuasiveness of statistical evidence compared to personal testimonials depends on the recipient's distance from the message issue Marina I. Wieluch | Sandra Praxmarer-Carus Universität der Bundeswehr München, Fakultät für Wirtschaftsund Organisationswissenschaften, Professur für Allgemeine Betriebswirtschaftslehre, insbesondere Marketing, Neubiberg, Germany Correspondence Sandra Praxmarer-Carus, Universität der Bundeswehr München, Fakultät für Wirtschaftsund Organisationswissenschaften, Professur für Allgemeine Betriebswirtschaftslehre, insbesondere Marketing, Werner-HeisenbergWeg 39, 85577 Neubiberg, Germany. Email: [email protected] Funding information Deutscher Verein für Versicherungswissenschaft e. V. (only Study 2) Abstract Statistical facts and personal testimonials (anecdotal evidence) are two common types of evidence used in health education, warning messages, and charitable appeals. While statistical evidence provides aggregated numerical information, messages that use anecdotal evidence typically describe an individual experience. Because research has not found a stable advantage of one type of evidence over the other, the literature has sought to identify moderators that predict when statistical or anecdotal evidence is more persuasive. This paper shows that the relative persuasiveness of statistical versus anecdotal evidence depends on the psychological distance between the message recipient and the message issue. An increase (decrease) in recipients' message-issue distance increases the relative persuasiveness of statistical (anecdotal) evidence. In addition, we show that message-issue distance determines how personally useful message recipients find statistical and anecdotal evidence. We also demonstrate that recipients' more abstract (concrete) thinking about the message issue prior to message exposure increases the persuasiveness of statistical (anecdotal) evidence. Based on our findings, we recommend that social marketers use statistical (anecdotal) evidence when the recipients' distance from the message issue is high (low) and the recipients' thinking about the message issue is abstract (concrete). Before deciding on the type of evidence, message designers may need to assess how abstract or concrete their target audience thinks about the message issue. The short measure used in Experiment 1b of this paper may be useful. It is adaptable to different message contexts and could easily be implemented in pretests to decide when to use statistical or anecdotal evidence. 1|INTRODUCTION Messages in health education, risk communication and charity advertising often use statistical evidence or personal testimonials to support their claims (Small et al., 2007; Zebregs et al., 2015). While statistical evidence provides aggregated numerical information, such as “30% of U.S. cancer deaths can be attributed to tobacco use” (Truth Initiative, 2022), messages that use personal testimonials typically describe an individual's experience, such as “John, 54, long-term smoker, has been diagnosed with lung cancer ….”The literature has used various terms for the latter type of information, most commonly anecdotal evidence (Cox & Cox, 2001), narrative evidence (Ma & Received: 18 May 2023 Revised: 4 July 2024 Accepted: 2 September 2024 DOI: 10.1002/cb.2404 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Author(s). Journal of Consumer Behaviour published by John Wiley & Sons Ltd. J Consumer Behav. 2025;24:75–91. wileyonlinelibrary.com/journal/cb 75 Nan, 2018), exemplars (Brosius & Bathelt, 1994), personal testimonials (De Wit et al., 2008), and identifiable lives (Slovic, 2007). This paper uses the term anecdotal evidence. Numerous studies have compared the effectiveness of statistical and anecdotal evidence (see Freling et al., 2020 for a review). While some studies report an advantage of statistical evidence over anecdotal evidence (e.g., Hornikx, 2005), other studies report the opposite (e.g., Winterbottom et al., 2008). Because research has not observed a general advantage of one type of evidence over the other, the literature has sought to identify moderators that predict when statistical or anecdotal evidence is more persuasive (e.g., Freling et al., 2020). Surprisingly, in the search for explanations of when statistical or anecdotal evidence has a persuasive advantage, the literature has paid relatively little attention to the psychological distance of the message issue from the message receiver (Trope & Liberman, 2010). We use the term message issue to refer to the problem or appeal that the message raises. To support their appeals, messages may present statistical or anecdotal evidence. For example, a fundraising appeal for malnourished children in a particular country (message issue) may feature percentages of malnourished children in that country (statistical evidence) or a single child (anecdotal evidence). Thus, from the perspective of message designers, the choice of evidence type may be independent of the message issue. However, we argue that the effectiveness of anecdotal and statistical evidence depends on the psychological distance of the message issue from the message receiver, and their more concrete or abstract thinking about the message issue. Drawing on the positive effects of construal-level fit (e. g., Dogan & Erdogan, 2020; Park et al., 2020; Reczek et al., 2018), we propose that messages that combine psychologically close message issues with anecdotal evidence and psychologically distant message issues with statistical evidence should be more persuasive than messages that combine close message issues with statistical evidence and distant message issues with anecdotal evidence. Ma and Nan (2018) and Kim and Nan (2019) put forward a very similar proposition. Their reasoning, and ours, is based on the arguments that recipients' message-issue distance is linked to constructing more highlevel/low-level representations of the message issue (Soderberg et al., 2015; Trope & Liberman, 2010) and that statistical evidence tends to be high-level information (Hastall & Knobloch-Westerwick, 2013), while anecdotal evidence tends to be low-level information (Kim & Nan, 2019; Yan & Sengupta, 2013). Relatively few studies have examined the moderating effect of recipients' message-issue distance on the effects of anecdotal and statistical evidence on behavior or behavioral intentions. Ma and Nan (2018) and Kim and Nan (2019) tested the interaction. However, Ma and Nan (2018) found no differences in the resulting intentions, and Kim and Nan (2019) did not show each of the absolute advantages they predicted. Thus, the empirical evidence for the expected interaction seems sparse, which may lead researchers to question the general proposition. Because the proposed interaction between evidence type and message-issue distance is theoretically convincing, this paper revisits and reexamines the effect. First, we will explain why the findings of Ma and Nan (2018) and Kim and Nan (2019) do not contradict their rationale and ours, and how this research differs from previous studies. Ma and Nan (2018) used an anecdotal message and a “non-narrative”message and found no differences in the resulting intentions. However, the “non-narrative”message used by Ma and Nan (2018) included details about the harmful effects of smoking. Yet, the interaction predicted here is based on the argument that anecdotal messages tend to be low-level information and statistical messages tend to be aggregated, high-level information. Thus, the “non-narrative” message used by Ma and Nan (2018) may have presented relatively concrete, low-level information rather than global, abstract information, and as such may have been inconsistent with the theoretical rationale the authors were testing. The present paper will use anecdotal and statistical evidence to test the rationale. Kim and Nan (2019) predicted that anecdotal messages would be more effective than nonnarrative messages in a close situation, while nonnarrative messages would be more effective than anecdotal messages in a distant situation. Kim and Nan (2019) found an interaction but did not show each of the absolute advantages they expected. We argue that such absolute advantages are desirable but not necessary to support the theoretical rationale because the manipulated psychological distance is relative. Figure 1illustrates our reasoning. For example, imagine two donation appeals that depict individuals in need. One depicts individuals from the same country as the recipient, and the other depicts individuals from a different country. Typically, the psychological distance between the message and the recipient should be smaller in the first situation (same country) than in the second (foreign country). Closer/less distant, however, does not necessarily mean that the former message issue (same country) is perceived by the receiver as psychologically close in absolute terms, which may be required to result in low-level construal and the predicted absolute advantage of anecdotal over statistical evidence. For an illustration of the above versus versus FIGURE 1 Prototypical effects of anecdotal and statistical evidence as a function of message-issue distance, as proposed in this research (in blue), and two examples of areas that might be considered with message issues of specific experiments (in grey). (As psychological closeness can be related to the personal relevance of the message issue, a main effect of psychological closeness on behavior or behavioral intentions may emerge, which is not illustrated in Figure 1). 76 WIELUCH and PRAXMARER-CARUS example, see the right of the two gray areas in Figure 1. Therefore, we argue that the theoretical rationale proposed by Ma and Nan (2018) and Kim and Nan (2019) and in this paper is already supported if a statistical message, compared to an anecdotal message, is relatively more effective in a distant situation than in a close situation, and an anecdotal message, compared to a statistical message, is relatively more effective in a close situation than in a distant situation. Consequently, this research will propose relative advantages. In addition, this paper tests whether recipients' concrete/abstract thinking about the message issue moderates the persuasiveness of statistical and anecdotal evidence. By explaining why the findings of the studies by Ma and Nan (2018) and Kim and Nan (2019) do not contradict their theoretical rationale, and by retesting the moderation with messages that align better with the theoretical explanation, our research contributes to theory and to a better understanding of the persuasiveness of anecdotal and statistical evidence. We explain and demonstrate that recipients' message-issue distance (Study 1a) and recipients' concrete/ abstract thinking about the message issue (Study 1b) determine the effectiveness of anecdotal and statistical evidence. This paper also shows that message-issue distance determines how personally useful message recipients perceive statistical and anecdotal evidence to be (Study 2). This article is organized as follows. First, we describe the basic characteristics of anecdotal and statistical evidence. We then explain how the persuasiveness of statistical and anecdotal evidence should depend on the psychological distance of the message recipient from the message issue and on the recipient's concrete/abstract thinking about the message issue. In addition, we explain why the perceived usefulness of the information should mediate the effect. Subsequently, we present three experiments. Finally, we discuss the findings of our studies and provide suggestions for practice and future research. 2|THE PERSUASIVENESS OF STATISTICAL AND ANECDOTAL EVIDENCE Both statistical and anecdotal evidence have specific strengths, and both types of evidence can persuade message recipients (Freling et al., 2020). Statistical evidence consists of a large number of representative cases. As such, it is highly predictive and is often considered more informative than anecdotal information (Greene & Brinn, 2003; Hoeken & Hustinx, 2009). Anecdotal evidence is vivid, specific, and easily imagined, and can make an issue feel closer and more relevant to message recipients (De Wit et al., 2008; Zillmann & Brosius, 2012). Furthermore, identifiable victims (anecdotal evidence) are emotionally engaging, whereas statistical information does hardly evoke affect (Cox & Cox, 2001; Slovic, 2007). As a result, anecdotal information is processed less analytically than statistical evidence (Escalas, 2007; Slovic, 2007; Small et al., 2007) and generates fewer counterarguments (Kang et al., 2020; Slater & Rouner, 1996). Consistent with the specific strengths of each type of evidence, studies that have compared the persuasiveness of statistical and anecdotal evidence report mixed results. While some studies show an advantage of statistical evidence over anecdotal evidence (e.g., Hoeken & Hustinx, 2009), other studies show the opposite (e.g., De Wit et al., 2008; Greene et al., 2010). Because of the inconsistent findings, the literature has sought to identify moderators that may predict when statistical or anecdotal evidence is more persuasive. Research has suggested several moderators, including value congruency (Slater & Rouner, 1996), message framing (e.g., Cox & Cox, 2001), and personal relevance (e.g., Freling et al., 2020). For literature reviews, see Freling et al. (2020) and Lee and Feeley (2016). This paper examines a moderator that has received relatively little attention in the literature, namely, the psychological distance between the message recipient and the message issue. Below we explain how the effectiveness of anecdotal and statistical evidence should depend on the psychological distance of the message recipients from the message issue. 3|HOW THE PERSUASIVENESS OF STATISTICAL AND ANECDOTAL EVIDENCE MAY DEPEND ON MESSAGE-ISSUE DISTANCE 3.1 |The moderating role of receivers' messageissue distance 3.1.1 | Theoretical explanations Message issues may be more or less psychologically distant from the message recipient (Trope & Liberman, 2010). Examples of message issues that may evoke different levels of psychological distance include risks to self-versus others, consequences that will occur today versus years from now, and near versus far regions for donation appeals (Soderberg et al., 2015). According to Construal Level Theory, individuals tend to form high level, that is, general and abstract mental representations, when an issue is psychologically distant (Liberman & Trope, 1998). In contrast, when an issue is psychologically close, individuals tend to form low level, that is, specific and detailed mental representations. Many studies have shown the link between psychological distance and how concretely or abstractly individuals think about an event or object (e.g., Fujita et al., 2006; Liviatan et al., 2008; Wakslak et al., 2006, see Soderberg et al., 2015 for a meta-analysis). The literature has also shown that information that aligns with consumers' high-level or low-level thinking is easier to process and more persuasive (Carter et al., 2020; Ein-Gar & Levontin, 2013; Han et al., 2016; Lee et al., 2010; Park et al., 2020; White et al., 2011). Thus, among message recipients who are confronted with a psychologically close message issue and engage in low-level thinking, supporting evidence that is specific and concrete should be more persuasive than supporting evidence that is global and on a superordinate level. Among recipients who are confronted with a psychologically distant message issue, supporting evidence that is global and superordinate should be more persuasive than evidence that is specific, low-level information. WIELUCH and PRAXMARER-CARUS 77 Because anecdotal evidence describes the fate of a single individual, it tends to be specific, low-level information (Kim & Nan, 2019). In contrast, statistical evidence uses aggregated numerical information and is therefore global, high-level information (Hastall & KnoblochWesterwick, 2013). For example, Soderberg et al. (2015, p. 526) describe “exemplars”and “situational information”(typically presented in anecdotal messages) as concrete, low-level descriptions of an event, and “aggregate information”(typically presented in statistical messages) as more abstract, high-level descriptions. Thus, in line with Ma and Nan (2018) and Kim and Nan (2019), we argue that messages that combine message issues that are psychologically close to the recipient with anecdotal evidence and message issues that are distant from the recipient with statistical evidence should create a construal-level fit and be more persuasive than messages that combine close message issues with statistical evidence or distant message issues with anecdotal evidence. 1 The above reasoning may implicitly suggest that message recipients first notice the message issue and then read the supporting evidence that the message provides. This temporal sequence may be true, but it is not relevant to our argument, which is based on the construal-level fit within the message. Our argument remains valid if the recipients of the message first read the anecdotal or statistical evidence and thereby become aware of the message issue. Not only the message issue, but also the type of evidence can determine the level of abstraction of the message recipient's thinking (Soderberg et al., 2015). Thus, when message receivers process anecdotal (statistical) evidence and engage in concrete (abstract) thinking, a close (distant) message issue raised by the message should be more construal-level congruent than a distant (close) issue. In addition, if close message issues are more emotionally engaging than distant message issues (Abraham et al., 2023; Mühlberger et al., 2008; Williams et al., 2014), affective/narrative versus analytical processing (Escalas, 2007) may provide a related explanation for the interaction between evidence type and message-issue distance. Low versus high emotional engagement leads consumers to think more analytically (Blanchette, 2006; Freling et al., 2020). Compared to affective processing, analytical processing is more likely to elicit critical thinking, so persuasion requires strong arguments (Escalas, 2007). Because an anecdotal message presents only one exemplar, its predictive power is weak and it can be perceived as a relatively weak argument (Hoeken & Hustinx, 2009). Taken together, the tendency to engage in analytical thinking should be higher for distant message issues, and here message recipients should be more likely to recognize that the informational value of an anecdotal message is weak. In contrast, when the message issue is close, recipients may be more likely to engage in affective processing and find an anecdotal message relatively meaningful because they are less likely to recognize its low informational value. 3.1.2 | Previous findings Ma and Nan (2018) and Kim and Nan (2019) have examined the effect of evidence type on behavioral intentions as a function of messageissue distance. We discussed both experiments in the Section 1. Kim and Nan (2019) found the proposed interaction but had predicted absolute advantages that their study did not reveal. The nonnarrative message that Ma and Nan (2018) used may not have been high-level information. Furthermore, the study asked nonsmoking students to imagine that either their best friend (close situation) or an average student (distant situation) smoked and to report their intentions to persuade the other person to quit. It is possible that this manipulation of psychological distance did not result in different thinking styles, since 69% of participants reported not having a close friend who smokes. Therefore, as argued earlier, the findings of Kim and Nan (2019) are consistent with the proposed interaction, and the findings of Ma and Nan (2018) do not challenge the proposed effect. Research by Ledgerwood et al. (2010) showed that temporal distance increases the weight consumers place on aggregate compared to individualized information about a product's effectiveness, which is consistent with the moderating effect predicted here. Ledgerwood et al. (2010) did not use anecdotal messages such as those studied here (describing a person's fate). However, the messages are related to the evidence types tested here and to the theoretical reasoning of this paper. Although Yan and Sengupta (2013) did not examine behavioral intentions, their findings are consistent with our rationale because they showed that consumers' reliance on statistical (anecdotal) information increased (decreased) with psychological distance. The meta-analysis by Freling et al. (2020) did not explicitly consider message-issue distance, but its findings are consistent with and related to the interaction proposed in this paper. Freling et al. (2020) report that anecdotal evidence proved more persuasive than statistical evidence when personal relevance and emotional engagement were high, whereas statistical evidence proved more persuasive when personal relevance was low. The personal relevance of a message issue can be related to psychological distance, as closer message issues may feel more relevant to individuals than distant message issues (Soderberg et al., 2015). 3.1.3 | Hypotheses We propose a hypothesis about the relative effectiveness of statistical versus anecdotal evidence (and vice versa), depending on the psychological distance of the message recipient from the message issue. This research examines psychological distance in relative terms because, 1 Our proposition is based on the argument that anecdotal evidence is concrete, low-level information, while statistical evidence is aggregated, high-level information. One reviewer pointed out that statistical evidence can be perceived as concrete information. For example, a probability of 25% means one in four people, which can be perceived as concrete when applied to a group of people. However, we argue that compared to exemplars, statistical information is at an aggregate level, more abstract, and more likely to evoke high-level thinking (Soderberg et al., 2015, p. 526). To further address this issue and potential concern, Experiment 1b of this paper tests whether recipients' more concrete/abstract thinking about the message issue (prior to exposure to the message) determines the persuasiveness of statistical and anecdotal evidence (see H1b). The positive effects of construal-level fit on persuasion have been repeatedly demonstrated (Carter et al., 2020; Ein-Gar & Levontin, 2013; Han et al., 2016; Lee et al., 2010; White et al., 2011). Thus, if recipients think concretely about the message issue and anecdotal evidence is relatively concrete information, then anecdotal evidence should be more persuasive than statistical evidence. If recipients think abstractly about the message issue and statistical evidence is relatively abstract information, then statistical evidence should be more persuasive than anecdotal evidence. 78 WIELUCH and PRAXMARER-CARUS for example, a “less distant”message issue is not necessarily close in absolute terms and therefore may not result in an absolute advantage of anecdotal over statistical evidence (see Figure 1). We propose: H1a. Recipients' message-issue distance moderates the effect of evidence type on compliance intention such that the relative persuasiveness of statistical evidence (compared to anecdotal evidence) increases with increasing message-issue distance, and the relative persuasiveness of anecdotal evidence (compared to statistical evidence) increases with decreasing message-issue distance. Along with Ma and Nan (2018) and Kim and Nan (2019), we have suggested that construal-level fit is a relevant explanation for why recipients' message-issue distance moderates the persuasiveness of statistical and anecdotal evidence. Thus, under the assumption that statistical (anecdotal) evidence is relatively abstract (concrete) information, the recipients' abstract/concrete thinking about the message issue is the actual explanatory variable of the theoretical rationale proposed here. Positive effects of construal-level fit on persuasion have been repeatedly demonstrated (Carter et al., 2020; Ein-Gar & Levontin, 2013; Han et al., 2016; Lee et al., 2010; White et al., 2011). If recipients think concretely about the message issue and anecdotal evidence is concrete information, then anecdotal evidence should be more persuasive than statistical evidence. If recipients think abstractly about the message issue and statistical evidence is abstract information, then statistical evidence should be more persuasive than anecdotal evidence. We propose: H1b. Recipients' concrete versus abstract thinking about the message issue moderates the effect of evidence type on compliance intention such that the persuasiveness of statistical evidence (compared to anecdotal evidence) increases with more abstract thinking, and the persuasiveness of anecdotal evidence (compared to statistical evidence) increases with more concrete thinking about the message issue. 3.2 |The mediating role of the perceived usefulness of the message information Message-issue distance may determine how personally useful message recipients perceive statistical and anecdotal evidence to be. The perceived usefulness of message information is, of course, subjective and context-dependent (DeLorme et al., 2009; Schwarz, 2010). When message information aligns with consumers' high-level or low-level thinking, it should be processed more fluently and less critically than when it does not (Cesario et al., 2004; Schwarz, 2004). For example, although a statistical message may be perceived as informative in principle, the perception of how personally useful such information is to the recipient at a given time may depend on the recipient's situational mindset. Similarly, while anecdotal evidence may be perceived as easy to imagine in principle, the perception of how personally useful such anecdotal information is to the recipient at a given time may depend on the recipient's mindset. A combination of close message issues with anecdotal evidence, as well as a combination of distant message issues with statistical evidence, should make the message “feel right”to the recipient (Cesario et al., 2004), evoke less critical thinking, and make the message feel more personally useful (Ledgerwood et al., 2010; Schwarz, 2004). Furthermore, when a message is processed more analytically due to greater message-issue distance, recipients are more likely to recognize that a single exemplar (anecdotal evidence) is not a strong argument and may perceive the message as less useful than a statistical message. Taken together, the perceived usefulness of statistical versus anecdotal evidence should depend on recipients' message-issue distance. Since the perceived usefulness of the information provided by a message determines persuasion (Schwarz, 2010), perceived usefulness should mediate the effects of evidence type and message-issue distance on intentions. We propose: H2a. The perceived usefulness of statistical evidence (relative to anecdotal evidence) increases with increasing message-issue distance and the perceived usefulness of anecdotal evidence (relative to statistical evidence) increases with decreasing message-issue distance. H2b. The effect of evidence type on behavioral intentions is mediated by the perceived usefulness of the message, and the effect of evidence type on the perceived usefulness of the message is moderated by message-issue distance (moderated mediation). 4|EMPIRICAL STUDIES This paper presents three experiments. Study 1a tests H1a, Study 1b tests H1b and Study 2 tests H1a,H2a, and H2b. Figure 2illustrates the effects tested in Experiment 1a. 4.1 |Experiment 1a 4.1.1 | Method Stimuli and design The study used a 2 (distance: same country vs. foreign country) 2 (evidence type: statistical vs. anecdotal) between-subjects design. The messages asked for donations and described a project to support vacation homes for children with cancer—either in Germany (same versus FIGURE 2 Effects studied in Study 1a. WIELUCH and PRAXMARER-CARUS 79 country) or in Japan (foreign country). The study was conducted in Germany. We chose Japan because Germany and Japan are comparable in terms of their health care systems and the number of children with cancer (Globocan, 2012). The statistical evidence message reported how many new cases of childhood cancer are diagnosed each year. The anecdotal message told the story of a little girl with cancer. Appendix Ashows the messages. Procedure and sample The study was programmed in the online survey software Unipark (www.unipark.com). The link to the online questionnaire was distributed by an online panel provider in Germany. The study randomly assigned participants to one of the messages. Because the panel provider administered the sample, we do not have information on response rates within the panel. Participants ranged in age from 17 to 88 (M age =28). The study included a control question asking in which country the project was located. Because the country (Germany or Japan) acted as a distance manipulation, we excluded participants who were unable to answer this question correctly (43), resulting in a final sample of 394 individuals. Overall, 44% of the sample was female, and the average age was 28. Three hundred sixty-eight individuals (93%) were German (16 individuals did not report their nationality, six were from Austria, one from Switzerland, and one was French). Overall 46% were employed or self-employed, 2% were homemakers, 39% were students, 5% indicated “other,”and 7% refused to report their occupation. Overall 47% of respondents had donated within the past 12 months. Measures We asked participants if they had donated in the past 12 months (yes =47%). Furthermore, we asked participants: “Have you ever donated for a cancer-related project or organization?”(yes =19%) because involvement with cancer-related projects is likely to determine individuals' intentions in the context of this study. After message exposure, we measured receivers' intention to donate with the item “How likely are you to donate to the project described?”(1 =very unlikely, 7 =very likely). The manipulation check of psychological distance read “I feel closely related to the children in …”[Japan or Germany] (1 =not at all, 7 =strongly). In addition, the study measured descriptive statistics and included a few questions not related to this article. 4.1.2 | Results Manipulation check Participants in the close condition expressed that they felt more closely related to the children than participants in the distant condition (M close =3.47; M distant =1.96; t=9.91, p=.000). Thus, the distance manipulation was successful. Covariate Whether participants had donated to a cancer-related project in the past indicated their past behavior and involvement with cancerrelated projects. Overall, 19.3% of participants reported having donated to a cancer-related project prior to the study. Because prior behavior was likely to determine donation intentions in this study, and the percentages differed across the four experimental groups (e. g., anecdotal message, close condition: 16%, statistical message, close condition: 25%), we controlled for the variable “previous donations to cancer-related projects”in our analysis and included it as a covariate. Below, we first report the relevant model with the covariate and then, for transparency, the model without the covariate. Test of H1a We performed analysis of covariance (ANCOVA) with donation intention as dependent variable, issue distance and evidence type as independent variables and prior cancer-related donations as covariate. Evidence type had no significant main effect (F(1, 389) =.36, p=.551, partial η 2 =.001). The main effect of issue distance was significant (F(1, 389) =27.93, p=.000, partial η 2 =.067). Individuals were more likely to donate when donation recipients were from the same country. This negative main effect of distance is not surprising because donors tend to prefer charities working in their own country (Chapman et al., 2022; Herzenstein & Posavac, 2019). As expected, the issue distance by evidence type interaction was significant (F (1, 389) =4.51, p=.034, partial η 2 =.011). Furthermore, the effect of the covariate (previous behavior) was significant; respondents who had previously donated to a cancer project were more likely to donate (F(1, 389) =29.10, p=.000, partial η 2 =.070). Table 1shows the mean values and estimated marginal means (EMM), Figure 3illustrates the negative main effect of distance and the interaction effect between evidence type and distance. Consistent with H1a, the interaction effect shows that an increase in message-issue distance increased the relative persuasiveness of statistical evidence, while a TABLE 1 Compliance intention across the conditions of Experiment 1a, means and estimated marginal means (EMM). Issue distance Evidence type Means (SD) EMM (SE) N Close condition Anecdotal evidence 2.97 (1.57) a ** 3.00 (.13) c,d ** 102 Statistical evidence 2.68 (1.50) b *2.63 (.14) c,e *94 Distant condition Anecdotal evidence 2.00 (1.20) a ** 1.99 (.13) d ** 105 Statistical evidence 2.18 (1.29) b *2.20 (.14) e *93 Note: Means with equal superscripts are significantly different at the level of .10 (at least), if they are additionally marked with *, they are significant at p< .05, if they are additionally marked with **, they are significant at p< .01. *At p< .05.**At p< .01. 80 WIELUCH and PRAXMARER-CARUS decrease in message-issue distance increased the relative persuasiveness of anecdotal evidence. The findings provide support for H1a. For transparency, we report the findings of an ANOVA without the covariate. Again, evidence type had no significant main effect (F (1, 390) =.14, p=.705, partial η 2 =.000). The main effect of issue distance was significant (F(1, 390) =27.07, p< .001, partial η 2 =.065). The issue distance by evidence type interaction was marginally significant at p<.10(F(1, 390) =2.80, p=.095, partial η 2 =.007). Note that the model controlling for past behavior (ANCOVA) should provide more reliable estimates of the effects than the model without the covariate because past behavior (the covariate) affected donation intentions and was not evenly distributed across the experimental groups. Pairwise comparisons We did not propose an absolute advantage of statistical evidence over anecdotal evidence in the distant condition or an absolute advantage of anecdotal evidence over statistical evidence in the close condition because, for example, “less distant”is not necessarily close (see Figure 1). Nevertheless, we report pairwise comparisons. In the close condition, anecdotal evidence tended to be more persuasive than statistical evidence at p=.056 (EMM anecdotal evidence =3.00; EMM statistical evidence =2.63, F(1, 389) =3.68). In the distant condition, statistical evidence was slightly more persuasive than anecdotal evidence but the difference was not significant (EMM anecdotal evidence =1.99; EMM statistical evidence =2.20, F(1, 389) =1.18, p=.278). An absolute advantage of statistical evidence over anecdotal evidence in the distant condition would require high-level or analytical thinking. However, it is possible that the children suffering from cancer evoked compassion in the recipients, which may have resulted in low-level construal, even in the relatively distant condition. In addition, a threatening subject such as cancer may evokenegativeemotions,whichmaylead to relatively concrete, lowlevel construal even in the distant condition (Labroo & Patrick, 2009). 4.1.3 | Discussion of Experiment 1a The interaction effect suggests that the effectiveness of anecdotal and statistical evidence depends on message-issue distance. Furthermore, anecdotal evidence was more effective than statistical evidence only in the close situation (p=.056), but not in the distant situation. In the distant situation, statistical evidence was as effective as anecdotal evidence (and tended to be more effective). We cannot exclude the possibility that the interaction effect between evidence type and issue distance in Experiment 1a was also driven by similarity and identification, since the effectiveness of the anecdotal message was much higher in the close than in the distant situation. The literature has demonstrated that anecdotal messages are more persuasive when the protagonists are similar to the recipients (De Graaf, 2014; Praxmarer-Carus & Wolkenstoerfer, 2018), and message recipients are more likely to identify with a person from their own country than with a person from a foreign country. However, the theoretical reasons why we expect the effectiveness of anecdotal and statistical evidence to depend on message-issue distance are more general (see Section 3) and are neither dependent on nor limited to protagonist similarity effects. To avoid such potential similarity and identification effects, Experiment 2, which also tests H1a, manipulates the recipients' message-issue distance as a temporal distance. Thus, similarity and identification will not be able to explain a potential interaction effect between evidence type and message-issue distance in Experiment 2. 4.2 |Experiment 1b This experiment tests H1b and examines whether recipients' concrete/abstract thinking about the message issue moderates the persuasiveness of statistical and anecdotal evidence (see Figure 4). This study does not manipulate message-issue distance but measures how abstractly/concretely individuals think about the message issue prior to message exposure. The experiment focuses on individuals' issuespecific thinking. While individuals may have a chronic tendency to think more abstractly or more concretely (Hong & Lee, 2010; Vallacher & Wegner, 1989), mental representations should also vary within individuals and across events, for example, because the event may be psychologically closer or further away from the individual (Liviatan et al., 2008; Soderberg et al., 2015; Wakslak et al., 2006). While chronic tendencies and issue-specific construal levels should be related, individuals' concrete/abstract thinking about the message issue should be more influential in explaining the persuasiveness of anecdotal and statistical evidence and is the focus of this study. 4.2.1 | Method Stimuli and design The study manipulated the type of evidence (statistical vs. anecdotal) and measured participants' concrete/abstract thinking about the FIGURE 3 The effects of anecdotal and statistical evidence on intentions dependent on message-issue distance (Experiment 1a). versus FIGURE 4 Effects studied in Study 1b. WIELUCH and PRAXMARER-CARUS 81 message-issue prior to exposure to the message. The context of the study was tick-borne encephalitis and the messages recommended vaccination for individuals living in a risk area (several federal states and districts in Germany are risk areas). The statistical message reported cases of tick-borne encephalitis and the like, and the anecdotal message told the story of a student who was infected by a tick bite. Appendix Bshows the messages. Procedure and sample The link to the online questionnaire was posted on the platform of a large student community in Germany. Participants received no compensation. The study randomly assigned participants to one of the messages, and the questionnaire was completed by 361 individuals. Because the German government's vaccine agency (RKI), and therefore German health professionals, recommend vaccination only for people living in a risk area (Robert Koch Institute, 2024), the messages did not apply to 94 participants who knew they did not live in a risk area. Since the vaccine is not recommended for them and the message recommendation did not apply to them, they had to be excluded from the study. The final sample consisted of 267 individuals (38.2% male, the average age was 24). As one vaccination is not sufficient and the vaccine needs to be boostered every 5 years, participants who had already been vaccinated against tick-borne encephalitis were not excluded (40% reported having been vaccinated). Measures Before we could ask participants how concretely/abstractly they thought about the message issue, we had to state the message issue. Participants read the following brief information: “Tick bites can infect humans with diseases such as tick-borne encephalitis, which can cause inflammation of the meninges and brain.”To measure how concretely/abstractly participants thought about the message issue, we adapted a self-report item from Venus et al. (2019). The item read “When I think about tick bites and the potential dangers they pose, I tend to have …concrete details in mind (=4); …in mind what it could mean for me in general (=+4). Participants responded on a 9point rating scale. We chose this item, suggested by Venus et al. (2019), for two reasons. First, it focuses on the core of the construct of interest (Bergkvist & Rossiter, 2007), which can be described as follows: Whereas low-level representations of events contain contextualized details, high-level representations of events achieve abstraction by omitting secondary details and focusing on their generalized meaning (Trope et al., 2007). Second, the short self-report measure may be useful in practice. It does not require coding of participants' openended responses, such as the linguistic category model (Semin & Fiedler, 1991), which involves controversy over how to interpret the terms expressed by individuals (Burgoon et al., 2013). And (if it explains the persuasiveness of anecdotal and statistical evidence) it could easily be used to pretest campaigns and decide when to use statistical or anecdotal evidence (see Section 5). After exposure to the message, we measured participants' intentions to get vaccinated with the item “How likely are you to get vaccinated?”(1 =very unlikely, 9 =very likely). The study also measured descriptive statistics, asked whether participants lived in a risk area, whether they had been vaccinated, and a few other questions not related to this article. 4.2.2 | Results Covariate Whether participants had been vaccinated against tick-borne encephalitis before taking part in our study indicates their general acceptance of vaccines. In the group that saw the statistical evidence, 44% had been vaccinated; in the anecdotal group, 36% had been vaccinated. We report the results with and without controlling for this variable (covariate). Test of H1b As the moderator was measured on a rating scale, we ran model 1 of Hayes' (2018) PROCESS macro to test H1b. Evidence type was the independent variable (statistical evidence =0, anecdotal evidence =1), concrete/abstract thinking about the message issue was the moderator (4=concrete; +4=abstract), and vaccination intention was the dependent variable. Table 2shows the findings. As TABLE 2 Model 1 of Hayes' (2018) PROCESS macro (Experiment 1b). Model with covariate Model without covariate bSE t(262) pb SE t(263) p Constant 4.148 .256 16.18 .000 5.364 .243 22.07 .000 Evidence type (EvT) .350 .310 1.130 .260 .118 .350 .336 .737 Concrete/abstract thinking (CL) .145 .091 1.588 .114 .107 .103 1.037 .310 EvT CL .391 .130 3.001 .003 .394 .147 2.672 .008 Covariate 2.724 .314 8.675 .000 - - - - Conditional effects of evidence type at values of CL Value of CL Effect SE pValue of CL Effect SE p 16th percentile 2.12 1.178 .441 .008 2.12 .952 .498 .057 84th percentile 3.00 .822 .460 .079 3.00 1.064 .527 .044 82 WIELUCH and PRAXMARER-CARUS intentions to obtain the HPV vaccine. Health Communication,30, 301–308. Park, S. Y., Kim, E., & Moon, M. J. (2020). Designing public information campaigns as an effective policy tool: Construal-level fit effects and evidence from an experimental study. Journal of Comparative Policy Analysis: Research and Practice,22(6), 579–592. Praxmarer-Carus, S., & Wolkenstoerfer, S. (2018). Enhancing the effectiveness of narratives among vaccine-skeptical parents. ACR North American Advances. Praxmarer-Carus, S., Wolkenstoerfer, S., & Dijkstra, A. (2022). Outcome presence and regulatory fit: Competing explanations for the advantage of gains and losses over non-gains and non-losses. Journal of Consumer Behaviour,21(2), 310–327. Reczek, R. W., Trudel, R., & White, K. (2018). Focusing on the forest or the trees: How abstract versus concrete construal level predicts responses to eco-friendly products. Journal of Environmental Psychology,57,87–98. Robert Koch Institut. (2024). Infektionsschutz. https://www.rki.de/DE/ Content/Infekt/Impfen/ImpfungenAZ/FSME/FSME.html. Schwarz, N. (2004). Metacognitive experiences in consumer judgment and decision making. Journal of Consumer Psychology,14(4), 332–348. Schwarz, N. (2010). Meaning in context: Metacognitive experiences. In B. Mesquita, L. F. Barrett, & E. R. Smith (Eds.), The mind in context (pp. 105–125). Guilford. Semin, G. R., & Fiedler, K. (1991). The linguistic category model, its bases, applications and range. European Review of Social Psychology,2(1), 1–30. Slater, M. D., & Rouner, D. (1996). Value-affirmative and value-protective processing of alcohol education messages that include statistical evidence or anecdotes. Communication Research,23(2), 210–235. Slovic, P. (2007). If I look at the mass I will never act: Psychic numbing and genocide. Judgment and Decision Making,2(2), 79–95. Small, D. A., Loewenstein, G., & Slovic, P. (2007). Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims. Organizational Behavior and Human Decision Processes,102(2), 143–153. Soderberg, C. K., Callahan, S. P., Kochersberger, A. O., Amit, E., & Ledgerwood, A. (2015). The effects of psychological distance on abstraction: Two meta-analyses. Psychological Bulletin,141(3), 525–548. Statista. (2011). Welche Gründe sprechen gegen eine Berufsunfähigkeitsversicherung? https://de.statista.com/statistik/daten/studie/199188/ umfrage/gruende-gegen-eine-berufsunfaehigkeitsversicherung/. Statista. (2022). Schwellenwerte für die Armutsgefährdung in Deutschland. https://de.statista.com/statistik/daten/studie/816302/umfrage/ monatlicher-schwellenwert-fuer-armutsgefaehrdung-in-deutschland2016/. Trope, Y., & Liberman, N. (2010). Construal-level theory of psychological distance. Psychological Review,117(2), 440–463. Trope, Y., Liberman, N., & Wakslak, C. (2007). Construal levels and psychological distance: Effects on representation, prediction, evaluation, and behavior. Journal of Consumer Psychology,17(2), 83–95. Truth Initiative. (2022). Tobacco and cancer: What are the survival rates? https://truthinitiative.org/research-resources/harmful-effectstobacco/tobacco-and-cancer-what-are-survival-rates. Vallacher, R. R., & Wegner, D. M. (1989). Levels of personal agency: Individual variation in action identification. Journal of Personality and Social Psychology,57(4), 660–671. Venus, M., Johnson, R. E., Zhang, S., Wang, X. H., & Lanaj, K. (2019). Seeing the big picture: A within-person examination of leader construal level and vision communication. Journal of Management,45(7), 2666–2684. Wakslak, C. J., Trope, Y., Liberman, N., & Alony, R. (2006). Seeing the forest when entry is unlikely: Probability and the mental representation of events. Journal of Experimental Psychology: General,135(4), 641–653. White, K., MacDonnell, R., & Dahl, D. W. (2011). It's the mind-set that matters: The role of construal level and message framing in influencing consumer efficacy and conservation behaviors. Journal of Marketing Research,48(3), 472–485. Williams, L. E., Stein, R., & Galguera, L. (2014). The distinct affective consequences of psychological distance and construal level. Journal of Consumer Research,40(6), 1123–1138. Winterbottom, A., Bekker, H. L., Conner, M., & Mooney, A. (2008). Does narrative information bias individual's decision making? A systematic review. Social Science & Medicine,67(12), 2079–2088. Yan, D., & Sengupta, J. (2013). The influence of base rate and case information on health-risk perceptions: A unified model of self-positivity and self-negativity. Journal of Consumer Research,39(5), 931–946. Yan, D., Sengupta, J., & Hong, J. (2016). Why does psychological distance influence construal level? The role of processing mode. Journal of Consumer Research,43(4), 598–613. Zillmann, D., & Brosius, H. B. (2012). Exemplification in communication: The influence of case reports on the perception of issues. Routledge. Zebregs, S., Van den Putte, B., Neijens, P., & De Graaf, A. (2015). The differential impact of statistical and narrative evidence on beliefs, attitude, and intention: A meta-analysis. Health Communication,30(3), 282–289. AUTHOR BIOGRAPHIES Marina I Wieluch served as a research assistant at the Bundeswehr University Munich, Department of Economics and Management and received her doctorate degree from the Bundeswehr University Munich, Germany. Her research interests include persuasion and advertising effectiveness. Her work has appeared in “Advances in Consumer Research”(proceedings of the ACR conferences). Present position: Head of Department Public Relations and Communication, Landesfeuerwehrverband (State Firefighters Association) Bayern e.V. Sandra Praxmarer-Carus is a Professor of Marketing at the Bundeswehr University Munich, Department of Economics and Management. Her research interests include persuasion and advertising effectiveness. She received her doctorate degree from the University of Augsburg, Germany. Before her appointment in Munich, she served as a Lecturer in Marketing at the University of Wollongong, Australia, and as an Assistant Professor at the University of Bamberg, Germany. She has published in, for instance, the Journal of Business Research, the Journal of Advertising Research, the International Journal of Advertising, and the Journal of Consumer Behaviour. How to cite this article: Wieluch, M. I., & Praxmarer-Carus, S. (2025). How the persuasiveness of statistical evidence compared to personal testimonials depends on the recipient's distance from the message issue. Journal of Consumer Behaviour,24(1), 75–91. https://doi.org/10.1002/cb.2404 WIELUCH and PRAXMARER-CARUS 89 APPENDIX A The messages that Experiment 1a used. Close Anecdotal Statistical Anna, 4, from Munich, is suffering from cancer. Our vacation homes in Germany offer the required care and resources, where she and her family can recover their strengths after their stay in hospital. Your donation helps heal Every year 1.800 kids under the age of 15 are diagnosed with cancer in Germany. Our vacation homes in Germany offer the required care and resources, where these kids and their families can recover their strengths after their stay in hospital. Your donation helps heal Distant Anecdotal Statistical Ayumi, 4, from Kyoto, is suffering from cancer. Our vacation homes in Japan offer the required care and resources, where she and her family can recover their strengths after their stay in hospital. Your donation helps heal Every year 1.800 kids under the age of 15 are diagnosed with cancer in Japan. Our vacation homes in Japan offer the required care and resources, where these kids and their families can recover their strengths after their stay in hospital. Your donation helps heal 90 WIELUCH and PRAXMARER-CARUS APPENDIX B The messages that Experiment 1b used. APPENDIX C The messages that Experiment 2 used. Anecdotal Statistical Lukas, 24, student, had a relaxing afternoon with friends in the park. Sunbathing, relaxing, and playing a bit of football… In the morning, he noticed a tick on his groin. He had it removed and that seemed to be the end of it. Nine days later, Lukas developed a fever and a headache. The tick had infected him with the tick-borne encephalitis virus and Lukas had meningitis. He was admitted to hospital, but his condition worsened. When the infection spread to his brain, Lukas' condition became critical. Lukas survived, but he uses a wheelchair because the infection has paralyzed his legs. His ability to concentrate has also been affected. Lukas has dropped out of university and his career dreams have faded into the distance. Protect yourself from tick bites and the diseases they can transmit! The Robert Koch Institute recommends vaccination for people living in areas at risk of TBE. It provides reliable protection for about 5 years. The vaccination would have protected Lukas. Ask your doctor for more information! It is estimated that 10 million people in Germany are bitten by ticks every year. Ticks can be found almost anywhere there are plants, including gardens and parks. Ticks can transmit the tick-borne encephalitis (TBE) virus to humans. Depending on the region, up to 5% of ticks carry the TBE virus. TBE can cause severe inflammation of the meninges, brain and spinal cord. Meningitis (inflammation of the meninges) with encephalitis (inflammation of the brain) leaves one in five people (20%) with permanent damage. This often takes the form of paralysis or problems with memory and concentration. 3% of people use a wheelchair after developing the disease. 2% of cases are fatal. Adults and young adults are more likely to develop severe cases than children. Protect yourself from tick bites and the diseases they can transmit! The Robert Koch Institute recommends vaccination for people living in areas at risk of TBE. It provides reliable protection for about 5 years. Ask your doctor for more information! Note: Male participants saw the male version of the anecdote (“Lukas”) and female participants saw the female version of the anecdote (“Lea”). Anecdotal Statistical Not getting disability insurance early on puts your financial existence at risk. Stefan's story: Stefan was only 24 and a student when, unexpectedly, he became severely ill. Due to the disease, Stefan became permanently unfit for work. Presumably, he will never be able to work normally. Because he was a student, he is not eligible for minor governmental disability pension payments. Unfortunately, Stefan had not taken out disability income insurance for students that would have afforded him monthly benefits. Stefan's future is uncertain, and his financial existence is at risk. Not getting disability insurance early on puts your financial existence at risk. In Germany, 2.8 Mio. young people are studying. The probability of a 20-year-old of becoming permanently unfit for work before retirement age is 43%. In 90% of the cases, a severe disease is the cause. If students get permanently unfit for work, they are not eligible for the governmental pension payments of 38% of one's last income. If students have not taken out disability income insurance, they do not receive monthly benefits from an insurance. If they become unfit for work their future is uncertain, and their financial existence is at risk. Close condition Your to-do for this month: find out about disability insurance for students Close condition Your to-do for this month: find out about disability insurance for students Distant condition Your to-do for 2020: find out about disability insurance for students Distant condition Your to-do for 2020: find out about disability insurance for students Note: The study was conducted in November 2019. Male participants saw the male version of the anecdote (“Stefan”) and female participants saw the female version of the anecdote (“Stefanie”). WIELUCH and PRAXMARER-CARUS 91