This is an open access article distributed under the terms of the Creative Commons
Attribution License (Associating immigration with several economic or cultural threats is a well-established approach to explaining racism and xenophobia (Cea D’Ancona, 2016). This perception tends to become more acute in periods of economic uncertainty (Quillian, 1995), which is taken advantage of by far-right political parties with openly anti-immigration discourses that serve as a vehicle for channeling that fear (Della Posta, 2013; Rydgren, 2003). In recent years, these types of political groups have experienced unprecedented success in Spain and in other European countries, and their emergence on the political scene with a clearly xenophobic discourse makes anti-immigration messages naturalize, ultimately further aggravating intergroup tensions (Arango et al., 2019). In this context of social tension, it is a priority to design effective strategies to reduce intergroup conflict.
People experience daily intergroup contact situations with immigrants through the media. Given that there is a tendency to emphasize the anecdotal, the out of the ordinary, and the most negative events, on many occasions media contact reinforces negative attitudes towards immigration (Beyer & Matthes, 2015). However, it has also been observed that mediated intergroup contact (Harwood, 2010; Park, 2012) is an effective instrument to improve attitudes towards stigmatized groups (such as immigrants), especially when narrative messages are used (Moyer-Gusé et al., 2019; Murrar & Brauer, 2018).
This work focuses on the impact of personal or testimonial narrative messages, in which a migrant person relates their life experience. Testimonial messages or personal narratives are common on social media, blogs, and websites1. Although they are considered to represent a less sophisticated form of narrative (Walter & Cohen, 2019), testimonials are easier to understand than didactic messages or those based on statistical arguments, and are especially effective when the involvement with the message topic is low (Braverman, 2008; De Wit et al., 2008). These types of messages can thus be used in media campaigns against racism and xenophobia.
Taking as a reference the research on narrative persuasion (Bilandzic & Busselle, 2013; Fitzgerald & Green, 2017), we carry out an experiment using short written testimonial messages describing the experience of an immigrant woman from Romania, since this group is one of the most stigmatized in Spain. In this context, the present work provides two innovations. First, the factors that increase the persuasive efficacy of testimonial messages aimed at improving attitudes towards stigmatized immigrants are studied. In particular, the effect of the similarity to the protagonist and of the narrative voice are analyzed as “narrative devices” that can increase identification with the protagonist (or the capacity to feel and adopt the point of view of the protagonist of the narrative; Cohen et al., 2020). Second, the indirect effects on the attitude towards immigration and on the intention to collaborate with non-government organizations (NGOs) to support immigrants through identification (as a primary mediator) and cognitive elaboration (as a secondary mediator) are analyzed.
Personal Migrant Stories as Persuasive Devices
Narratives have been defined as messages including at least one character who experiences or faces an event (or broader set of situations) in a specific space–time frame. Furthermore, the narrative must communicate something relevant to the audience, who can take away a lesson from it all (Braddock & Dillard, 2016; Hoeken et al., 2016).
Testimonial messages, in particular, are persuasive devices that have been widely used in health campaigns (e.g., Chen et al., 2017; Banerjee & Greene, 2012; De Wit et al., 2008), and we thus consider their potential for the prevention of racism and xenophobia. In a testimonial message of this type, we propose the following properties should be present: two different states should be alluded to, namely life before emigrating and after becoming an immigrant, with both events being connected through a causal and temporal sequence. Furthermore, in the narrative the immigrant person mentions the reasons that led them to leave their country, tells their personal story in the host country, alludes to their feelings of belonging, and communicates their expectations for the future (the intentions and goals of the character).
Testimonials offer several advantages over other, more elaborate narrative formats. First, by focusing attention on a single protagonist, the underlying persuasive message can be communicated more effectively (Walter & Cohen, 2019). Second, their production cost is lower than that of entertainment–education audiovisual messages (Murrar & Brauer, 2018). And, third, as they are short messages, they can be conceived as narrative vaccines and be spread more easily through social media (Igartua et al., 2017).
Meta-analysis studies have shown that narrative interventions produce significant effects, although it is also observed that not all narratives are effective (Braddock & Dillard, 2016; Zebregs et al., 2015). An important question in this field is therefore to determine which are the ingredients of narratives that are most effective from a persuasive point of view (e.g., Cohen & Tal-Or, 2017; Tukachinsky, 2014). In this work we focus on two characteristics of narrative messages that can condition their persuasive effectiveness: audience–character similarity and narrative voice.
Audience–Character Similarity
Similarity occurs when the person who is exposed to a narrative message shares certain characteristics with its protagonist. There is an open debate on the role of similarity in increasing identification and, indirectly, fostering persuasive impact. According to the similarity–identification hypothesis (Cohen et al., 2018), people form more intense affective and cognitive bonds with protagonists of narrative messages who are similar in demographic, social, or psychological terms. However, the studies carried out to date have shown inconclusive results on this hypothesis (Tukachinsky, 2014). In this context, our proposal consists of making a theoretical adjustment to the concept of similarity, and secondly, analyzing the conditions under which similarity increases identification with the protagonist.
Most of the works in which character similarity has been manipulated have concentrated on objective demographic attributes such as sex or age (e.g., Cohen et al., 2018; de Graaf, 2014; Hoeken et al., 2016), whereas there are few studies on the effect of similarity based on psychological or subjective characteristics, such as personality, attitudes, values, or biographical experiences (Cohen & Hershman-Shitrit, 2017). We consider that, when messages are focused on stigmatized protagonists, more powerful narrative resources are necessary to create affective and cognitive ties than mere demographic matching, since these types of characters elicit, by default, less identification (Chung & Slater, 2013; Igartua & Frutos, 2017). We therefore propose that other dimensions of similarity linked more directly with social identity should be activated.
The theory of social identity and the theory of self-categorization (Tajfel, 1982; Turner, 1985) posits that people tend to categorize those perceived to be similar as members of the ingroup, while those perceived to be dissimilar are considered to be members of the outgroup. This leads to people reacting differently towards members of the outgroup (Gaertner et al., 1993). However, since the ingroup–outgroup categorization is context-dependent, it is possible that, if certain cues are activated in the message, a case belonging to an outgroup can be seen as a member of the ingroup (Turner et al., 1987). Moreover, according to the Common Ingroup Identity Model (Gaertner et al., 1993), a person belonging to an outgroup is classified as such because they differ in certain features that define social identity, such as ethnicity, national origin, or any other group categorization. However, if a shared sense of identity is stimulated, creating a supraordinate category (“we”), the ingroup–outgroup separation is more likely to disappear (Dale & Moyer-Gusé, 2020).
In this context, our research focuses on the effect of similarity in terms of social identity, which implies highlighting in the message what people share regarding subjective aspects such as national feelings and attachment to cultural elements (Igartua et al., 2019). For example, similarity to the host-country audience can be activated if an immigrant makes a positive comment in the message about the host-country’s food. This does not prevent the protagonist from safely retaining their identity of origin, but it puts them in a different perspective before the host-country audience, who will perceive a shared common social identity. We consider that this subtle activation of subjective similarity could deactivate the ingroup–outgroup categorization process and, consequently, facilitate identification with the protagonist of the narrative. However, our work does not propose a main effect of similarity, but rather explores the role of the narrative voice as a second (formal) narrative attribute that could facilitate identification with the protagonist.
Narrative Voice
The narrative voice is a formal feature related to the perspective adopted when depicting a story, differentiating messages written in the first, second, or third person (Banerjee & Greene, 2012; Chen et al., 2017; Nan et al., 2015). In first-person narratives, the narrator is inside the story and communicates their vision on a specific issue or their experience. In this way, the audience has direct access to the internal life of the protagonist, which facilitates the involvement process. On the other hand, in the third-person messages, the narrator recounts the experience of the protagonist from an external or spectator’s perspective. For this reason, third-person messages establish a kind of psychological firewall that separates the individual from the protagonist of the story. Finally, second-person narratives identify the reader of the narrative as the protagonist, but they are rarely used in health campaigns or to solve social problems (de Graaf et al., 2016).
In research on narrative persuasion, insufficient time has been invested in analyzing the effect of the narrative voice, despite its being a very relevant formal resource for the design of narrative messages (Chen et al., 2016; de Graaf et al., 2016). What has been observed in previous studies is that first-person messages (as opposed to third-person narratives) are perceived as more personal, are easier to understand, and increase perspective-taking (Chen et al., 2017; Nan et al., 2015).
However, studies that combine the narrative voice and a second independent variable in the same experiment have not yielded consistent results (Kaufman & Libby, 2012; Nan et al., 2017). For example, in the context of public service advertisements promoting human papillomavirus (HPV) vaccination, Nan et al. (2017) carried out an online experiment employing a 3 (message type: first-person narrative, third-person narrative, or non-narrative) x 2 (modality: text-based vs. audio-based) between-subjects design. The results revealed that the first-person narrative did not result in greater identification compared to the third-person narrative. However, Kaufman and Libby (2012) observed an interaction effect between the narrative voice (first or third person) and the similarity to the protagonist (belonging or not to the same university as the participants) on identification, whereby reading a story written in the first person and narrated by a character from the ingroup significantly increased identification. That study serves as a reference to propose our first hypothesis. We argue that the similarity to the protagonist (a Romanian immigrant woman) may constitute a necessary, but not sufficient, condition to increase identification. In this way, an additional attribute is needed in the narrative to enhance identification through the narrative voice.
H1. Similarity to the protagonist will increase identification only when the message is written in the first person.
Inside the Black Box: Identification With the Protagonist and Cognitive Elaboration
The main theoretical models on narrative persuasion consider that identification fosters the persuasive impact, since this process inhibits resistance to the persuasive proposal of the message, facilitating its attitudinal impact (Moyer-Gusé, 2008; Slater & Rouner, 2002). Nevertheless, the empirical evidence on the role of counterarguing (the evocation of critical cognitive responses that refute the content of the message; Niederdeppe et al., 2012) and reactance (reacting negatively to the message when perceiving that freedom of choice or opinion is being threatened; Rains, 2013) is not conclusive, so alternative models that give greater weight to cognitive elaboration have been proposed (de Graaf & Van Leeuwen, 2017; Igartua & Vega, 2016; Walter & Cohen, 2019). In any case, the role of cognitive elaboration in narrative persuasion processes has been less well investigated to date than the role of message resistance processes (de Graaf & Van Leeuwen, 2017). For this reason, our work makes a significant contribution to understanding the role of identification and cognitive elaboration in narrative persuasion processes.
Identification constitutes an imaginative process that has been defined as a multidimensional construct linked to emotional empathy, cognitive empathy, and the feeling of merging with the character and adopting their goals (Cohen, 2001; Igartua & Barrios, 2012). It constitutes a psychological phenomenon by which the members of the audience mentally adopt the position of the protagonist of the narrative (Cohen & Tal-Or, 2017). Meanwhile, cognitive elaboration is defined as a process of reflection around the content of the message and constitutes a measure of the intensity of such reflection during the reception process (Petty & Cacioppo, 1986; Igartua & Vega, 2016). Cohen (2001) points out that identification constitutes a process of temporary involvement with the message, enabling an increase in cognitive elaboration during its reception. In this context, we consider that the experience narrated in the message by a positive role model can serve as inspiration and stimulate deep cognitive processing in people, such that they question their previous opinions and adjust their attitudes about immigration.
Previous evidence on this topic is scarce, but corroborates this approach, revealing that identification increases cognitive elaboration and thereby the attitudinal impact (Igartua & Vega, 2016). However, those results were obtained in the context of research on entertainment–education, using complex messages of an audiovisual nature that incorporated attitudinal proposals with health benefits for individuals. In contrast, this research uses testimonial-type narrative messages that incorporate an attitudinal proposal without apparent or direct benefits for individuals (for example, improving intergroup attitudes). In this context, our work aims to provide evidence on the role of cognitive elaboration as a secondary mediator mechanism dependent in turn on the identification process. We consider that, by narrating in the first person the testimony of a migrant who presents themselves as similar to the audience, the host-country population can more easily identify with the protagonist and find these messages inspiring, in turn leading to an attitudinal impact. In this context, it is assumed that identification can facilitate persuasive impact through cognitive elaboration, establishing the following mediational hypothesis (see Figure 1):
H2. The indirect effect of similarity to the protagonist on the attitude towards immigration (H2a) and the intention to collaborate with NGOs that support immigrants (H2b) will be serially mediated by identification and cognitive elaboration, but only when the message is written in the first person.
Figure 1
Hypothesized Moderated Serial Multiple Mediation Model (H2)
Note. Dashed lines represent non-hypothesized relationships between the variables included in the model.
Method
Design and Participants
An online experiment was carried out using Qualtrics to access a panel of 464 people of Spanish origin and whose parents were also Spanish. Of these, 21 were removed for failing the manipulation check of narrative voice manipulation (correctly remembering which type of message had been read, i.e., first or third person). The remaining sample consisted of 443 participants (50.3% women). Participant age ranged from 18 to 65 years old (M = 41.06, SD = 12.51, Mdn = 41.00). It is important to note that the inclusion of the participants who failed the manipulation check did not change the results of the present study. Both sets of results are available in the Open Science Framework (OSF) together with the rest of the materials (https://osf.io/79kcx/).
The questionnaire used was divided into three blocks: pre-test measures, experimental manipulation, and post-test measures (see Supplementary Materials for a detailed description of the measures). In the pre-test measures, sociodemographic variables, political self-positioning (from 0 = left to 10 = right, M = 4.42, SD = 2.39), and national identification with Spain (see below) were measured. After completing this block of questions, the participants were randomized to four experimental treatments according to a 2 (similarity to the protagonist: low versus high) × 2 (narrative voice: first versus third person) between-subjects factorial design. After reading the narrative, the post-test measures were presented with questions to test the effectiveness of the two experimental manipulations, the mediating variables (identification with the protagonist and cognitive elaboration), and the dependent variables (attitude towards immigration and intention to collaborate with NGOs to support immigrants).
Since Qualtrics allows a series of quality controls to be implemented, the questionnaire was designed in such a way that it could only be completed in a single session. In addition, only the results from participants who took between 6 and 45 minutes to complete the questionnaire (M = 14.69 minutes, SD = 5.92), took between 60 and 420 seconds to read the narrative (M = 127.52, SD = 53.66), and correctly answered a control question included in the questionnaire were counted as valid cases (additional information on Qualtrics’ quality control process is available in the OSF repository).
Stimulus Materials and Independent Variables
Examples of testimonies from the web pages of immigrant support organizations were taken as a reference to construct a written narrative by a Romanian immigrant woman. In the narrative, 30-year-old Alina (a popular female name in Romania) recounted her experiences since arriving in Spain 4 years previously. The underlying persuasive goal of the message was to improve attitudes towards immigration in Spain; furthermore, the protagonist indicated that information campaigns should be developed to reduce stereotypes (narrative messages are available in the OSF repository).
Four versions of the testimonial message were created by manipulating the similarity to the protagonist and the narrative voice. To manipulate the similarity to the protagonist, various aspects were considered (to reinforce this effect; Kim et al., 2016): the feeling of belonging (“I’ve been living in Spain for four years, and I feel almost Spanish” versus “I’ve been living in Spain for four years, but I don’t feel Spanish”), the similarity in values (“my values are very similar to those of most Spaniards” versus “my values are very different from those of most Spaniards”), having many (or some) Spanish friends, attachment to cultural elements such as food (“I feel very close to this country; its landscape and food attract me a lot” versus “I feel distant from this country; I am not attracted much to its landscape or food”), the use of the Spanish language (“I think I speak Spanish very well” versus “I think I speak Spanish very poorly”), and the perception of similarity to the majority of Spaniards (“I feel that I have many things in common with the people of Spain” versus “I feel that I have few things in common with the people of Spain”).
To manipulate the narrative voice, the elements marking the grammatical person in the written narratives, such as the choice of personal pronouns, were modified (Banerjee & Greene, 2012; Chen et al., 2016). In addition, in the first-person narrative, the protagonist presented herself with her own voice and with her name at the beginning (“My name is Alina Dimitru, I’m 30 years old, and I’m from Brasov in Romania”). On the other hand, in the third-person narrative, an external observer introduced the protagonist and narrated the story (“Alina Dimitru is 30 years old and is from Brasov in Romania”).
All four versions of the narrative message were identical in all other aspects of their content. In addition, in all the versions, the protagonist made positive comments about her life in Spain and ended her story by stating “I want Spain to be my home and my daughter’s” (“Alina wants Spain to be her home and her daughter’s”). The four narratives used had a similar length (between 550 and 560 words), which is the most common length used in this type of work (Dahlstrom et al., 2017). A pilot study was carried out in which 120 people participated (57.5% women, M = 19.94 years, SD = 2.03) to confirm the validity of the experimental manipulation of the similarity to the protagonist. As expected, the participants who read the high-similarity narrative scored significantly higher on the perception of similarity than those who had read the narrative designed to stimulate low similarity (results are available in the OSF repository).
Measures
Mediating Variables
Identification With the Protagonist
Identification was assessed using an 11-item scale (Igartua & Barrios, 2012) that measures the degree of identification with a specific character (e.g., “I felt as if I were Alina”; 1 = not at all, 5 = very much; α = .93, M = 3.33, SD = 0.86).
Cognitive Elaboration
An adapted version of the Igartua and Vega (2016) cognitive elaboration scale was used, made up of three items (e.g., “As I read, I tried to draw conclusions to adjust my opinions about immigration”; from 1 = strongly disagree to 7 = strongly agree; α = .81, M = 5.03, SD = 1.17).
Dependent Variables
Attitudes Towards Immigration
These were evaluated with a scale used in previous studies (e.g., Igartua et al., 2019) and composed of four items (e.g., “Do you think that, in general terms, immigration is rather positive or rather negative for Spain?”; 1 = very negative, 7 = very positive; α = .78, M = 4.81, SD = 1.20).
Intention to Collaborate With NGOs Supporting Immigrants
A scale developed from Murrar and Brauer (2018), composed of four items (e.g., “I would be willing to donate money to an NGO to finance campaigns against racism and xenophobia”, 1 = strongly disagree, 7 = strongly agree; α = .93, M = 3.55, SD = 1.61), was used.
Experimental Manipulation Check
Perceived Similarity to the Protagonist
This was evaluated using a scale made up of six items (e.g., “Alina has many things in common with the people of Spain”; from 1 = strongly disagree to 7 = strongly agree; α = .90, M = 4.83, SD = 1.48).
Remembering the Narrative Voice
Participants were asked: Do you remember whether the story you just read was written in the first person or in the third person? (1 = it was written in the first person, “My name is Alina, I’m 30 years old ...”; 2 = it was written in the third person, “Alina is 30 years old ...”).
Covariates
National identification With Spain
This was evaluated using a scale elaborated from the works of Finell et al. (2013) and Pehrson et al. (2009) and composed of four items (e.g., “I feel proud to be Spanish”; from 1 = strongly disagree to 7 = strongly agree; α = .91, M = 4.92, SD = 1.68).
Results
Preliminary Analysis
Randomization was successful, the conditions did not differ significantly on gender: χ2(3, N = 443) = 2.99, p = .393, age: F(3, 439) = 1.29, p = .276, political ideology: F(3, 439) = 0.25, p = .857, or national identification: F(3, 439) = 0.29, p = .833. Perceived similarity manipulation was successful. An independent-sample t-test found that participants who read a high-similarity personal narrative considered that Alina was similar to most Spaniards (M = 5.75, SD = 0.96) to a significantly greater extent than participants who read the low-similarity testimonial, M = 3.89, SD = 1.31; t(441) = -17.04, p < .001.
Effect of Similarity to the Protagonist and Narrative Voice on Identification (H1)
The first hypothesis suggests that similarity to the protagonist would increase identification only when the message was written in the first person. To test this hypothesis, the PROCESS macro by Hayes (2018) was used, using model 1 (simple moderation)2. A statistically significant interaction effect between the similarity and the narrative voice on the identification with the protagonist was found (B = 0.31, SE = 0.15, p = .041). The analysis of the conditional effects showed that the similarity significantly increased the identification with the protagonist when participants read the version of the message written in the first person (θSimilarity → Identification|(First-person narrative) = .27, SE = .11, p = .011), but not the third person (θSimilarity → Identification|(Third-person narrative) = -.03, SE = .10, p = .732). These results supported H1.
Testing a Moderated Serial Multiple Mediation Model (H2)
H2 predicts that the effect of the similarity to the protagonist on attitudes towards immigration and on the intention to collaborate with NGOs supporting immigrants would be serially mediated by the identification and cognitive elaboration, but that this effect would only occur when the message was written in the first person. To test this hypothesis, the PROCESS macro for SPSS was used (model 83; 10,000 boostrapping samples to generate 95% confidence intervals by the percentile method; Hayes, 2018). This procedure makes it possible to estimate the indirect (conditional) effects of similarity on the dependent variables through identification (primary mediator) and cognitive elaboration (secondary mediators), at the different levels of the moderating variable (narrative voice).
Identification was found to be associated with greater cognitive elaboration (B = .85, SE = .05, p < .001). Furthermore, cognitive elaboration was significantly associated with a more favorable attitude towards immigration (B = .18, SE = .05, p < .001; see Figure 2) and with a greater intention to collaborate with an NGO to support immigrants (B = .44, SE = .06, p < .001; see Figure 3).
Figure 2
Results of the Moderated Serial Multiple Mediation Analysis (H2) – Dependent Variable: Attitudes Towards Immigration
Note. The figure shows the non-standardized regression coefficients, B. The coefficient of the direct effect appears in parentheses. Dashed lines represent nonsignificant coefficients.
+p < .10. *p < .05. **p <.01. ***p <.001.
Figure 3
Results of the Moderated Serial Multiple Mediation Analysis (H2) – Dependent Variable: Intention to Collaborate With NGOs Supporting Immigrants
Note. The figure shows the non-standardized regression coefficients, B. The coefficient of the direct effect appears in parentheses. Dashed lines represent nonsignificant coefficients.
+p < .10. *p < .05. **p < .01. ***p < .001.
Two statistically significant conditional indirect effects of the similarity to the protagonist were observed on both dependent variables, when the message was written in the first person (see Table 1): through identification (attitudes towards immigration: Effect = .1481, SE = .0709, 95% CI [.0284, .3063]; intention to collaborate with an NGO that supports immigrants: Effect = .1721, SE = .0783, 95% CI [.0340, .3421]) and through the serial mediation of identification and cognitive elaboration (attitudes towards immigration: Effect = .0448, SE = .0226, 95% CI [.0074, .0953]; intention to collaborate with an NGO supporting immigrants: Effect =. 1053, SE = .0451, 95% CI [.0217, .2007]). Thus, H2 received empirical support.
Table 1
Conditional and Unconditional Indirect Effects of Similarity to the Protagonist on the Attitudes Towards Immigration (H2a) and on the Intention to Collaborate With NGOs Supporting Immigrants (H2b) Through Identification and Cognitive Elaboration. Moderated Mediation Models With PROCESS
| Conditional and unconditional indirect effects | Effect | Boot SE | Boot 95% CI |
|---|---|---|---|
| (a) Dependent Variable: Attitudes towards Immigration | |||
| Similarity → Identification → Attitudes towards immigration | |||
| - Third-person narrative | -.0119 | .0564 | [-.1335, .0921] |
| - First-person narrative | .1481 | .0709 | [.0284, .3063] |
| IMM = .1681 (95% CI: .0106, .3702) | |||
| Similarity → Cognitive elaboration → Attitudes towards immigration | -.0228 | .0180 | [-.0636, .0076] |
| Similarity → Identification → Cognitive elaboration → Attitudes towards immigration | |||
| - Third-person narrative | -.0006 | .0176 | [-.0430, .0283] |
| - First-person narrative | .0448 | .0226 | [.0074, .0953] |
| IMM = .0509 (95% CI: .0028, .1167) | |||
| (b) Dependent Variable: Intention to Collaborate with NGOs Supporting Immigrants | |||
| Similarity → Identification → Intention to collaborate with NGOs supporting immigrants | |||
| - Third-person narrative | -.0232 | .0647 | [-.1501, .1064] |
| - First-person narrative | .1721 | .0783 | [.0340, .3421] |
| IMM = .1953 (95% CI: .0136, .4144) | |||
| Similarity → Cognitive elaboration → Intention to collaborate with NGOs supporting immigrants | -.0539 | .0388 | [-.1340, .0178] |
| Similarity → Identification → Cognitive elaboration → Intention to collaborate with NGOs supporting immigrants | |||
| - Third-person narrative | -.0142 | .0400 | [-.0941, .0649] |
| - First-person narrative | .1053 | .0451 | [.0217, .2007] |
| IMM = .1195 (95% CI: .0079, .2448) | |||
Note. Significant indirect effects in bold. IMM = index of moderated mediation (difference between conditional indirect effects).
Discussion and Conclusions
The present work extends previous research on the role of identification with the protagonist as a determining mechanism of the attitudinal impact of narrative messages (Cohen et al., 2020). In this context, our study makes a significant contribution in this field and clarifies two important questions. First, we find it is possible to increase identification through the use of messages that feature (stigmatized) characters who present themselves as similar to the audience, as long as they are presented in the first person. Second, we find a serial mediation process that allows us to conclude that identification is associated with greater cognitive elaboration, which, in turn, is linked with a more positive attitude towards immigration and a greater intention to collaborate with NGOs supporting immigrants.
Given that ingroup–outgroup categorization is a context-dependent process (Turner et al., 1987), we considered that, by activating certain signals in a narrative message about the social identity of its protagonist (highlighting the traits they share with the audience), they could be seen as a member of the ingroup through increasing identification. However, this work was not based on an analysis of the main effects of similarity. For this reason, we suggested that a formal resource that is widely used in the creation of testimonial messages, namely the narrative voice (presenting the written message in the first person), could help to increase the effect of similarity on identification. The results of our study corroborate this approach, revealing that the similarity to the protagonist increased identification only when the presented message was written in the first person. That is, this combination allowed the audience greater access to the internal life of the protagonist, making it easier to adopt their perspective and merge with it. This result is in accordance with those obtained by Kaufman and Libby (2012), although in our case we have focused on a more relevant dimension of similarity in the context of reducing prejudice, namely similarity in terms of social identity. Thusly, our similarity manipulation highlighted what people have in common by referring to subjective aspects such as national feelings and attachment to cultural elements (such as making a positive comment about the host-country’s food).
A key element in understanding how persuasive influence occurs through narrative messages is the analysis of explanatory mechanisms (Bilandzic & Busselle, 2013). The role of counterarguing and reactance has been highlighted by the main theoretical models (Moyer-Gusé, 2008; Slater & Rouner, 2002), although the empirical evidence is not conclusive, so alternative models in which a greater role is given to cognitive elaboration have been proposed (de Graaf & Van Leeuwen, 2017; Igartua & Vega, 2016; Walter & Cohen, 2019). After all, it is perhaps difficult to counterargue and experience reactance when a person shares their biography and recounts their migration process in a testimonial message. It this sense, the experience narrated by the protagonist, who is also presented as a positive role model, can serve as inspiration and stimulate reflection. The process of merging with the protagonist generates a state of temporary involvement that can stimulate cognitive elaboration and, in turn, lead people to question their previous opinions, adjust their attitudes about immigration, and even be willing to collaborate with NGOs supporting immigrants. The results of the present research corroborate this approach, since it was found that identification was associated with greater cognitive elaboration and this, in turn, predicted a more positive attitude towards immigration and a greater intention to collaborate with NGOs supporting immigrants. We consider that the result of serial mediation through identification and cognitive elaboration is robust, being observed in both dependent variables considered herein.
The most important shortcoming of our work is related to the operationalization of similarity. The concept of similarity in terms of social identity is central to this work, and we ponder it to be an innovative contribution to the research on narrative persuasion in which only objective or demographic similarity has been manipulated (Tukachinsky, 2014). However, one may suggest that our manipulation of similarity may be partially linked to acculturation orientations (Berry, 1992): in the high-similarity condition, the protagonist could be displaying an integration profile (high identification with the culture of origin and also with the host culture) or assimilation (high identification with the host culture but low with the culture of origin), while in the low-similarity condition the protagonist could be revealing a separation strategy (high identification with the culture of origin but low with the host culture). For this reason, in future research, it would be interesting to accomplish a better operationalization of similarity in terms of social identity that does not engender doubts about what is being manipulated (similarity versus acculturation strategies).
Regarding this point, we believe that it is reasonable that an immigrant belonging to a highly stigmatized migrant group would experience certain emotional or attitudinal ambivalence, as an expected result of the intricate process of adjustment to a new culture (which offers opportunities but also discriminates against them). This means that it is very likely that a person with these characteristics will have experienced ambivalent and contradictory feelings regarding their life in the host-country, alternating between integration and assimilation and even, on occasions, between separation and marginalization.
It is also necessary to bear in mind that multicultural individuals (such as immigrants, defined as such by having lived in at least two different cultures) tend to change their cultural register based on certain contextual signals, a process that has been defined as cultural frame-switching (Hong et al., 2000). For this reason, people undergoing a process of acculturation may choose to actively control the accessibility of their cultural frameworks through priming behaviors of their self. For example, individuals who wish to adapt quickly to a new culture could surround themselves with symbols and situations that would activate that culture´s system of meanings (e.g., reading local print media). On the contrary, when such people want to reaffirm and experience their ethnic or cultural background, they may surround themselves with stimuli that would activate that culture (e.g., consuming certain types of foods or socializing with other people of the same ethnic or cultural group). In this case, a person could be perceived as “separate” when consuming food or reading written press from their culture of origin or socializing mostly with people from their own group, when in reality they would be acting like this in a temporary way, for example, to deal with nostalgia towards their culture of origin or the stress of acculturation caused by lived experiences of discrimination or rejection. In this sense, one must consider that, in today’s global (eminently mobile) society, cultures are increasingly fluid and dynamic and do not behave so much as an “operating system” but rather as a “set of applications.” In this way, individuals absorb (voluntarily or not) influences from multiple cultures through the media, travel, and intercultural contact, making the relationship between people and culture both partial and plural (Benet-Martínez & Repke, 2020).
The irruption on the political scene of far-right political parties with a clearly xenophobic discourse has made the immigration issue a point of differentiation in Spanish politics. In recent years, the polarization of the immigration debate has intensified, such that agreement between liberals and conservatives on the matter has become increasingly improbable (Pardos-Prado, 2019). This situation results in the stalemate of any legislation that requires a minimum of bipartisan commitment, and it has dire consequences for the integration of immigrants. For this reason, the development of strategies focused on reducing prejudice towards the immigrant collective is necessary. In this context, our work makes a relevant contribution to understanding how to increase the persuasive effectiveness of testimonial messages designed to reduce prejudice toward stigmatized immigrants. Although constituting a less sophisticated form of narrative (Walter & Cohen, 2019), the potential of testimonials is high in the new communication ecosystem, and they could be used in combination with more complex narrative approaches based on entertainment–education formulas (e.g., Moyer-Gusé et al., 2019) and thus form part of campaigns to reduce racism and xenophobia.