Benefits of matching consumers’ personality: Creating perceived trustworthiness via actual self‐congruence and perceived competence via ideal self‐congruence

Zogaj, Adnan; Tscheulin, Dieter K.; Olk, Stephan Article — Published Version Benefits of matching consumers’ personality: Creating perceived trustworthiness via actual self‐congruence and perceived competence via ideal self‐congruence Psychology & Marketing Provided in Cooperation with: John Wiley & Sons Suggested Citation: Zogaj, Adnan; Tscheulin, Dieter K.; Olk, Stephan (2020) : Benefits of matching consumers’ personality: Creating perceived trustworthiness via actual self‐congruence and perceived competence via ideal self‐congruence, Psychology & Marketing, ISSN 1520-6793, Wiley, Hoboken, NJ, Vol. 38, Iss. 3, pp. 416-430, https://doi.org/10.1002/mar.21439 This Version is available at: https://hdl.handle.net/10419/233687 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/ Received: 12 March 2020 | Revised: 5 October 2020 | Accepted: 27 November 2020 DOI: 10.1002/mar.21439 RESEARCH ARTICLE Benefits of matching consumers’personality: Creating perceived trustworthiness via actual self‐congruence and perceived competence via ideal self‐congruence Adnan Zogaj |Dieter K. Tscheulin |Stephan Olk Department of Marketing and Health Care Management, University of Freiburg, Freiburg im Breisgau, Germany Correspondence Dieter K. Tscheulin, Department of Marketing and Health Care Management, University of Freiburg, Rempartstr. 16, 79098 Freiburg im Breisgau, Germany. Email: [email protected] Abstract Endorsers such as social media influencers who aim to match consumers’personality to strengthen their credibility must consider that actual self‐congruence and ideal self‐congruence have different effects on the main dimensions of credibility: perceived trustworthiness and perceived competence. In our study, we conceptually argue that actual self‐congruence is associated more strongly with a closer psychological distance and, in turn, with trustworthy information, and we empirically show that perceived trustworthiness solely mediates the effect of actual self‐ congruence on consumer behavior. Subsequently, we conceptually argue that ideal self‐congruence is more strongly related to ideal skills and therefore should be more strongly related to perceived competence, and we empirically demonstrate that perceived competence solely mediates the effect of ideal self‐congruence on consumer behavior. Our findings explain the mechanism behind the effects of actual and ideal self‐congruence and demonstrate important differences in authentic and aspirational marketing strategies with regard to creating the perception of trustworthiness and competence via social media influencers. KEYWORDS actual self‐congruence, competence, credibility, ideal self‐congruence, purchase intentions, similarity, trustworthiness 1|INTRODUCTION In recent years, so‐called social media influencers have played an important role in social media channels. An increasing number of consumers watch videos and view posts in social media channels to receive product information because they perceive influencers as an important source of credible advice (Casaló et al., 2018; Lou & Yuan, 2019; Schouten et al., 2020). Compared to traditional (offline) opinion leaders, social media influencers generally have a much greater online presence (Casaló et al., 2018; Lyons & Henderson, 2005). Consequently, social media influencer marketing can be found on social media networks such as Instagram, TikTok, Twitter, Facebook, and Snapchat. Brands have already recognized the important role of social media influencers in affecting consumer behavior. In 2018, 39% of 5726 marketers worldwide said that they were working with social media influencers as part of their marketing activities (Stelzner, 2018). Moreover, it is anticipated that the social media influencer Psychol Mark. 2021;38:416–430.416 | wileyonlinelibrary.com/journal/mar 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. © 2020 The Authors. Psychology & Marketing published by Wiley Periodicals LLC marketing market will grow in the future. That is, growth from USD 5.5 billion (2019) to USD 22.3 billion (2024) can be expected (Markets and Markets, 2019). However, for social media influencers, a new challenge has emerged. Social media influencers’credibility has been decreasing (Mscience, 2019; Wavemaker, 2018). In 2019, a study by social media influencers, 46% of followers (13% more than in 2018) stated that they perceived influencers as not being credible (Mscience, 2019). One reason for this change is that an increasing number of consumers have begun to perceive influencers as “paid spokespersons”who are mainly interested in making money through advertising. This challenge raises the question of how social media influencers can make consumers perceive them as being more credible, because credibility is generally an important driver of consumers’perceptions and behaviors (Hussain et al., 2020). Based on the similarity–attraction theory, individuals favor other individuals who are similar to them because perceived similarity leads to cognitive consistencies (Byrne, 1971; Festinger, 1957). On this basis, several studies have found that self‐congruence (i.e., perceived similarity between one's personality and another's personality) positively affects consumer behavior (Bekk et al., 2016; Dolich, 1969; Huber et al., 2018; Shan et al., 2019; Sirgy, 1982). However, another personality (e.g., that of the social media influencer) can match one's actual (i.e., actual self‐congruence) or ideal (i.e., ideal self‐congruence) personality and self‐concept (Malär et al., 2011,2018; Zhu et al., 2019). Several studies in the field of self‐ congruence have supported the positive effects of both actual and ideal self‐congruence on consumer behavior (Ahn, 2019; Dolich, 1969; Ericksen, 1997; Kressmann et al., 2006; Tan et al., 2019; Van Quaquebeke et al., 2019). In addition to the positive effect on consumer behavior, studies have also examined the influence of self‐congruence on credibility (Ismagilova et al., 2020; Nienstedt et al., 2012) and its main dimensions, that is, perceived trustworthiness (e.g., Lou & Yuan, 2019; Yoon & Kim, 2016) and perceived competence, or similar outcomes, such as the evaluation of functional and utilitarian attributes (e.g., Kressmann et al., 2006; Sirgy et al., 1991). However, how the main dimensions (i.e., perceived trustworthiness and perceived competence/expertise) of perceived credibility are connected to actual and ideal self‐congruence and how they mediate the effect of self‐ congruence on consumer behavior remain unclear. That is, no study has examined the relationship of the main dimensions of credibility (i.e., perceived trustworthiness and perceived competence) and the two dimensions of self‐congruence (i.e., actual and ideal self‐ congruence) to explain the different mediating roles of perceived trustworthiness and perceived competence in the effects of actual and ideal self‐congruence on consumer behavior. In this study, we address this study gap by providing a novel model that suggests that perceived trustworthiness mediates the influence of actual self‐congruence on consumer behavior (i.e., purchase intentions), while perceived competence mediates the influence of ideal self‐congruence on consumer behavior. This approach indicates that perceived trustworthiness and perceived competence have competing roles in mediating the effects of actual and ideal self‐ congruence on consumer behavior. Consequently, we contribute to self‐congruence theory by making three important contributions to the self‐congruence literature. First, based on our model, we show how self‐congruence triggers perceived credibility. Doing so enables us to explain the mediating role of credibility and its dimensions in the effect of self‐ congruence on consumer behavior. Moreover, we answer the question of how social media influencers can trigger credibility (i.e., trustworthiness and competence) via matching consumers’ personality and self‐concept. Second, we link construal level theory and psychological distance (Trope & Liberman, 2010) to actual and ideal self‐congruence. We argue that the actual self is related to a closer psychological distance than ideal self‐congruence. Doing so enables us to show the link that explains why authentic marketing strategies and brands are more strongly related to perceived trustworthiness than aspirational marketing strategies and brands. Third, based on prior research (Malär et al., 2018; Sirgy, 1985), we conceptually argue why ideals are linked more strongly to ideal skills than individuals or other objects that match the actual self. Accordingly, we answer the question of why perceived competence should be a better mediator of the effect of ideal self‐congruence on consumer behavior. Doing so enables us to explain the connection between aspirational brands and perceived competence. Overall, providing a novel model that examines the mediators in the effect of actual self‐congruence (through perceived trustworthiness) and ideal self‐congruence (through perceived competence) on consumer behavior enables us to explain the mechanism behind the effects of actual and ideal self‐congruence. 2|CONCEPTUAL AND THEORETICAL BACKGROUND 2.1 |Self‐congruence and consumer behavior Self‐congruence is the perceived match between a consumer's self‐ image and another's image (e.g., an influencer's; see, e.g., Japutra et al., 2019; Liberatore & Tscheulin, 2011; Sirgy, 1982), and it can be divided into two main dimensions: actual self‐congruence and ideal self‐congruence (Malär et al., 2011; Zhu et al., 2019). The actual self (i.e., the basis for actual self‐congruence) describes a person's own self‐image, that is, how one perceives him/herself in the “here and now.”In contrast, the ideal self (i.e., the basis for ideal self‐ congruence) describes the person whom one would like to be (Gough et al., 1978). Consequently, the matching degree of consumers’actual (ideal) personality and an influencer's personality is called actual (ideal) self‐congruence (Huber et al., 2018; Tan et al., 2019). Self‐congruence is of vital importance because similarity–attraction theory indicates that perceived self‐congruence positively affects individuals’perceptions and behavior (Byrne, 1971; Festinger, 1957). That is, perceived similarities—here based on personality and ZOGAJ ET AL. | 417 self‐concept—positively influence individual behavior. Several studies in different contexts (e.g., brand context, tourism context) have supported the positive effects of actual and ideal self‐congruence (see, e.g., Beerli et al., 2007; Bekk et al., 2016; Graeff, 1996; Japutra et al., 2019; Rodrigo et al., 2019). Moreover, studies in the social media context have also shown that self‐congruence between consumers and influencers affects consumer behavior (see, e.g., Shan et al., 2019). The resultsofpriorstudiesthatexaminedself‐endorser congruence or self‐ seller congruence also support the hypothesis that actual and ideal self‐ congruence influence consumer behavior (e.g., Choi & Rifon, 2012;Dion et al., 1995;Smith,1998; Xu & Pratt, 2018). However, although plenty of studies show that actual and ideal self‐congruence positively influence consumer behavior, the mechanism behind the effects of actual and ideal self‐congruence still remains unclear (see, e.g., Malär et al., 2011,2018;Zhuetal., 2019). Indeed, several studies have not found any direct effect of actual or ideal self‐congruence on consumer behavior (see, e.g., Ahn et al., 2013; Choi & Rifon, 2012;Hosany&Martin,2012;Xu& Pratt, 2018). With respect to the endorsers’context, Pradhan et al. (2016) found no direct effect on purchase intentions or brand attitude. In contrast, Shan et al. (2019) showed that ideal self‐congruence with an influencer had a positive effect on consumer behavior (i.e., brand attitude, purchase intentions, and parasocial identification). These ambiguous results can be explained by the coexistence of competing mediator variables that transmit the effects of actual and ideal self‐congruence on consumer behavior (Hayes, 2018; Malär et al., 2011,2018). That is, a nonsignificant direct effect can mean that there are coexisting mediators (see, e.g., Hayes, 2018). Consequently, to understand why and how actual and ideal self‐ congruence influence consumer behavior, it is essential to examine the factors that mediate the effects of actual and ideal self‐ congruence (Hayes, 2018; Malär et al., 2018). 2.2 |The mediating role of credibility Source credibility is an important factor in explaining the influence of endorsers such as social media influencers on consumer behavior (Hussain et al., 2020; Ohanian, 1991). In prior studies, credibility has been treated as one‐dimensional or multidimensional (Wilson & Sherrell, 1993).Becauseweanticipate that actual and ideal self‐congruence will have different effects on the main dimensions of credibility, we use a two‐dimensional approach to examine the mediating role of credibility. The main dimensions of source credibility are perceived expertise/competence and perceived trustworthiness (Ismagilova et al., 2020; Lou & Yuan, 2019; Schouten et al., 2020). Consequently, source credibility represents the extent to which a source is perceived as being trustworthy and competent (Bhattacherjee & Sanford, 2006). Using these two credibility dimensions, we will analyze the indirect effects of actual and ideal self‐congruence on consumer behavior by showing that actual self‐congruence triggers perceived trustworthiness, while ideal self‐congruence triggers perceived competence. Studies have examined the role of trustworthiness to understand why self‐congruence influences consumer behavior (e.g., Badrinarayanan et al., 2014; Dion et al., 1995; Lou & Yuan, 2019; Nyffeneger et al., 2015; Smith, 1998; Zhang et al., 2015). In doing so, several studies in different contexts have been able to support the positive relationship between self‐congruence and perceived trustworthiness and have thus shown that perceived self‐concept similarity positively affects trustworthiness (see, e.g., Badrinarayanan et al., 2014; Dion et al., 1995; Fu et al., 2019; Nienstedt et al., 2012; Zhang et al., 2015). The positive effect of self‐congruence on trust accords with similarity–attraction theory because perceived similarity leads to cognitive consistencies, which make people perceive more trustworthiness (Byrne, 1971; Festinger, 1957; Malär et al., 2011). Further, drawing on construal level theory will answer the question of why both actual and ideal self‐congruence should positively influence perceived trustworthiness. Construal level theory describes the relationship between psychological distance and mental abstractions (Liberman & Trope, 1998; Trope & Liberman, 2010; Trope et al., 2007). Accordingly, everything that does not happen in the here and now can be made visible through abstract imaginings. These construed imaginings are more abstract for future situations (or further perceived situations) than for closer situations (Liberman & Trope, 1998). However, individuals imagine not only future situations but also, for example, other perspectives, remote locations, and counterfactual alternatives. The reference point for construed imaginings is always the self in the here and now, and the further the distance to a situation is (i.e., in time, in space, in social distance, and hypothetically), the more abstract the construed imagining (Trope & Liberman, 2010). Thus, the smaller the psychologically perceived distance, the more reliable and trustworthy the related information will be (Trope et al., 2007). As the perception of both types of self‐congruence decreases the perception of psychological distance, it can be stated that perceived actual as well as perceived ideal self‐congruence increases perceived trustworthiness. This argument finds support in Choi and Reid (2016) who analyzed the effect of self‐congruence on claim believability (i.e., the perception of information trustworthiness). In doing so, they treated self‐ congruence as two‐dimensional (i.e., actual and ideal self‐ congruence). Their results showed that self‐congruence positively influences the perception of believability. Furthermore, trust strengthens the relationship between two actors because it decreases perceived uncertainty and, in turn, helps one person rely on another (Morgan & Hunt, 1994). Consequently, trust triggers positive behavioral responses such as purchase intentions (Badrinarayanan et al., 2014; Chaudhuri & Holbrook, 2001; Hussain et al., 2020). Summarizing the above considerations on the relationship between self‐congruence, trust, and purchase intentions, we hypothesize the following: H1. Perceived trustworthiness mediates the effect of (a) actual self‐congruence and (b) ideal self‐congruence on purchase intentions. 418 | ZOGAJ ET AL. Given that both actual and ideal self‐congruence influence perceived trustworthiness, whether actual or ideal self‐congruence has a stronger effect on perceived trustworthiness must be asked to understand whether perceived trustworthiness better mediates the effect of actual or ideal self‐congruence on consumer behavior. Thinking of the construal level theory and considering that individuals have an actual and an ideal self raises the question of which of these constructs is closer (i.e., stronger related) to the here and now. As the ideal self refers to the person whom one would like to be (Sirgy, 1985; Tan et al., 2019), the ideal self is psychologically further away from the here and now (Malär et al., 2011). Consequently, the actual self is related to a smaller psychological distance than the ideal self. As we know from construal level theory, each abstraction of something/somewhere/someone that is not directly related to the here and now means that information is lost because, with a greater distance, less reliable and trustworthy information is available (Trope et al., 2007). In turn, because more reliable and trustworthy information is linked to the here and now (i.e., actual self), actual self‐congruence has a stronger influence on perceived reliability and trustworthiness than does ideal self‐congruence. This hypothesis finds support in Tan et al. (2019), who found that actual self‐ congruence is more important than ideal self‐congruence in present‐ focused situations. Summarizing the above considerations on the relationship between self‐congruence and construal level theory, we hypothesize the following: H2. The effect of actual self‐congruence on purchase intentions via perceived trustworthiness is stronger than the effect of ideal self‐ congruence on purchase intentions via perceived trustworthiness. Karjaluoto et al. (2019) found that self‐congruence is positively related to the perceived utilitarian value of services. More specifically, the authors showed that perceived congruence biased consumers in evaluating utilitarian value because consumers with higher self‐congruence evaluated the perceived utilitarian value of mobile financial services better than those with low self‐congruence. Utilitarian value refers to functional and practical benefits that are based on perceived performance and quality benefits (Aw et al., 2019; Sweeney & Soutar, 2001). Other studies support the positive relationship between self‐congruence and perceived performance and quality by showing that self‐congruence positively affects the evaluation of functional attributes and perceived expertise (see, e.g., Aw et al., 2019; Yoon & Kim, 2016). The positive relationship between self‐congruence and the evaluation of functional and utilitarian attributes (i.e., perceived performance, expertise, and quality) occurs because individuals evaluate self‐congruence more subconsciously and before they evaluate functional and utilitarian characteristics (Sirgy et al., 1991). That is, self‐congruence drives individuals to form a favorable attitude and to favorably evaluate perceived utilitarian and functional aspects (Kressmann et al., 2006). Many studies have proven this bias evaluation effect (Choi & Reid, 2016; Karjaluoto et al., 2019; Kressmann et al., 2006; Sirgy et al., 1991). Furthermore, a better evaluation of functional/utilitarian attributes and perceived competence positively affects consumer behavior (e.g., satisfaction, commitment, and revisit intentions; Aw et al., 2019; Karjaluoto et al., 2019; Sirgy et al., 1991; Yoon & Kim, 2016). The positive effect of utilitarian and functional value on consumer behavior has also been supported in the context of influencers (see, e.g., Ismagilova et al., 2020; Lou & Yuan, 2019). Summarizing the above considerations on the relationship between self‐congruence, perceived competence, and consumer behavior, we hypothesize the following: H3. Perceived competence mediates the effect of (a) actual self‐ congruence and (b) ideal self‐congruence on purchase intentions. Again, the question of whether actual or ideal self‐congruence may have a stronger effect on the mediator arises. While actual self‐ congruence refers to consumers’personality traits in the here and now, ideal self‐congruence refers to consumers’ideal skills (Malär et al., 2018; Sirgy, 1982,1985). Because ideal self‐congruence is related to ideal skills, it is plausible that ideal self‐congruence has a stronger effect on perceived functional value than does actual self‐ congruence (Huber et al., 2010; Hung & Petrick, 2011; Sirgy et al., 1991). The stronger effect of ideal self‐congruence on the evaluation of functional attributes is supported in several studies (see., e.g., Huber et al., 2010; Hung & Petrick, 2011; Sirgy et al., 1991). Reiterating that functional value is related to competence (Aw et al., 2019; Sweeney & Soutar, 2001) and summarizing the above considerations, we hypothesize the following: H4. The effect of ideal self‐congruence on purchase intentions via perceived competence is stronger than the effect of actual self‐congruence on purchase intentions via perceived competence. 3|STUDY 1 3.1 |Research design and participants To test our hypotheses, we conducted a study with a social media influencer. Social media influencers use social media networks such as Facebook, Instagram, TikTok, and Twitter and exist in a variety of fields such as fashion, health and fitness, beauty, and electronic devices. They shape audiences, especially via tweets, blogs, and other kinds of posts (Freberg et al., 2011). In 2018, a social media influencer, who evaluates Samsung products and presents them to his followers, uploaded a smartphone review video to Facebook (>12,000 followers) and Twitter (>10,000 followers). This social media influencer runs an unofficial Samsung blog and creates reviews of Samsung products from all areas, which he then posts on various social media channels. In those reviews, products are evaluated holistically with a focus on technology according to the opinion of the social media influencers. On the homepage of the social media influencer, it is further mentioned that these subjective reviews cannot be influenced by Samsung as it is not an official Samsung blog ZOGAJ ET AL. | 419 but an independent one. Consequently, his followers are individuals who are interested in electronic devices such as smartphones. After uploading the Samsung smartphone video, he posted a link to our online survey on Facebook and Twitter and asked his followers to participate. Data were collected for three weeks. His followers participated voluntarily and without receiving any incentives. A total of 223 subjects participated in the study; however, 14 questionnaires were not completed. Further, 12 questionnaires were deleted to ensure that the participants completed the survey in a reasonable time. In doing so, we followed the procedure of a globally active data collection institute and deleted the subjects who took less than half the median time to complete the survey. Therefore, the final sample consists of 197 followers. Seventy‐four percent of the subjects were male, and the average age was 30 (min = 16; max = 58; median = 28). Furthermore, 33.5% of our sample were employed, 29.4% were students, 11.2% were officials, 9.1% were self‐employed, 8.1% were apprentices, and the rest had another job status. The salary distribution (net per month) is as follows: 32.5% earned ≤1000€, 14.7% earned 1001–1500€, 11.7% earned 1501–2000€, 22.8% earned 2001–2500€, 12.2% earned more than 2500€, and 6.1% did not answer this question. 3.2 |Measures We used previously empirically validated constructs to measure the subjects’responses. Based on a 7‐point Likert scale, the answers ranged from 1 = strongly disagree to 7 = strongly agree. Solely perceived competence was measured based on a 7‐point bipolar scale. At the beginning of the survey, actual and ideal self‐congruence were measured directly and globally according to Malär et al. (2011) and Sirgy et al. (1997) because, as Sirgy et al. (1997) showed, measuring self‐congruence directly is more valid than measuring self‐ congruence based on discrepancy indices by using predetermined personal characteristics. According to the literature, before asking self‐congruence questions, the subjects were asked to think about the person in the video (i.e., the influencer) and his personality. Next, we measured actual self‐congruence with two items: 1. The personality of the person in the video is consistent with how I see myself; and 2. The person in the video is a mirror image of me. Subsequently, we also measured ideal self‐congruence with two items: 1. The personality of the person in the video is consistent with how I would like to be; and 2. The person in the video is a mirror image of the person whom I would like to be. In turn, we measured trust with three items based on Ohanian (1990) and Chaudhuri and Holbrook (2001): 1. The person in the video is trustworthy; 2. What the person says in the video is reliable; and 3. The person in the video is honest. Next, based on Voss et al. (2003) and Fang (2014), we measured the perceived competence of the source (i.e., functional and practical benefits). Therefore, we asked subjects to evaluate how “…” they perceived the video and its content: 1. not functional/functional, 2. unnecessary/necessary, 3. useless/useful, and 4. not sensible/sensible. Finally, we measured the dependent behavioral variable (i.e., purchase intentions) based on Schnebelen and Bruhn (2018): 1. The next time I buy a smartphone, I will (again) buy one from the “X”brand; 2. I plan to continue to buy this brand's products; and 3. The next time I buy a smartphone, I plan to buy one from “X.” 3.3 |Results We ran a reliability analysis in SPSS 25 and a confirmatory factor analysis in AMOS 25 to test the model fit and to evaluate the psychometric properties of the items. In doing so, all items were simultaneously considered in one model. Furthermore, the items were restricted to load onto specific factors. The results are presented in Appendix A. The Cronbach's αvalues and composite reliability (CR) values were above 0.80, indicating internal consistency in the scale items (Nunnally, 1978). Moreover, all standardized regression weights had appropriate values and were significant (actual self‐ congruence ≥0.940; ideal self‐congruence ≥0.956; perceived trustworthiness ≥0.744; perceived competence ≥0.645; purchase intentions ≥0.771). Furthermore, the average variance explained (AVE) values were higher than the maximum shared variance (MSV) values and higher than the threshold of 0.50. Additionally, the Fornell–Larcker criterion was met because the square root of each AVE was higher than the correlations between the constructs. Consequently, based on the statistics of the reliability and confirmatory factor analyses, the constructs show good discriminant and convergent validity (Hair et al., 2010). Next, we checked the goodness of fit of the five‐factor model based on the recommended indices of Byrne (2016) and Schreiber et al. (2006). The χ 2 /df was 1.769 (χ 2 = 118.529; df = 67) and lower than the threshold of 3.0. The standardized root mean square residual (RMR) was 0.048 and lower than the threshold of 0.08. The comparative fit index (CFI) value was 0.976, the incremental fit index (IFI) value was 0.976, and the Tucker–Lewis index (TLI) value was 0.968. These values were higher than the threshold of 0.95. Finally, the root mean square error of approximation (RMSEA) was 0.063 (<0.08), and the pof close fit (PClose) was 0.128 (>0.05). Accordingly, the indices show that the proposed five‐factor model fits the data well. Next, based on Japutra et al. (2019), two models (i.e., the full mediation model and the partial mediation model) were tested. In the full mediation model, the paths from actual self‐congruence and ideal self‐congruence to purchase intentions are not available. In addition, we considered age and sex as control variables. The results are presented in Table 1. However, before analyzing the results, we checked the goodness of fit of the models based on the indices recommended by Byrne (2016) and Schreiber et al. (2006). First, we checked the model fit of the full mediation model: the χ 2 /df was 1.726 (χ 2 = 150.129; df = 87), the CFI value was 0.971, the IFI value was 0.972, the TLI value was 0.960, and the RMSEA was 0.061 (<0.08). Then, we checked the model fit of the partial mediation 420 | ZOGAJ ET AL. model: the χ 2 /df was 1.744 (χ 2 = 148.244; df = 85), the CFI value was 0.971, the IFI value was 0.972, the TLI value was 0.959 (>0.95), and the RMSEA was 0.062 (<0.08). As shown in Table 1, actual self‐congruence (SRW = 0.324; p< 0.001) has a stronger effect on perceived trustworthiness than does ideal self‐congruence (SRW = −0.037; p> 0.05). The results also indicate that the influence of ideal self‐congruence on perceived competence (SRW = 0.217; p< 0.05) is stronger than the influence of actual self‐congruence on perceived competence (SRW = 0.098; p> 0.05). In conclusion, these results show that actual self‐ congruence is linked to perceived trustworthiness, while ideal self‐ congruence is linked to perceived competence. Furthermore, the results of this study show that both perceived trustworthiness (SRW = 0.250; p< 0.01) and perceived competence (SRW = 0.227; p< 0.05) positively affect purchase intentions. This finding indicates the mediating role of perceived trustworthiness and perceived competence in the effect of actual self‐congruence (i.e., perceived trustworthiness) and ideal self‐congruence (i.e., perceived competence) on purchase intentions. Furthermore, the results show that actual and ideal self‐ congruence do not directly influence purchase intentions. That is, perceived trustworthiness and perceived competence fully mediate the effects of actual and ideal self‐congruence on purchase intentions. However, according to Hayes (2018), nonsignificant direct effects can mean that there are other (competing) mediators. In this regard, Malär et al. (2018) showed that, for example, envy is a negative mediator in the effect of ideal self‐congruence on consumer behavior. To further test the hypotheses, we calculated specific indirect effects of the partial mediation model based on a bias‐corrected 95% bootstrap confidence interval (CI) and 5000 bootstrap samples in AMOS 25 as recommended by the literature (Arbuckle, 2017; Hayes, 2018; Zogaj et al., 2019). To calculate the specific indirect effects of actual and ideal self‐congruence on purchase intentions via perceived trustworthiness and perceived competence, we defined the specific estimands (specific indirect effects) based on Arbuckle (2017). The results are presented in Table 2. As shown in Table 2, the CI for the indirect effect of actual self‐ congruence on purchase intentions via perceived trustworthiness (b= 0.054*; 95% CI = 0.009–0.137) and the CI for the indirect effect of ideal self‐congruence on purchase intentions via perceived competence (b= 0.025**; 95% CI=0.005–0.068) exclude zero. However, the CI for the indirect effect of actual self‐congruence on purchase intentions via perceived competence (b= 0.015 n.s. ; 95% CI = −0.010 to 0.062) and the CI for the indirect effect of ideal self‐congruence on purchase intentions via perceived trustworthiness (b=−0.005 n.s. ; 95% CI = −0.049 to 0.015) include zero. As hypotheses H2 and H4 indicate a statistical difference between paths, we assessed the statistical difference between the parameter estimates even if one of the paths was not significant (Rodríguez‐Entrena et al., 2018). The results indicate that there is only a significant difference in the paths of actual and ideal self‐congruence via trustworthiness (95% CI = 0.051–0.308]), but not through competence (95% CI = −0.005 to 0.201), based on a 95% CI. However, with respect to competence, it can be seen that the lower bound is near zero. Therefore, next, we assessed the statistical difference based on a 90% CI. As expected, the results showed that the difference is significant based on a 90% CI (90% CI = 0.011–0.182). In summary, perceived trustworthiness solely mediates the effect of actual self‐congruence on purchase intentions, while perceived competence solely mediates the effect of ideal self‐ congruence on purchase intentions. Consequently, hypothesis H1a is TABLE 1 Results of full and partial mediation in Study 1 Full mediation Partial mediation Path SRW CR SRW CR Actual self‐congruence →trustworthiness 0.329 4.281*** 0.324 4.228*** Ideal self‐congruence →trustworthiness −0.049 −0.608 n.s. −0.037 −0.465 n.s. Actual self‐congruence →competence 0.097 1.193 n.s. 0.098 1.211 n.s. Ideal self‐congruence →competence 0.219 2.480*0.217 2.475* Trustworthiness →purchase intentions 0.249 2.754** 0.250 2.576** Competence →purchase intentions 0.203 2.251*0.227 2.465* Actual self‐congruence →purchase intentions 0.035 0.397 n.s. Ideal self‐congruence →purchase intentions −0.121 −1.391 n.s. Squared multiple correlations Purchase intentions 14.6% 15.7% Note:N= 197. Abbreviations: CR, composite reliability; n.s., not significant; SRW, standardized regression weight. *p< 0.050. **p< 0.010. ***p< .001. ZOGAJ ET AL. | 421 supported because perceived trustworthiness mediates the effect of actual self‐congruence on purchase intentions. Since perceived trustworthiness does not mediate the effect of ideal self‐congruence on purchase intentions, hypothesis H1b must be rejected. Further, we tested if the effect of actual self‐congruence on purchase intentions via trustworthiness is stronger. The statistical difference between the effects could be supported on a 90% CI. It follows that the effect of actual self‐congruence on purchase intentions via trustworthiness is stronger. Therefore, hypothesis H2 is supported. Furthermore, hypothesis H3b is supported, and hypothesis H3a must be rejected because perceived competence solely mediates the effect of ideal self‐congruence on purchase intentions. Finally, hypothesis H4 is supported because the effect of ideal self‐congruence on purchase intentions via perceived competence is significantly stronger. Moreover, the standardized total effects of actual and ideal self‐ congruence on purchase intentions are not significant. That is plausible as both actual and ideal self‐congruence have no direct effects on purchase intentions; moreover, actual (ideal) self‐congruence has no indirect effect through competence (trustworthiness). These (significant and nonsignificant) direct and indirect effects result in a nonsignificant total effect (see, e.g., Hayes, 2018). 4|STUDY 2 4.1 |Research design and participants To extend the generalizability of our findings from Study 1, we conducted Study 2. That is, we tested the model in a different, more general social media influencer context. Thus, we provided the survey link on Instagram as Instagram is the most used social media platform with respect to following an influencer (Casaló et al., 2018; Mscience, 2019). The survey link was shared using the “snowball method.”This means that we asked student assistants to post the link in their bio on Instagram and draw attention to it in their stories. These student assistants also asked acquaintances to post the link in their bio on Instagram and to draw attention to it in their stories. Incentives were not given. In contrast to Study 1, in Study 2 we aimed to generate answers based on the perceptions of several influencers due to more generalization. Therefore, we asked Instagram users whether they follow an influencer. However, first, we asked them if they know what a social media influencer is. Subsequently, they gave the name of the social media influencer and answered the question based on their perception of that influencer. Thus, in Study 2, we not only focus on one social media influencer and his followers but on several social media influencers and several product types to test the model in a more general social media influencer context. A total of 437 subjects participated in the study; however, 18 subjects answered that they do not know what a social media influencer is and 110 subjects answered that they do not follow any social media influencers. Two more responses were deleted as they took less than half of the median time to answer the survey. Therefore, the final sample consists of 307 subjects. Sixty percent of the subjects were female, and the average age was 27 (min = 16; max = 65; median = 26). Furthermore, 42.3% of our sample were employed, 31.3% were students, 4.2% were officials, 4.6% were apprentices, 4.6% were self‐employed, 1.3% were mini‐ jobbers, and the rest had another job status. The salary distribution (net per month) is as follows: 29.6% earned ≤1000€, 20.5% earned 1001–2000€, 18.9% earned 2001–3000€, 9.1% earned 3001–4000€, 8.8% earned more than 4000€, and 13.0% did not answer this question. 4.2 |Measures We again used previously empirically validated constructs to measure the subjects’responses. Based on a 7‐point Likert scale, the answers ranged from 1 = strongly disagree to 7 = strongly agree. Actual and ideal self‐congruence were measured based on Malär et al. (2011) and Sirgy et al. (1997). That is, we first asked the subjects to think about the social media influencer. Next, we measured actual self‐congruence with two items: 1. The personality of “name” is consistent with how I see myself (my actual self); and 2. The personality of “name”is a mirror image of me (my actual self). Subsequently, we also measured ideal self‐congruence with two items: 1. The personality of “name”is consistent with how I would like to be TABLE 2 Indirect effects in Study 1 Bias‐corrected CI (95%) Path Specific indirect effect pLower Upper Actual self‐congruence via trustworthiness 0.054*0.019 0.009 0.137 Ideal self‐congruence via trustworthiness −0.005 n.s. 0.484 −0.049 0.015 Actual self‐congruence via competence 0.015 n.s. 0.203 −0.010 0.062 Ideal self‐congruence via competence 0.025** .009 0.005 0.068 Note: N = 197. Abbreviation: CI, confidence interval. *p< 0.050. **p< 0.010. 422 | ZOGAJ ET AL. (my ideal self); and 2. The personality of “name”is a mirror image of the person I would like to be (my ideal self). In turn, we measured perceived trustworthiness and competence based on Ohanian (1990)withfouritemsbasedona7‐point bipolar scale. To measure trustworthiness, we asked the subjects to think about the influencer and to respond to the following items: 1. Dishonest/honest, 2. Untrustworthy/trustworthy, 3. Insincere/ sincere, and 4. Unreliable/reliable. With respect to perceived competence, the subjects were asked to respond to the following items: 1. Not an expert/expert, 2. Unknowledgeable/knowledgeable, 3. Unqualified/qualified, and 4. Unskilled/skilled. Finally, we measured purchase intentions based on two items as recommend by Ki and Kim (2019): 1. In the future, I am likely to try one of the products/services “name”posted on his/her Instagram, and 2. In the future, I am likely to try one of the brands “name”recommended/ posted on his/her Instagram. 4.3 |Results The results of the reliability analysis and confirmatory factor analysis are presented in Appendix A. The Cronbach's αvalues and CR values were above 0.80, indicating internal consistency in the scale items (Nunnally, 1978). Moreover, all standardized regression weights had appropriate values and were significant (actual self‐ congruence ≥0.873; ideal self‐congruence ≥0.921; trustworthiness ≥0.845; competence ≥0.726; purchase intentions ≥0.931). Furthermore, the AVE values were higher than the MSV values and higher than the threshold of 0.50. Additionally, the Fornell–Larcker criterion was met because the square root of each AVE was higher than the correlations between the constructs. Consequently, based on the statistics of the reliability and confirmatory factor analyses, the constructs show good discriminant and convergent validity (Hair et al., 2010). Next, we checked the goodness of fit of the five‐factor model based on the recommended indices of Byrne (2016) and Schreiber et al. (2006). The χ 2 /df was 2.113 (χ 2 = 141.593; df = 67) and lower than the threshold of 3.0. The standardized RMR was 0.035 and lower than the threshold of 0.08. The CFI value was 0.978, the IFI value was 0.979, and the TLI value was 0.971. These values were higher than the threshold of 0.95. Finally, the RMSEA was 0.060 (<0.08), and the PClose was 0.107 (>0.05). Accordingly, the indices show that the proposed five‐factor model fits the data well. Next, we estimated the two models (i.e., the full mediation model and the partial mediation model) as recommended by Japutra et al. (2019). In doing so, we again considered age and sex as control variables. Further, we considered familiarity with the influencer as control variables as familiarity could influence the effects of actual and ideal self‐congruence on consumer behavior (see, e.g., Malär et al., 2011) The results are presented in Table 3. However, first, we assessed the goodness of fit of the models based on the indices recommended by Byrne (2016) and Schreiber et al. (2006). We started with the model fit of the full mediation model: the χ 2 /df was 1.964 (χ 2 = 188.539; df = 96), the CFI value was 0.974, the IFI value was 0.974, the TLI value was 0.963 (>0.95), and the RMSEA was 0.056 (<0.08). Then, we checked the model fit of the partial mediation model: the χ 2 /df was 1.861 (χ 2 = 174.924; df = 94), the CFI value was 0.977, the IFI value was 0.977, the TLI value was 0.967, and the RMSEA was 0.053 (<0.08). As shown in Table 3, actual self‐congruence (SRW = 0.283; p< 0.01) has a stronger effect on trustworthiness than ideal self‐ congruence does (SRW = −0.010; p> 0.05). The results also indicate TABLE 3 Results of full and partial mediation in Study 2 Full mediation Partial mediation Path SRW CR SRW CR Actual self‐congruence →trustworthiness 0.286 2.991** 0.283 2.957** Ideal self‐congruence →trustworthiness −0.010 −0.105 n.s. −0.010 −0.111 n.s. Actual self‐congruence →competence −0.137 −1.300 n.s. −0.141 −1.330 n.s. Ideal self‐congruence →competence 0.382 3.664*** 0.383 3.663*** Trustworthiness →purchase intentions 0.238 3.521*** 0.152 2.179* Competence →purchase intentions 0.229 3.420*** 0.232 3.410*** Actual self‐congruence →purchase intentions 0.204 1.984* Ideal self‐congruence →purchase intentions 0.036 0.360 n.s. Squared multiple correlations Purchase intentions 23.9% 27.8% Note:N= 307. Abbreviations: CR, composite reliability; n.s., not significant; SRW, standardized regression weight. *p< 0.050. **p< 0.010. ***p< 0.001. ZOGAJ ET AL. | 423 APPENDIX A Tables A1–A4. TABLE A1 Descriptive parameters – Study 1 Construct Mean scores SD Variance Kurtosis SE Kurtosis VIF Actual self‐congruence 4.523 1.432 2.049 −.045 0.345 1.410 Ideal self‐congruence 3.531 1.829 3.343 −1.062 0.345 1.342 Trustworthiness 5.953 0.877 0.770 5.119 0.345 1.393 Competence 5.585 0.983 0.966 2.515 0.345 1.336 Purchase intentions 5.865 0.927 0.859 4.807 0.345 DV Abbreviations: DV, dependent variable; SD, standard deviation; SE, standard error; VIF, variance inflation factor. TABLE A2 Construct reliability and validity –Study 1 Construct αCR AVE MSV F1 F2 F3 F4 F5 F1 0.962 0.963 0.928 0.223 0.963 F2 0.979 0.980 0.961 0.223 0.472*** 0.980 F3 0.861 0.862 0.677 0.243 0.406*** 0.267*** 0.823 F4 0.883 0.884 0.660 0.243 0.216** 0.308*** 0.493*** 0.812 F5 0.886 0.889 0.729 0.112 0.121 0.006 0.334*** 0.306*** 0.854 Note: F1 = actual self‐congruence; F2 = ideal self‐congruence; F3 = trustworthiness; F4 = competence; F5 = purchase intentions; AVE = average variance extracted; CR = composite reliability; MSV = maximum shared variance; in bold: square root of the AVE; in italics: correlations between constructs; significance: **p< 0.010, ***p< 0.001. TABLE A3 Descriptive parameters – Study 2 Construct Mean scores SD Variance Kurtosis SE Kurtosis VIF Actual self‐congruence 2.910 1.396 1.949 −.778 0.277 1.912 Ideal self‐congruence 3.676 1.640 2.691 −.951 0.277 1.923 Trustworthiness 5.467 1.159 1.344 1.454 0.277 1.444 Competence 5.293 1.194 1.425 −.096 0.277 1.349 Purchase intentions 3.555 1.824 3.329 −1.121 0.277 DV Abbreviations: DV, dependent variable; SD, standard deviation; SE, standard error; VIF, variance inflation factor. TABLE A4 Construct reliability and validity –Study 2 Construct αCR AVE MSV F1 F2 F3 F4 F5 F1 0.868 0.869 0.768 0.562 0.876 F2 0.921 0.921 0.853 0.562 0.749*** 0.923 F3 0.923 0.923 0.751 0.268 0.390*** 0.362*** 0.866 F4 0.914 0.921 0.746 0.268 0.192** 0.294*** 0.518*** 0.864 F5 0.931 0.931 0.870 0.161 0.365*** 0.338*** 0.402*** 0.393*** 0.933 Note: F1 = actual self‐congruence; F2 = ideal self‐congruence; F3 = trustworthiness; F4 = competence; F5 = purchase intentions; AVE = average variance extracted; CR = composite reliability; MSV = maximum shared variance; in bold: square root of the AVE; in italics: correlations between constructs; significance: **p< 0.010, ***p< 0.001. 430 | ZOGAJ ET AL.