XXX Impact of The Perceived Risk

Technological Forecasting & Social Change 184 (2022) 121997
Contents lists available at ScienceDirect
Technological Forecasting & Social Change

journal homepage: www.elsevier.com/locate/techfore
Impact of the perceived risk in influencers' product recommendations on

their followers' purchase attitudes and intention
L. Javier Cabeza-Ramírez *, Sandra M. Sánchez-Cañizares, Luna M. Santos-Roldán,
Fernando J. Fuentes-García
Department of Business Organization, University of Córdoba, Spain
A R T I C L E I N F O A B S T R A C T
Keywords: The impact of the use of influencers in recommending certain products to the audience has been highlighted in
Influencer marketing recent literature on social media marketing. However, this literature has barely considered the possible
Purchase intention perception of risk associated with these recommendations on the part of consumers or the involvement of users
Social media
with the product in their purchase intention. This paper fills this gap in the literature by analysing the possible
Perceived risk
Product endorsement
interconnection between observational learning, product involvement and perceived risk on the effectiveness of
influencers' recommendations. To this end, we propose a new exploratory model applying an approach grounded
in observational learning and an extension of the theory of planned behaviour, analysed by means of PLS-SEM in
a sample of 948 followers of influencers who recommend fashion products. The results have new implications for
influencer marketing research by reporting the impact of risk on product attitude and purchase intention,
although their predictive power was lower than expected. Furthermore, perceived authenticity of the message
exerted the highest positive effect on attitude, and it was established as a significant mitigator of user-perceived
risk. Finally, product involvement, a construct unexplored in previous studies, revealed a considerable influence
on both perceived authenticity of the message and attitude.
1. Introduction influencer marketing seeks to create direct, organic interactions through

favourable word-of-mouth on social networks (eWOM), and promoted
Given the global rise of what is known as the digital economy, social by users with a high capacity to exert influence over other people (the
media marketing has become an omnipresent activity and an essential audience), while shaping perceptions, feelings and experiences
element of the promotional strategies of almost any company (AlSharji regarding a recommended product or service (Martínez-López et al.,
et al., 2018; He et al., 2017; Yuan et al., 2021). The mass diffusion of all 2020; Petrescu et al., 2018). Contrary to what it may seem, the use of
kinds of content through established social media platforms such as influencers is not a new phenomenon in marketing but an adaptation to
Facebook, Twitter, Instagram and Youtube (Arora et al., 2019), the the particularities of social media (Bugshan and Attar, 2020). Until not
continuing emergence of increasingly dynamic and versatile new plat too many years ago, it was common to rely on famous celebrities from
forms such as Twitch or TikTok (Cabeza-Ramirez et al., 2021; Zhang, different fields such as film, music, sport or fashion to influence con
2021), their unprecedented incorporation into people's daily lives sumer behaviour (Masuda et al., 2022). Social media influencers are
(Tafesse and Wood, 2021), as well as the increase in their popularity, has independent third-party sponsors who help form the attitudes of their
provided companies and organizations with new opportunities to raise followers through videos, photos, comments, blogs, tweets and other
brand awareness, and a novel way to attract customers and improve posts (Freberg et al., 2011). These are people who are very active on one
relationships with them (Braojos-Gomez et al., 2015; Khan et al., 2014; or more of the aforementioned platforms and who have a well-
Lou and Yuan, 2019). This is due in part to the unique features of social established identity reflected in a growing audience and considerable
media, including ease of access and openness, strong engagement, and online fame (Arora et al., 2019; Hudders et al., 2021).
content sharing between brand and user as well as among users (Khan The latest literature reviews in the field of influencer marketing
et al., 2014). An emerging field has harnessed some of these benefits: research reveal certain knowledge gaps (De Veirman et al., 2019; Enke
* Corresponding author at: Faculty of Law, Economic and Business Sciences, University of Córdoba, Puerta Nueva s/n, E-14071 Córdoba, Spain.
E-mail address: [email protected] (L.J. Cabeza-Ramírez).
https://doi.org/10.1016/j.techfore.2022.121997
Received 22 January 2022; Received in revised form 29 June 2022; Accepted 24 August 2022
Available online 5 September 2022
0040-1625/© 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
L.J. Cabeza-Ramírez et al. Technological Forecasting & Social Change 184 (2022) 121997
and Borchers, 2019; Hudders et al., 2021; Vrontis et al., 2021). One of (Chetioui et al., 2020; Shin and Lee, 2021; Wiedmann et al., 2010).
the most prominent lines of research focuses on the effects of recom
mendations made by influencers on the attitudes and purchase intention 2. Literature review and hypothesis development
of their audience (Chetioui et al., 2020; Dwidienawati et al., 2020;
Halim et al., 2020; Hudders et al., 2021). However, this approach has 2.1. Background
barely taken into account the presence of risks arising from sponsorship,
apart from pointing out some danger linked to advertising in disguise, or Global social media usage continues its upward trend: 4.55 billion
the occasional lack of ethics in terms of the message conveyed (Boerman people around the world now use social media and 400 million new
et al., 2017; Casalo et al., 2020; Hudders et al., 2021). Nevertheless, users have come online within the last 12 months; specifically, in
from the perspective of risk theory, it seems unlikely that an influencer's October 2021, 57.6 % of the world population were daily users of social
followers do not foresee any kind of possible anticipated loss related to networks, spending an average time of 2 h and 27 min a day (Smar
the purchase of a product (Forsythe and Shi, 2003; Taylor, 1974). In this tInsights, 2021). Another way of looking at it is that social media plays a
regard, previous research has mainly considered adjustments related to significant role in the lives of more than half of the world's population.
the positive triggers that activate the complex psychological process of This has spurred exponential growth in the use of influencers as strategic
purchase, including: credibility, attractiveness, parasocial interaction brand ambassadors making sponsored product recommendations
and congruence (Enke and Borchers, 2019; Hudders et al., 2021), or how (Boerman, 2020; Martínez-López et al., 2020). In this regard, Saima and
interactive numerical cues (likes, shares, comments) have the potential Khan (2021) note that when a brand hires or sponsors an influencer it
to increase the perceived authenticity of messages (Chen et al., 2011; does so because the influencer has become relevant in a specific domain,
Zafar et al., 2021). It is difficult to find studies that analyse the impact of and the influencer's followers readily accept their recommendations.
risk, or at least shed light on the mechanisms that mitigate the danger WeAreSocial (2020), an agency specialized in marketing and commu
derived from an influencer's recommendations (Chopra et al., 2020; nication, recently reported that during the toughest months of the
Deshbhag and Mohan, 2020). On the other hand, research has been pandemic (April and May 2020) “median engagement rates for influencers
undertaken on different types of products: food, fashion, lifestyle, luxury were a massive 43 times higher per post for influencers than for advertisers
brands and others (Chetioui et al., 2020; Coates et al., 2019; Trivedi and [...] (259,307 versus 5,906), according to [...] comparisons of the top 3,500
Sama, 2020). Yet, the possible effects of the product involvement of the influencers versus the top 3,500 brands on Instagram.” Furthermore, these
users receiving the recommendations have not been assessed either people, who are considered micro-celebrities (Gaenssle and Budzinski,
(Mou et al., 2020; Zaichkowsky, 1985). For example, if an influencer 2021), have a special connection with their audience, generate persua
recommends a green product and the audience is highly environmen sive content (Sokolova and Perez, 2021), and disseminate recommen
tally conscious, it will seem likely that their attitudes and intentions dations to highly engaged followers (Hudders et al., 2021). All this
would be more influenced, and they would even perceive their message enables certain brands to reach specific niches of consumers who are
as more authentic. In this context, researchers and marketers alike need familiar with the use of social networks (Raun, 2018). Under these cir
more information on additional factors that affect the success or failure cumstances, more and more companies are attempting to gain a
of influencer marketing campaigns on social media (Hudders et al., competitive advantage through campaigns using influencers (Trivedi
2021; Vrontis et al., 2021). In particular, the complex psychological and Sama, 2020). In some cases, networks can even become saturated
process of attitude and purchase intention is presumably affected by the with sponsored posts, diminishing the persuasive power of influencer
interconnections of these three elements (observational learning, prod marketing and generating scepticism about the recommendations (Lin
uct involvement, and perceived risk). That is, it would be feasible to et al., 2017; Martínez-López et al., 2020; Petrescu et al., 2018).
think that product involvement, and the number of likes, comments and In the field of marketing it is striking to observe how the figure of the
shared content generated, would enhance the authenticity of the mes influencer fits perfectly with the definition of marketing (Kotler and
sage. In turn, these could be important elements when it comes to Zaltman, 1971), becoming a strategic communication and persuasion
mitigating the possible risks derived from the recommendation of an tool that raises awareness of a product and makes it more acceptable and
influencer. From this point, a question guides the research: Are the in desirable for the audience (Enke and Borchers, 2019). The use of
terconnections between observational learning, product involvement influencers is linked to the informal communication process that sparks
and perceived risk determinants for the effectiveness of influencer rec the interest of the potential client (Schwemmer and Ziewiecki, 2018). As
ommendations (attitudes and purchase intentions)? such, it is aligned with the new paradigm in which users of social net
The present study seeks to fill this gap in the research and aims to works act as independent brand ambassadors (Boerman, 2020). How
examine the impact of perceived risk as an antecedent of the purchase ever, the emerging literature on influencer marketing has yet to provide
attitudes and intentions influencers encourage in their followers. To that a comprehensive understanding of the more negative aspects that could
end, we propose a new exploratory theoretical model based on a com influence purchase intention and behaviour (Enke and Borchers, 2019;
bination of psychosocial theories of behaviour: observational learning Hudders et al., 2021), particularly those related to the risk perceived by
(OL) (Bandura, 1977; Bikhchandani et al., 1998) and an extension of the the audience concerning the recommendations received. The issue of
theory of planned behaviour (ETPB) (Ajzen, 1991; Meng and Choi, risk, which has been widely explored in studies on product acquisition
2016). Furthermore, several risk mitigating factors are explored in and purchasing behaviour (Cox and Rich, 1964; Forsythe and Shi, 2003;
depth: on the one hand, the social interactions of the observation process Taylor, 1974), has received little attention in this area. Risk refers to the
(Kim and Yang, 2017); and on the other hand, the follower's involve uncertainty felt by the consumer when he/she considers a purchase but
ment with the type of product supported (Mou et al., 2020) as well as the believes it may not live up to expectations (Biswas et al., 2006; Kim
perceived authenticity of the message conveyed by the influencer et al., 2008). Nevertheless, in recent years some studies have addressed
(Hudders et al., 2021; Martínez-López et al., 2020). The resulting hy this subject; the audience usually has a degree of awareness about the
potheses are tested using a sample of 948 followers of influencers on financial sponsorship that the influencer receives when making pur
social networks. The context is Spain, which is characterized by high chase suggestions (Martínez-López et al., 2020; Stubb et al., 2019),
rates of internet adoption (91 %) and use of social networks (80 % of the which could generate doubts about the credibility and impartiality of
population) (WeAreSocial and Hootsuite, 2021). Although the applica the message, leading to scepticism and a possible perception of risk
tion of the model to any consumer product can be theoretically justified, (Boerman et al., 2017). This issue may have been overlooked as most
we focus on recommendations of fashion products, as this is the most studies are based on an absence of risk and assume that recommenda
representative category in influencer marketing (InfluencerMarketing tions are always perceived as reliable—probably because such studies
Hub, 2021a), and one of the most extensively researched in this domain work under the assumption that the audience holds the influencer to be
2
someone who is relatable, reliable and highly knowledgeable about the the behaviours and decisions of others. Zafar et al. (2021) point to social
product (Casalo et al., 2020). Although some authors such as Chopra interaction as an essential element for understanding behaviour through
et al. (2020) conclude that risk is of minimal importance in a study of OL, and indicate that it contains discrete signals expressed by the actions
followers of influencers, other authors, such as Deshbhag and Mohan of others. Most studies of influencer marketing apply an audience-
(2020), report contrasting results, indicating that the audience perceives focused approach and examine the responses to content suggested by
a risk and this has an impact on their purchase intentions. opinion leaders (celebrities/micro-celebrities), the reactions of their
Hudders et al. (2021) and Borchers (2019) highlight a series of followers, and how the audience attributes meanings that play a
business valuation indicators to opt for influencer marketing in product prominent role in evaluations of the suggestions received (Hudders
promotion, for example, the credibility and experience of the influencer, et al., 2021). An influencer's follower base reflects their degree of
the congruence with the brand and the potential customer, as well as a commitment to sponsored content through their reactions on social
set of quantitative metrics such as the number of followers, the amount networks (Bashari and Fazl-Ersi, 2020; Francalanci and Hussain, 2017).
of shared content they generate, the comments or the number of “likes” Recent research on social media sales reveals that there are particular
that those contents reach (Borah and Xiao, 2018). These metrics are elements on the different networks that affect the final purchase inten
linked to the degree of audience engagement (Zafar et al., 2021). tion through mechanisms such as recommending or sharing content,
Quantitative metrics are an inherently positive element when the in including ratings or “likes”, making comments, posting reviews and
teractions generated between influencer and audience, or followers with recommendations, using emojis of agreement and disagreement (Das
each other, are positive and encourage engagement with the endorsed et al., 2019), or any other version thereof that generates valuable in
product or service (Borah and Xiao, 2018; Zafar et al., 2021). In essence, formation and influences the audience and their purchase decision
comments, subscriptions and “likes” of posts promote two-way (Hajli, 2015).
communication, and support interactions (Zafar et al., 2021). Conse In the context of social media sales, e-Word-of-Mouth (eWOM) and
quently, these manifestations contribute to increasing relational trust OL are compatible concepts (Wang and Yu, 2017), as they are forms of
and foster credibility of sources (Borah and Xiao, 2018; De Vries, 2019; communication that yield information in different communities; that is,
Hilverda et al., 2018; Seo et al., 2019). As noted by Tolson (2010), the the content generated by the audience of the influencer transmits in
ability of the audience to participate in the creation and dissemination of formation about the product or service that is recommended (Kon
content posted by an influencer supports the authenticity of the mes stantopoulou et al., 2019; Wang and Yu, 2017). Consumers tend to be
sage. To this extent, observational learning theory offers potentially affected by social interactions; Chen et al. (2011) claim that actions on
useful information (Kim and Yang, 2017; Zafar et al., 2021), in terms of social networks are perceived as credible and their quantitative in
its ability to explain how influencers' followers acquire knowledge from dicators are taken into account as part of the observation and learning
the cues generated in the social media communication process, and this process (Kim and Yang, 2017). In the same vein, Zafar et al. (2021) argue
affects their final purchase intention. Therefore, observational learning that interactive numerical signals on social media—the number of
(OL) can impact on the authenticity of the message conveyed (Borah and “likes”, “shares”, and “comments”—comprise a simplified, heuristic
Xiao, 2018; Zafar et al., 2021). In particular, recent research on influ approach integrated into decision-making, and form part of this
encer marketing uses observational learning (OL) through metrics to learning. We thus propose the first research hypothesis:
improve the understanding of credibility, and the purchase process
H1 (a, b, c). Quantitative indicators of interactions (the number of
prompted by online celebrities (Zafar et al., 2021), although the
“likes,” “shares” and “comments”) positively influence the perceived
simultaneous impact it may have on mitigating potential perceived risks
authenticity of the message conveyed by the influencer.
to recommendations has so far not been considered. Perceived risk is
precisely an unknown in influencer research. It is not known whether an Moreover, these quantitative metrics are frequently used by mar
influencer's followers trust their recommendations as an opinion leader keting specialists (Childers et al., 2019). In the field of research on social
about the product in question and it is an irrelevant construct, or media influencers, it has been shown that e-WOM is linked with pur
whether the danger associated with purchasing the product is reduced chase behaviour through the perception of source credibility and the
by other factors, such as reactions generated by other users. Addition reduction in the perceived risk, since the followers of influencers refer to
ally, the involvement with the recommended product in particular could reviews on platforms like Instagram to reduce the potential hazards
be a factor implicated in the recommendations made by influencers detected in the recommendations (Djafarova and Rushworth, 2017). In
(Mou et al., 2020; Zaichkowsky, 1985). In this vein, the three constructs the same vein, Martensen et al. (2018) point to “likes” as a form of social
mentioned: observational learning (OL), perceived risk and product proof or conformity that offers security and legitimacy when it comes to
involvement could be integrated from one of the most consolidated making fashion purchases suggested by influencers. Based on the above,
theoretical frameworks in the analysis of purchase behaviour and we propose the second hypothesis:
intention, the theory of planned behaviour (Ajzen, 1991; Sánchez-
H2 (a, b, c). Quantitative indicators of interactions (number of “likes,”
Cañizares et al., 2021). Based on this approach, a new exploratory model
“shares” and “comments”) reduce the perception of general risk about
could be built to offer new theoretical insights to the growing research
the recommendations made by the influencer.
field dedicated to influencer marketing, as well as to make recommen
dations to marketers on how to develop more effective influencer
collaboration campaigns. 2.3. The perceived authenticity of the influencer's message as a means of
mitigating the perceived risk; an antecedent of purchase attitude and
2.2. Observational learning as an antecedent of the perceived authenticity intention
of the message conveyed by the influencer and the risk perceived by the
audience The credibility of the source and the promotional message are
noteworthy constructs in the emerging research on influencers (Hudders
Observational learning (OL) is a significant construct that can be et al., 2021; Martínez-López et al., 2020). Kowalczyk and Pounders
found in research on purchasing processes (Zafar et al., 2021). Its origins (2016) define the authenticity of a celebrity as the presentation of
lie in the theory of observational learning (Bandura, 1977; Bikhchan genuine aspects of the self through their posts. In this regard, Zafar et al.
dani et al., 1998), according to which people gather information from (2021) point out that the authenticity of an influencer lies in being
others when they take on new tasks. This enables them to reduce the sincere and truthful with his or her follower base. Recent literature notes
mental effort in decision-making. According to Wang and Yu (2017), it that so-called online celebrities are perceived as more credible (Nouri,
can be manifested through the social interaction derived from observing 2018). In addition, the endorsements they provide are taken as truthful
3
and influential (Djafarova and Rushworth, 2017). In this study, we of risk that followers of an influencer may experience in response to
define the perceived authenticity of the message as the feeling among an recommendations in a non-restrictive sense, that is, we define the
influencer's followers that a recommendation is genuine and sincere, not perception of risk as the expected negative utility associated with the
false (Gilmore and Pine, 2007). Furthermore, we assume that the purchase of a specific product (Tsiros and Heilman, 2005). Therefore, as
perceived authenticity of the message varies depending on personal and this study is an initial approach to risk in the field of influencers' rec
parasocial interactions, persuasive appeal and feelings of intimacy ommendations, we address the general risk (Barcelos et al., 2018). In
(Sokolova and Kefi, 2020). In the field of influencer marketing, the in this regard, recent literature has reported that influencers come from
fluence of the authenticity of the message on purchase attitude and different domains: sport, lifestyle, fashion, entertainment, food, and
intention is reflected in positive impacts on both. A prominent core of others (Hudders et al., 2021). Their recommendations are perceived as
the research has focused on recommendations of fashion products; for genuine partly because they come from a particular field in which they
example, Monsalve (2017) show how the opinion leadership exerted by are assumed to be highly knowledgeable and credible (Casalo et al.,
influencers through their messages is transmitted to the final purchase 2020). However, as far as we know, in the novel research field focused
intention. Woobin and Jung (2019) reveal that honesty and experience on influencer marketing there has been very little empirical analysis of
are more important than social or physical attractiveness. Chetioui et al. the risk stemming from influencer endorsements, with it almost
(2020) claim that perceived credibility is the most critical factor, fol invariably being assumed that no harm can come from following their
lowed by experience and trust. Based on all this, we propose the recommendations. For example, Chopra et al. (2020) conducted a
following hypotheses: qualitative study with data collected from in-depth interviews with the
target audience and marketing experts, concluding that this construct
H3. The perceived authenticity of the sponsored message positively
was of little importance and that on discovering untrue statements the
affects the attitude towards the influencer's recommendations.
audience would simply stop following influencer, meaning the impact
H4. The perceived authenticity of the message exerts positive effects would be limited. Along the same lines, Silva et al. (2020), in a quan
on the purchase intention promoted by influencers. titative analysis of 261 surveyed social network users aged over 16,
reported that these users did not experience risk when taking account of
In contrast to the research on the credibility of the message
suggestions beyond the danger involved in purchasing any product with
conveyed, there is still relatively little research on perceived risk in the
or without such a recommendation. Conversely, studies such as that by
field of influencer marketing. Previous studies have highlighted how, as
Biswas et al. (2006) do assign some importance to the construct,
product-related risks increase, the consumer's desire for information
reporting differential effects with respect to the perceived risk according
increases as a step prior to purchasing (Hilverda and Kuttschreuter,
to whether the product endorsement comes from a celebrity or an expert
2018; Kamalul Ariffin et al., 2018). Consequently, consumers turn to
in a particular area. Expert endorsers are found to have a stronger effect
influential sources that they see as relevant in order to minimize the risk.
than celebrity endorsers in terms of reducing consumer risk perceptions,
It therefore seems likely that the endorsement of a product by an
depending mainly on the type of product and the buyers' level of
influential personality in a social environment, who is understood to
knowledge about the product. In any case, the audience usually has a
have expertise in a specific field (e.g. video games, sport, fashion), re
certain degree of awareness about the financial sponsorship that the
duces the perception of risk; this does not imply that the risk is totally
influencer receives when making suggestions (Martínez-López et al.,
eliminated, but it does represent an additional way of trying to
2020; Stubb et al., 2019), which could generate doubts about the
neutralize it (Biswas et al., 2006; Dean and Biswas, 2001). In this regard,
credibility and impartiality of the message, leading to scepticism and
Dean and Biswas (2001) highlight how endorsements can signal the
possible perception of risk (Boerman et al., 2017). Furthermore,
reliability of a product and inform the consumer. This is in line with
Deshbhag and Mohan (2020) recently showed how perceptions of risk
signalling theory (Boulding and Kirmani, 1993), according to which
influenced attitudes and intentions to purchase consumer goods. Based
brands can reduce consumer uncertainty and risk perception by sending
on the theory of perceived risk and the above arguments, we propose the
pre-purchase signals. The authors' note, however, that to be considered
following hypotheses:
credible, endorsements should contain a binding component and entail a
potential cost to the provider of the endorsement if the signal is false and H6. The perception of risk in influencers' suggestions has a negative
the product does not meet expectations. When an influencer promotes a impact on purchase attitudes.
product he/she sends a signal, accepting a risk of damage to his/her
H7. The perception of risk in influencer's suggestions has a negative
reputation if the message conveyed about the product does not turn out
impact on purchase intention.
to reflect reality. As a result, the message tends to be perceived as
credible and will diminish the perception of risk. In light of the above,
2.5. Effect of product involvement on the perceived authenticity of the
we propose the following hypothesis:
message, and as an antecedent of purchase attitude and intention
H5. The perceived authenticity of the sponsored message reduces the
perceived risk of influencers' recommendations. In the 1980s, Zaichkowsky (1985) developed a scale to capture the
concept of involvement with a product, defining it as a person's
2.4. Effects of the perceived risk of influencer recommendations on perception of the relevance of specific product, based on his/her
purchase attitude and intention inherent needs, values and interests. Mou et al. (2020) point out that the
theory of involvement was initially proposed by Sherif and Cantril
Risk theory has often been used in studies of purchasing behaviour (1947) to indicate that the involvement of a person's ego in a given
(Cox and Rich, 1964; Forsythe and Shi, 2003; Taylor, 1974). Perceived situation raises the interest in the stimulus. Since then, the term has been
risk is defined as the possibility of loss anticipated by consumers in used in the literature with various definitions and applications (Fried
connection with the purchase of a product or service; it refers to the mann and Lowengart, 2019). In this respect, involvement with a product
degree and amount of uncertainty that consumers perceive when they has been viewed as a general attitude reflecting its importance to the
are considering a purchase and believe that it may not turn out to be as consumer, a function of the frequency of purchase, social risk, financial
expected (Biswas et al., 2006; Kim et al., 2008). Forsythe and Shi (2003), risk and the cost entailed (Nkwocha et al., 2005). The most widely-
in a pioneering study about the risks stemming from online purchases, accepted definition is probably that put forward by Zaichkowsky
identified six components of the risk related to purchases in general: (1985), in which involvement constitutes a motivational construct that
financial loss, product performance or quality, psychological, time and/ depends in part on consumers' values and personal needs (Mou et al.,
or convenience, social and physical. This paper addresses the perception 2020). Moreover, it is considered a key theoretical construct that can
4
yield a better understanding of the main concepts used to explain con incorporated comply with three principles: they must be imperative
sumers' attitudes, intention and behaviour (Xu et al., 2004). Different factors that affect the decision-making process, they must be concep
studies have tested this theory by conducting empirical analyses on a tually independent of the original, and they must be potentially appro
wide variety of products: alligator leather accessories (Xu et al., 2004), priate for gaining an understanding of a specific behaviour (Ajzen, 1991;
hotels, wines and ecological cars (Rahman, 2018), luxury cruises (Han Meng and Choi, 2016). The extended model has been widely used,
and Hyun, 2018) and smartphones (Faisal et al., 2020) among others. including constructs that are critical in different contexts (Sánchez-
In this paper, we take the definition proposed by Zaichkowsky Cañizares et al., 2021) including influencer marketing (Chetioui et al.,
(1985), and we follow the suggestions made by Xu et al. (2004), who 2020; Chopra et al., 2020).
indicate that people's involvement with a certain product can be Taking the above into account, we formulate the last hypotheses and
measured through variables—antecedent to intention and attitu present the proposed model in Fig. 1:
de—related to buyers' degree of commitment to the specific product; in
H11. Perceived control over purchasing products endorsed by influ
their case, fashion involvement (Shim et al., 1989). In this study we have
encers positively affects the intention to purchase recommended
chosen to use fashion consciousness, defined as a person's level of in
products.
terest in displaying a fashionable appearance through the use of clothes
and accessories (Gutman and Mills, 1982; Lertwannawit and Man H12. Attitude towards influencers' recommendations has a positive
dhachitara, 2012; Nam et al., 2007). Influencers can thus be seen as impact on the intention to purchase recommended products.
leaders who post messages with recommendations on the latest trends,
H13. The subjective norm has a positive impact on the intention to
expressing their individuality, while making the audience aware of
purchase products recommended by influencers.
current styles they can adopt to maintain their individual status in the
social network in which they are participating (Lertwannawit and
3. Research method
Mandhachitara, 2012).
Product involvement has been proven to play a decisive role in
The research, which centres on an exploratory descriptive analysis, is
consumer behaviour (Mou et al., 2020). Previous research focusing on
performed using non-probability sampling techniques; namely, conve
different areas, contexts and products has shown that involvement
nience sampling and snowball sampling. These two procedures allow us
constitutes an antecedent of behaviour influencing both attitudes and
to harness the power of social media to easily and cheaply reach a broad
behavioural intentions (Mou et al., 2020; Xu et al., 2004). Furthermore,
set of eligible units. Non-probability sampling techniques are particu
degree of involvement has been shown to be positively correlated with
larly suitable for an initial study of a given phenomenon, since they
cognitive participation, (Zhu et al., 2019). This implies that in the
contribute to a better understanding of it while new research questions
emotional stage on a platform, for example, when people are looking at
are being developed and the methods applied are being refined (Li and
an influencers' recommendation, if their understanding of the product or
Walejko, 2008). A questionnaire was designed in line with the different
their interests align with the recommendations, it could increase the
approaches identified in the literature described above. It addresses the
perception of authenticity of the message conveyed and increase their
hypotheses generated from the model in Fig. 1 and covers the aim of this
commitment to the possible purchase. However, involvement remains a
study. The survey was distributed online, which by default excludes
largely unexplored construct within the field of influencer marketing.
people without an internet connection and who would thus find it more
An exception is the study by Parmar et al. (2020), whose results reveal
difficult to follow recommendations made by influencers through social
that the presence of a celebrity boosts the purchase impulse, and its
media.
effectiveness increases in the case of products with which the consumer
feels a strong involvement, as may well be the case with fashion. Based
on this, we propose the following hypotheses: 3.1. Methodological justification
H8. Product involvement through fashion consciousness positively
As far as the data collection technique is concerned, bias is the main
influences the perceived authenticity of the influencer's message.
criticism made when data are collected non-randomly with convenience
H9. Product involvement through fashion consciousness exerts posi samples. In this sense, the representation of the sample can be ques
tive effects on attitudes towards products recommended by influencers. tioned (this is included later as a limitation of this work). Precisely,
probability sampling aims to mitigate the possible effects of unbalanced
H10. Product involvement through fashion consciousness has a posi
covariates through random selection. However, in our analysis we opted
tive impact on the intention to purchase products suggested by
for convenience sampling as the target population (followers of influ
influencers.
encers) is not well known, besides considerations such as those of
Borchers (2019): “SMIs (Social Media Influencers) have a wide reach in
2.6. Extended theory of planned behaviour attractive and contested publics, in particular teenagers and young adults,
and acceptance rates for SMIs' promotional activities are high”. Moreover,
The theory of planned behaviour (TPB) is an extension of the theory Schreuder et al. (2001) pointed out that non-probability samples can be
of reasoned action (TRA) (Sánchez-Cañizares et al., 2021). The original very useful for describing specific subgroups. Based on the above, the
model holds that a person's intention to perform any behaviour is recommendations of Lehdonvirta et al. (2021) on the use of online
determined by their attitude, that is, by the individual's personal surveys for exploratory research on emerging topics, new phenomena
conviction and feelings towards the behaviour; and by the subjective such as influencers, were noted, as they provide early research evidence,
norm, or the individual's perception of what his/her salient referents help to probe the characteristics of new relevant subpopulations and can
would think about the performance of the behaviour in question. reach genuine members of the population.
However, Ajzen (1991) found that behaviour also depends on volitional As for the proposed methodology, the pioneering research of Duncan
control, or on the individual's ability to perform the behaviour. The (1966) is positioned as a precursor of structural equation modelling, by
inclusion of the latter construct gave rise to the current TPB, which in seeking alternatives to experimental manipulation through the analysis
dicates that the intention to perform a behaviour is determined by of paths or dependencies, in other words, it proposed the analysis of
attitude, the subjective norm and perceived behavioural control. causality through the specification of a set of significant variables in a
Nevertheless, Ajzen (1991) also indicated that his model is open to ex theory, deriving the relationships between the effects of the theory, in
tensions through the inclusion of additional constructs that can help to order to finally estimate the size of these effects There are currently two
capture a larger proportion of the variance, as long as the new variables approaches to structural equation modelling, one based on covariance
5
Fig. 1. Research model.*

*In order to facilitate the monitoring of the proposed research model, the constructs that configure observational learning (likes, shares & comments) have been
grouped together. The items integrated as antecedents in the theory of planned behaviour have also been marked in blue. The order of the hypotheses described in
this figure follows the sequence described in the Literature review and hypothesis development. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
(CB-SEM) and the other based on partial least squares (PLS-SEM) (Hair (fashion consciousness); PBC = Perceived behavioural control; SN =
et al., 2021). The most salient aspect of each lies in the differences in the Subjective Norm. The specification of each of these constructs is
recommended use of each of them. In essence, the choice of approach addressed by means of a reflective measurement model based on the
depends on the objective of the research: while the first approach is considerations made by Coltman et al. (2008): the indicators of each
indicated for the verification and testing of theories, the second construct can be interchangeable and each specific item could be
approach is recommended for exploring or predicting unestablished omitted without changing the meaning of the construct as long as it has
theories or relationships (Hair et al., 2017). Based on the aforemen sufficient reliability.
tioned, the analysis of the responses obtained was carried out by means
of partial least squares based structural equation modelling (PLS-SEM).
3.2. Data collection and sample design
As Chin (1998) indicated, this technique is suitable for evaluating
complex models such as the one described above (see Fig. 1). “Concep
The target population for this study is Spanish people aged 18 and
tually and practically, PLS-SEM is similar to using multiple regression anal
over, who are followers of an influencer in the field of fashion, and who
ysis. The primary objective is to maximize explained variance in the
have made a purchase based on an influencer's recommendations. A
dependent constructs but additionally to evaluate the data quality based on
total of 1224 responses were collected, although 57 incomplete ques
measurement model characteristics” (Hair et al., 2017). Moreover, it is a
tionnaires were excluded along with 219 respondents who answered
particularly useful empirical research tool when there is no coherent
“no” to the questions “Do you follow any Spanish influencers in the
explanation, or new constructs included in models developed from
fashion sector?” and “In the last year have you bought any fashion
established theories have not been sufficiently explored (Dash and Paul,
product based on the recommendations of an influencer?” Other than
2021). This methodological development has demonstrated its potential
those just noted, no additional restrictions were introduced in the se
in the field under study, more specifically when testing knowledge
lection criteria in order to ensure a high degree of randomness. The final
related to consumer feelings and perceptions (Jibril et al., 2019). In its
sample was made up of 948 participants.
application, the indications developed by Hair et al. (2021) were fol
The data collection procedure entailed several steps. First, a draft of
lowed, and its specific use for the field of social sciences (Dash and Paul,
the questionnaire was written, with reference to the previous literature
2021).
on the constructs used. This draft was then reviewed by a group of five
The framework of the study was configured from 10 constructs
researchers in the area of business and marketing to ensure the validity
justified in the previous review: L = Likes; S = Shares; C = Comments
and reliability of the measures and the logical order of the questions.
(Observational Learning (OL)); PMA = Perceived Message Authenticity;
Next, a pilot test was carried out using a convenience sampling pro
PR = Perceived Risk; A = Attitude; Int = Intention; Inv = Involvement
cedure, with an initial link to the questionnaire distributed via social
6
Table 1
Sociodemographic profile of the sample.
Variable Categories Male (N = 314) Female (N = 634) Total (%)
Age (years) 18–25 243 447 690 (72.8 %)

26–30 26 43 69 (7.3 %)
31–40 14 33 47 (5.0 %)
41–50 12 49 61 (6.4 %)
>50 19 62 81 (8.5 %)
Education Primary level or lower 2 3 5 (0.5 %)
Middle school 16 42 58 (6.1 %)
High school 131 234 365 (38.5 %)
University 137 306 443 (46.7 %)
Master's 23 43 66 (7.0 %)
PhD 5 6 11 (1.2 %)
Employment status Business owner/self-employed 22 23 45 (4.7 %)
Paid employee 44 87 131 (13.8 %)
Civil servant 16 43 59 (6.2 %)
Student 221 424 645 (68.0 %)
Homemaker 1 21 22 (2.4 %)
Retired 5 7 12 (1.3 %)
Currently unemployed 5 29 34 (3.6 %)
Household disposable income (net euros/month) <€1000 64 139 203 (21.4 %)
€1001–2000 97 214 311 (32.8 %)
€2001–3000 76 146 222 (23.4 %)
€3001–4000 42 74 116 (12.2 %)
>€4001 35 61 96 (10.2 %)
Frequency of use Every day 267 539 806 (85.0 %)
Several times a week 30 47 77 (8.1 %)
Once a week 4 9 13 (1.4 %)
Several times a month 6 7 13(1.4 %)
Once a month 0 4 4 (0.4 %)
Several times a year 7 28 35 (3.7 %)
Way of shopping Only traditional 78 98 176 (18.6 %)
Only online 20 28 48 (5.1 %)
50 %–50 % 77 183 260 (27.4 %)
Both (+online) 55 115 170 (17.9 %)
Both (+traditional) 84 210 294 (31.0 %)
networks (n = 50). On the basis of the results, some questions were re- questionnaire they were giving their informed consent. The entire pro
worded. The data collected using the first draft of the questionnaire were cess and the final questionnaire were developed following consultation
not included in the analysis to avoid biases in the results. with the ethics committee of the authors' university, and the recom
The final collection of the sample was conducted through the Sur mendations and guidelines of the Declaration of Helsinki were also
veyMonkey platform. A link and a QR code were generated and used to observed. Lastly, it should be noted that although this is a convenience
collect responses. Social networks and the Moodle learning management sample, the sociodemographic data obtained are similar to the audience
tool were used to distribute the link or the code in different universities profiles for the top 10 Spanish fashion influencers sourced from two of
in southern Spain. The questionnaire remained open during the months the most popular platforms for influencer analytics, which provide in
of April and May 2021 (once the strictest mobility restrictions imposed formation on their reach, demographic characteristics and audience
in response to the Covid-19 pandemic had been relaxed). Potential (HypeAuditor.com, thesocialflame.com). For example, in December
participants viewed the link on social networks or received invitation 2021 the two most representative Spanish fashion influencers on
emails through Moodle, with messages from the researchers using Instagram (InfluencerMarketingHub, 2021b) were María Pombo
snowball sampling techniques to encourage respondents to distribute (@mariapombo) with 2.2 M followers and Aida Domenech (@dulceida)
the message through social networks and among their circle of close with 2.3 M. They have a similar audience profile, indicating that their
contacts (Li and Walejko, 2008). In both the questionnaire and the recommendations primarily influence young (24 %/29 % aged between
messages, participants were told approximately how long it would take 14 and 20; 47 %/49 % between 21 and 35), female (79 %/83 %), regular
(8–12 min), and were informed of the objectives of the study and their internet users (TheSocialFlame, 2021).
rights regarding voluntary participation, guarantees of anonymity and Table 1 shows the profile of the final sample, where women can be
confidentiality. They were also told that by clicking on the link to the seen to predominate (two thirds, compared to one third men), mostly
aged between 18 and 30 (80.1 %), with a high educational level. Stu
Table 2 dents and paid employees are the most represented groups. The level of
Sample profile on social networks. household disposable income varies widely, although more than half of
the respondents stated that they did not earn more than €2000 net per
Mean social network Mean number of influencers followed on
usage the network month. With regard to the use of social networks, both men and women
reported that they use social networks on a daily basis. A similar profile
Male Female Male Female
is observed with respect to the preferred way of shopping, with only 5.1
Instagram 5.60 5.73 4.51 4.95* % of people stating that they only buy online. The sample leans towards
Youtube 5.58 4.44* 4.28 3.27*
Twitter 3.65 3.00* 2.84 2.20*
traditional shopping in a physical establishment (18.6 %), or both but
Facebook 3.12 3.56* 2.15 1.99 with more of an emphasis on traditional shopping (31 %).
TikTok 3.05 3.33 2.73 2.96 Regarding the most used networks and the number of influencers
Twitch 2.60 1.34* 2.68 1.33* that respondents follow on each, Table 2 shows some notable differences
Linkedin 1.82 1.56* 1.43 1.21*
between men and women. The measurement scales went from 1- I don't
*
p < 0.05.
7
Table 3
Measurement model. Factor loadings.
Constructs Items Factor Mean / (Standard Adapted from
loading Deviation)
Likes (L) L1. I look to see whether the influencers' comments/posts/videos/ 0.838 3.110 (2.093) (Zafar et al., 2021)
photos get a lot of likes
L2. I compare the number of likes for the comments/posts/videos/ 0.893 2.292 (1.807)
photos published by influencers
L3. I check which fashion products get the most likes 0.883 2.488 (1.883)
Shared content (S) S1. I take notice of the most shared comments/posts/videos/photos 0.889 2.576 (1.901)
on the influencer's channel or social network
S2. I take notice of the number of times an influencer's comment/ 0.903 2.216 (1.715)
post/video/photo has been shared
S3. I take notice of the number of reactions to a fashion product 0.895 2.506 (1.861)
presented by an influencer
Comments generated (C) C1. I think it's worth checking out the comments that other people 0.812 4.025 (2.268)
make about the fashion product presented/recommended by the
influencer
C2. I take notice of the number of comments made by other people 0.908 3.224 (2.120)
about the influencer's opinion of the fashion product
C3. I take notice of which products recommended by the influencer 0.861 2.949 (2.013)
have generated the most comments from other users.
Perceived risk (PR) PR1. It's risky to buy products recommended/promoted by 0.884 3.991 (1.774)
influencers
PR2. Buying products recommended/promoted by influencers adds 0.850 4.055 (1.834) (Alalwan et al., 2018; Singh et al., 2021)
to the uncertainty about the results I will get from buying the product
PR3. Using influencers' recommendations exposes me to a general 0.827 3.828 (1.835)
risk about the results of the product
Perceived authenticity of PMA1.I perceive influencers' fashion suggestions as authentic 0.874 3.192 (1.690) (Martínez-López et al., 2020)
messages (PMA) PMA2. Fashion influencers' online posts seem genuine to me 0.908 3.252 (1.707)
PMA3. Influencers' opinions about fashion are reliable 0.878 3.200 (1.677)
Involvement (fashion Inv1. I usually have one or more items of clothing that are currently 0.747 4.563 (2.145) (Lertwannawit and Mandhachitara, 2012;
consciousness) (Inv) on-trend Xu et al., 2004; Zaichkowsky, 1985)
Inv2. When I choose between two items of clothing, I base the 0.779 3.436 (2.002)
decision more on what's in fashion than on comfort.
Inv3. Wearing the latest fashions is important to me 0.878 3.469 (2.018)
Inv4. It is important to me that my clothes are as current as possible 0.844 3.409 (1.998)
Attitude (A) A1. I like to buy fashion items recommended by influencers 0.867 2.896 (1.810) (Ajzen, 1991; Hwang and Kim, 2020; Kim
A2. It is worth buying fashion items endorsed by influencers 0.888 2.996 (1.677) and Karpova, 2009)
A3. It's good to buy fashion items used by influencers 0.808 2.584 (1.692)
Perceived control (PBC) PBC1. I have the necessary knowledge to buy most of the fashion 0.867 3.273 (1.954)
items recommended by influencers
PBC2. If I wanted to, I could buy the fashion items recommended by 0.669 4.131 (2.049)
influencers
PBC3. I am sufficiently social media savvy to buy fashion items 0.816 4.015 (2.133)
recommended by influencers
Subjective norm (SN) SN1. I believe that I let myself be guided by influencers when it comes 0.657 3.093 (1.905)
to making my fashion-related purchases
SN2. The people important to me (friends/family) think I should buy 0.785 2.245 (1.612)
fashion items recommended by influencers
SN3. People around me buy fashion items recommended by 0.722 3.352 (1.813)
influencers
SN4. I feel social pressure to buy fashion items recommended by 0.731 2.009 (1.565)
influencers
Purchase intention (Int) Int1. I intend to buy fashion products recommended by influencers 0.897 2.545 (1.770)
Int2. In the future I will try to buy products endorsed by influencers 0.928 2.503 (1.707)
Int3. I will make an effort to buy fashion products recommended by 0.894 2.129 (1.611)
influencers
The text already explains that the main diagonal (in bold) presents "the square root of the variance shared between each construct and its measures".
use it at all/I don't follow any influencers here to 7- I use it a lot/I follow 3.3. Measurement instrument and scales
7 or more influencers. As can be seen, Instagram is the most used
network for both genders. Men use Youtube, Twitter, Twitch and Link To carry out the cross-sectional analysis, the questionnaire was
edIn more than women do, and the differences are statistically signifi divided into six sections: sociodemographic characteristics (including
cant according to the difference of means T-student test with 95 % online shopping profile, social networks and influencers followed); OL
confidence level. Conversely, women use Facebook and TikTok more (including “likes”, “shares” and “comments”); perceived authenticity of
often, although the difference in the mean between the two groups is not the influencer's message; perceived risk in the suggestions received;
significant for TikTok. Regarding the number of influencers followed, product involvement; and planning of behaviour (including intention
Instagram and TikTok are the only social media where women follow and the three TPB constructs that affect it). The variables used in the
more influencers, although in the latter case, the difference is not sta model were operationalized by adapting measures developed in previ
tistically significant. ous studies on purchasing behaviour or the use of social networks
(Table 3). The latent variables were measured on a seven-point Likert
scale (1-Totally disagree; 7-Totally agree) (Table 3).
8
4. Results Table 5
Structural model. R2 and Q2.
4.1. Common method bias and multicollinearity test Variable R2 Q2
A 0.459 0.332
A full collinearity test was performed to determine whether the re Int 0.591 0.479
sults were affected by common method bias (CMB) (Kock and Lynn, PMA 0.334 0.260
2012), which would adversely affect their validity. None of the variance PR 0.041 0.028
inflation factors (VIF) for the latent variables in the model exceeded the
threshold of 5, a value proposed as indicating the existence of multi
model all values of the HTMT ratio are below the value 0.90 suggested
collinearity and CMB (Kock, 2015). Therefore, it can be concluded that
by Gold et al. (2001), a threshold considered acceptable when the latent
the empirical analysis is free of these possible biases.
variables measure similar constructs, implying that the model is well
adjusted (heterotrait correlations are lower than monotrait correla
4.2. Measurement model tions). In addition, a full bootstrapping procedure is carried out, as
suggested by Hair et al. (2021) to contrast the null hypothesis HTMT ≥
The proposed theoretical model was estimated using structural 1. The confidence intervals obtained do not contain values equal to or
equation modelling (SEM) with partial least squares (PLS). The software >1, so discriminant validity is achieved. In short, the 10 constructs of
SmartPLS 3.0 was used to analyse the relationships proposed in the the proposed model were found to be statistically different and can be
hypotheses. The measurement model with all constructs estimated used to estimate the structural model, since the measurement model met
reflectively was validated by assessing the reliability of the individual the criteria of internal consistency, reliability and convergent and
indicators through their loadings (Table 3), as well as the internal con discriminant validity.
sistency and convergent and discriminant validity of the constructs
(Table 4).
The outer loadings exceeded the cut-off value of 0.707 suggested by 4.3. Structural model
Carmines and Zeller (1979), with the exception of two items (one on
perceived control and the other on the subjective norm). However, since Following the analysis and validation of the measurement model, the
they were in different constructs and were well above the value of 0.4 next step was the estimation of the structural model to examine the
suggested by Henseler et al. (2009) as the cut-off for the removal of an standardized path coefficients and to test whether the proposed hy
indicator, they were left in the model. This can help in extracting useful potheses are supported.
information about the indicator and generating a better score for the According to Table 5, the R2 values for the endogenous variables in
latent variable, as well as for the rest of the measurement indicators that the structural model lie between 30 and 40 %, except in the case of PR,
verify the discriminant validity (Table 3). Moreover, a bootstrapping which barely reaches 4.1 %. R2 values of 0.67, 0.33, and 0.19 indicate
with 5000 iterations confirmed the statistical validity of the loadings substantial, moderate, and weak model, respectively (Chin, 1998).
through the Student-t-test, with all p-values being <0.01. Therefore, the model's strength in predicting attitude, purchase inten
Regarding internal consistency and convergent validity, Table 4 tion and perceived authenticity of the message is moderate. The latter is
presents the reliability and validity measures assessed through Cron smaller because there is only one significant explanatory variable for PR
bach's Alpha, the rho correlation coefficient, composite reliability and (Table 6). In any case, since this is an exploratory rather than a pre
average variance extracted (AVE) for each of the constructs in the dictive model, the percentage of variance explained does not have major
model. The first three measures exceeded the value of 0.7 suggested by implications in terms of validity.
Nunnally and Bernstein (1994) as an appropriate level for acceptable Additionally, Q2 (predictive relevance) is also reported. A blind
reliability. With regard to convergent validity, all the constructs sur folding procedure was applied to obtain Q2 values. According to Hair
passed the cut-off of 0.5 suggested by Fornell and Larcker (1981) for the et al. (2021) values greater than zero suggest that the exogenous con
AVE criterion; that is, each construct was shown to explain at least 50 % structs possess predictive relevance for the endogenous construct in
of the variance of the assigned indicators (Henseler et al., 2009). The question, which is fulfilled in all cases in the present model.
discriminant validity requirement was also fulfilled, as the square root of After calculating the R2 and Q2 values, significance of the path co
the variance shared between each construct and its measures (AVE), efficient was tested. Table 6 shows the results for each hypothesis ac
shown in the values on the main diagonal (in bold), exceeded the cor cording to the corresponding path coefficient βi, together with the
relations between each construct and any other (the rest of the matrix) respective p-value of the Student-t-test, obtained through bootstrapping,
(Hair et al., 2014). However, given that the Fornell-Larcker criterion which confirmed the statistical significance of each parameter.
may not be sensitive enough to detect possible problems in discriminant The estimated structural model shows how the path coefficients
validity, the Heterotrait - Monotrait (HTMT) ratio of correlations associated with the three OL hypotheses are statistically significant in
method was also applied (Henseler et al., 2016). In this measurement two cases (likes and comments) but not for shares. Therefore, the
Table 4
Internal consistency, convergent validity and discriminant validity of the measurement model.
Cronb. Alpha Rho_A (CR) AVE Fornell-Larcker Criterion
L S C PR PMA Inv A PBC SN Int
L 0.842 0.846 0.904 0.760 0.872

S 0.877 0.880 0.924 0.803 0.825 0.896
C 0.825 0.830 0.896 0.742 0.592 0.666 0.861
PR 0.818 0.858 0.890 0.729 0.013 0.046 0.107 0.854
PMA 0.864 0.864 0.917 0.787 0.466 0.442 0.428 − 0.108 0.887
Inv 0.829 0.836 0.887 0.662 0.310 0.287 0.277 0.001 0.435 0.814
A 0.815 0.823 0.890 0.731 0.541 0.556 0.490 − 0.110 0.645 0.465 0.855
PBC 0.709 0.783 0.830 0.622 0.276 0.305 0.379 0.016 0.383 0.233 0.457 0.789
SN 0.701 0.707 0.816 0.526 0.517 0.542 0.425 − 0.021 0.505 0.419 0.698 0.398 0.725
Int 0.892 0.894 0.933 0.822 0.551 0.569 0.460 − 0.102 0.557 0.397 0.733 0.388 0.663 0.907
9
Table 6
Structural model. Path coefficients and results for the hypotheses.
Hypothesis Independent variable Dependent variable Path coefficient (p-value) Result f2
L PMA 0.224 (0.000)*** Supported 0.024*
H1a
S PMA 0.054 (0.375) Not supported 0.001
H1b
C PMA 0.176 (0.000)*** Supported 0.025**
H1c
L PR − 0.043 (0.505) Not supported 0.001
H2a
S PR 0.040 (0.556) Not supported 0.000
H2b
C PR 0.186 (0.000)*** Not supported 0.019*
H2c
PMA A 0.539 (0.000)*** Supported 0.431***
H3
PMA Int 0.106 (0.003)*** Supported 0.015
H4
PMA PR − 0.185 (0.000)*** Supported 0.027**
H5
PR A − 0.052 (0.045)** Supported 0.005
H6
PR Int − 0.037 (0.094)* Supported 0.003
H7
Inv PMA 0.301 (0.000)*** Supported 0.121***
H8
Inv A 0.231 (0.000)*** Supported 0.080***
H9
Inv Int 0.022 (0.401) Not supported 0.001
H10
PBC Int 0.030 (0.184) Not supported 0.002
H11
A Int 0.441 (0.000)*** Supported 0.176***
H12
SN Int 0.279 (0.000)*** Supported 0.094***
H13
L = Likes; S = Shares; C = Comments; PMA = Perceived Message Authenticity; PR = Perceived Risk; A = Attitude; Int = Intention; Inv = Involvement (fashion
consciousness); PBC = Perceived behavioural control; SN = Subjective Norm.
***
p < 0.01.
**
p < 0.05.
*
p < 0.1.
hypothesis which posits that these indicators have a positive influence their intention to purchase it. This result is fairly logical in that if users
on PMA is supported in the case of H1a and H1c, but H1b—which relates find the fashion product endorsed by the influencer does not inspire a
the number of shares with the perceived authenticity of the content—is degree of security (for whatever reason, coupled with risk perception),
not supported. Respondents do associate the number of likes and com they will have an unfavourable attitude, and their purchase intention
ments that the influencer has received for a post/photo/video, etc., with will also be reduced.
this perception of authenticity; as such, the more information users The user's involvement with the product—in this study, measured
receive through reactions of this type, the more reliable they consider through their fashion consciousness—positively influences the
the content to be. perceived authenticity of the content and attitude towards the product,
On the other hand, the influence of OL on the perceived risk is sta meaning Hypotheses 8 and 9 are supported. Conversely, no direct
tistically significant only in the case of comments, but this relationship relationship is found between this involvement and the user's purchase
has a positive coefficient—the opposite sign to the expected one. intention, which means Hypothesis 10 should be rejected. However,
Consequently, H2a and H2b are not supported, indicating that likes and although no direct effect is found, there is a positive and significant
shared content are not related to the perception of risk about the fashion indirect effect (coefficient 0.209; p = 0.000) between this construct and
product that the influencer endorses or advertises. However, H2c is not purchase intention through the two positive effects on attitude and
supported either: more comments do not negatively influence the perceived authenticity of the content, which in turn have a direct and
perceived risk, as proposed. Rather the relationship is significant and significant influence on purchase intention.
positive, such that content that has generated more comments increases Lastly, Hypotheses 11, 12 and 13 refer to the relationships suggested
the user's perception of risk. This could be because comments about the by the TPB; H12 and H13 are supported by the positive and significant
product are negative or point out disadvantages associated with it. coefficients linking attitude and subjective norm to purchase intention.
Hypotheses 3, 4 and 5—on the influence of the perceived authen In contrast, H11 is not supported, since perceived control does not in this
ticity of the influencer's messages on attitude, purchase intention and case exert a notable influence on the intention of buying the fashion
perceived risk—are all supported. Therefore, if users perceive that the product advertised by the influencer. Attitude is the construct with the
content developed by the influencer is authentic and genuinely reflects highest coefficient and the one that has the most important effect on the
the quality and characteristics of the product, their attitude towards it intention to purchase.
will be more favourable and their purchase intention will similarly in Table 6 additionally presents the effect size (f2) which estimates the
crease. Likewise, their perception of the risk associated with said fashion contribution of each exogenous construct to the R2 of an endogenous
product is reduced when they trust in the authenticity of the influencer's latent variable. The f2 values of 0.02, 0.15 and 0.35 indicate a small,
recommendations. moderate and large effect respectively. According to the results ob
Hypotheses 6 and 7 are also supported: perceived risk is negatively tained, the effect of each variable on attitude would be small for
and significantly related to both users' attitude towards the product and perceived risk (0.005), small to moderate (0.080) for user involvement
10
with the product, and large (0.431) in the case of perceived message (under and over 25 years of age).
authenticity (PMA), which would therefore be the variable with the Table 7 presents the result of the MGA using two alternative ap
greatest influence in contributing to the R2 of attitude and, conse proaches: Henseler's MGA (Henseler et al., 2009) and the Welch-
quently, to the explanation of its variance. With respect to purchase Satterthwaite test. Since both procedures yielded the same result, we
intention, user involvement, perceived control and perceived risk have only show the statistically significant difference between the path co
almost no effect in explaining their variance, the effect is small in the efficients for the two age groups. Specifically, the relationship between
case of perceived message authenticity (0.015), small to moderate for product involvement (fashion consciousness), and attitude is positive
subjective norm (0.094) and moderate to large for attitude (0.176), the and significant in both age groups. However, the influence is signifi
most influential construct in this case on intention (which can also be cantly greater among respondents younger than 25.
deduced from its high path coefficient). Analysing the effects of the It should be noted, however, that these results should be taken with
variables affecting the perception of authenticity of the message, it can caution because the data come from a convenience sample obtained
be seen that shared content (S) has no effect, while Likes and Comments using snowball sampling. Consequently, it would not be possible to
have a small effect, with practically equal values (0.024 and 0.025), claim that the conclusions drawn here concerning the effects of gender
while user involvement is the most influential construct with a and age on the variables in the model are replicable in the full
moderate-high effect (0.176). Finally, risk perception, with a lower R2 population.
coefficient, as shown in Table 5, is only influenced by the low effect of
Comments (0.019) and the low to moderate effect of the perception of 5. Discussion and conclusions
authenticity of the message (0.027), with the effect of Likes and Shares
being non-existent. In a promotional context dominated by social networks, we use an
To finalize the validation of the structural model, the standardized audience-focused approach based on the ETPB (Ajzen, 1991; Meng and
root mean square residual (SRMR) has a value of 0.061; as it is below Choi, 2016) and OL (Bandura, 1977; Bikhchandani et al., 1998) to shed
0.10, this figure indicates the goodness of fit of the model (Ringle et al., light on the new paradigm of influencer endorsements in the fields of
2009). academia and business. This research develops and tests a new explor
atory model of the impact of recommendations on purchase intention.
The results point to the presence of general risk perceived from the
4.4. PLS-MGA multi-group analysis endorsements (Barcelos et al., 2018), which is mitigated by the
perceived authenticity of the message conveyed, based on the audience's
After analysing the model for the complete sample, we checked the involvement with the product and two key strategic metrics (“likes” and
effect of the sociodemographic variables gender and age on the path “comments”). The results reported here add to the emerging research in
coefficients of the structural model. To that end, we applied the the field of influencer marketing and are consistent with current studies
bootstrap-based partial least squares multi-group analysis (PLS-MGA) to on source credibility (Kim and Kim, 2021). We can highlight the
each of the variables, in order to analyse the differences in the co following outcomes:
efficients depending on the categories of each of these possible moder The first important finding concerns OL, measured through the
ating variables. number of “likes” and “comments” (H1a, H1c) that the followers of
To apply and interpret the PLS-MGA, we first had to check for influencers see, as it proved to be a reasonable predictor of greater
measurement invariance. This requirement must be met in order to be perceived message authenticity (PMA). This result was not surprising,
able to make an appropriate comparison between the two groups in and is in line with studies such as that by Borah and Xiao (2018), which
terms of the estimations of the standardized path coefficients in the shows how messages from a source considered an expert in a specific
structural relationships of the composites. To that end, we applied the field, with a high number of “likes”, are considered more credible,
three-step method used to assess the measurement invariance of com especially based on the social endorsement from other users, which in
posite models (MICOM) developed by Henseler et al. (2016). The three turn improves the favourable cognitive evaluation of the authenticity of
steps are as follows: 1) compliance with configural invariance; 2) the online message (Metzger and Flanagin, 2013). In this regard, Kim
compliance with compositional invariance; 3) equality of means and and Yang (2017) point out the distinctive nature of the three indicators:
variances. Meeting the requirements in the first two steps indicates a “like” is a spontaneous affective impulse linked with feelings of
partial measurement invariance, and if the third step is also verified it approval, a “comment” is a cognitively triggered behaviour, while a
indicates total measurement invariance. “share” is a combination of the two. This may be why the latter predictor
The application of MICOM for the two factors whose moderating can raise more doubts among the audience, and thus why the hypothesis
effect is under analysis confirmed that compositional invariance was not about shared content (H1c) was not supported. From an OL perspective,
established for gender. Therefore, it would not make sense to interpret this finding would indicate that the number of times content is shared
the results of the multigroup analysis for this variable. That is, the re seems to be a more confusing metric, as the motivations for the audience
spondent's gender cannot be considered to have a moderating effect on sharing a message might not be as clearly identifiable; for example,
the relationships between the constructs of the proposed model. Gender content may be shared out of interest, but also for fun or out of concern.
does not cause heterogeneity in the results of the structural model These types of effects have previously been identified in the field of
estimated for the entire sample. influencer marketing, specifically in the context of impulse buying
In the case of age, compositional invariance was established but not (Zafar et al., 2021), where it was found that the consumer identified
equality of means and variances, indicating partial measurement more with positive reviews, and consequently would be more affected
invariance. The next step was thus to compare the two groups proposed
Table 7
Multi-group analysis by age.
Age Path coefficients Path coefficient difference Henseler's MGA Welch-Satterthwaite test
<25 years >25 years β1-β2 p-value Student-t p-value
INV ➔ A 0.259*** 0.134*** 0.125** 0.05** 1.985 0.048**

***
p < 0.01.
**
p < 0.05.
11
by the numerical cues that can be more accurately identified. Previous line with the effects identified by Huang et al. (2010) in their study of
literature has contributed to the widely-held belief that a greater num the involvement of travel blog followers and their purchase intention. In
ber of “likes” emits a signal of trustworthiness (De Vries, 2019; Seo et al., said research, involvement positively influenced the attitude towards
2019), and could mitigate risks. However, the second of our findings both sponsorships and the sponsoring brands. At the same time, as in the
seems to somehow contradict this; indeed, we were surprised that OL study by Huang et al. (2010), we found no relationship between product
(based on the observation of quantitative metrics) was not found to involvement and purchase intention. This may well be due to the fact
mitigate perceived risk (H2a, b and c). That said, this finding does make that people who are very involved with fashion products follow influ
sense in a way, when viewed from the perspective of the dangers asso encers only in order to keep an eye on trends. Huang et al. (2010)
ciated with the use of influencers for product promotion (Campbell and explain how the level of product involvement in purchasing decisions
Farrell, 2020): it has been pointed out that the historical focus on can be mixed (Park and Lee, 2008). That is, less involved consumers pay
quantitative metrics has prompted some influencers to pay for bots to more attention to the popularity of the endorsement, thus improving
inflate the numbers they consider strategically relevant. Therefore, a their attitude and intention, while those who are very involved would
plausible explanation could be that the audience is increasingly aware of also see an improvement in their attitude, but would need to process
the sponsorship and the financial orientation of the influencer (Folkvord more information, letting themselves be guided by their involvement
et al., 2019; Konstantopoulou et al., 2019). before making a purchase decision. A more in-depth examination leads
Third, we explore the perceived authenticity of the influencer's us to one of the few studies to address involvement in celebrity en
message from a triple perspective: as a determinant of the attitude to dorsements, which highlights the need to explore the degree of
wards the product (H3), as a determinant of purchase intention (H4), involvement, as well as the application of the construct to different
and as a mitigating factor of the risk perceived by the audience (H5). The product categories (Parmar et al., 2020). Also noteworthy is the direct
results indicate a strong impact on attitudes and a significant effect on effect between product involvement and the perceived authenticity of
intention. Consequently, they add to the abundant previous literature the message (H8). This adds to the findings reported by Xue and Zhou
that identifies the trustworthiness, authenticity and credibility of the (2010), who indicate how greater product involvement leads to greater
message conveyed as the main causes of the impacts on purchase atti trust in the source, although they also point out that the greater the
tudes and intention in different contexts (Chakraborty and Bhat, 2018; involvement, the more likely people are to give credibility to negative
Yoon and Kim, 2016). For example, in the field of influencer marketing, reviews over positive ones (for less involved people).
pioneering studies such as that by Chu and Kamal (2008) found that the Finally, regarding the influence on intention exerted by the three key
credibility of bloggers' message significantly influenced consumers' at elements of the TPB—perceived control (H11), attitude (H12), and
titudes towards brands mentioned in their blogs. More recently, the subjective norm (H13)—the results are consistent with recent work in
study by Xiao et al. (2018) reported the same effects on Youtube, as did the field of influencers pointing to significant impacts of attitudes and
Kim and Kim (2021), but without focusing on a specific social network. subjective norm (Chetioui et al., 2020; Choi and Rifon, 2012; Chopra
Regarding the influence of the authenticity of the message on perceived et al., 2020), and non-significant relationships in the case of perceived
risk, this relationship was also confirmed, in line with the findings of control (H11). As indicated by Choi and Johnson (2019) this situation is
Kim and Lennon (2013), who showed how credibility based on reputa likely to occur in samples of users with a great deal of experience and a
tion is a determinant of consumers' emotions and has a significant high level of control over their ability to buy online, as shown by the
negative effect on perceived risk. Similarly, Hussain et al. (2017) in the sociodemographic analysis of the sample. A similar effect was recently
context of electronic word-of-mouth (eWOM), found that several di found in the study by Jain et al. (2017) focusing on luxury fashion items,
mensions of source credibility (expertness, trustworthiness and objec which would support this possible explanation.
tivity) contribute substantially to reducing perceived risk.
The fourth relevant finding relates to the influence of perceived risk 5.1. Theoretical implications
on both attitude towards recommendations (H6) and purchase intention
(H7). In the field of influencer marketing, there have been very few The main theoretical implication of this study lies in the inclusion of
studies that have considered the perception of risk as a factor with a risk perception when analysing the impact of influencers' recommen
significant influence on the purchase decision promoted by influencers dations on their followers' purchase intentions. In this line, Hypotheses 6
(Hudders et al., 2021). Moreover, some conflicting results have been and 7 confirmed the negative and statistically significant influence of
reported (Casalo et al., 2020; Deshbhag and Mohan, 2020). Perceived perceived risk on product attitude and purchase intention. Nevertheless,
risk is a well-established construct in the field of research on purchasing the associated f2 values showed a very small effect of risk on the
processes (Rehman et al., 2020). In this study, we show its influence on explanatory variance of these two variables. This result implies that
the two main antecedents of the eventual purchase behaviour, finding theoretical models of influencers' recommendations and their effects on
that it reduces both attitudes and intention. In a way, this finding un their followers should not neglect the inclusion of the risk variable as an
derscores the need to analyse the construct in future research, as it could antecedent of product attitude. However, it would be necessary to
be a mistake to assume that influencers' endorsements are always seen as reconsider the way in which this risk is measured, since a generic
reliable (Casalo et al., 2020). The impact was found to be greater on analysis of it, as considered in this study, has not had a significant effect.
attitudes than on intention, although it was significant in both cases. In contrast, its observation through the identification of its components,
Furthermore, the influence was moderated by the perceived authenticity as suggested in the study by Forsythe and Shi (2003), could generate
of the message, which acted as a mitigating factor (H5); as such, more relevant results when it comes to determining how its influence on
authenticity can be seen as a key tool for lessening the perceived risk. In product attitude and purchase intention is articulated.
our research, as in the study by Barcelos et al. (2018), we assessed the Furthermore, to the best of our knowledge, this study is one of the
general risk stemming from following the influencer's recommenda few that addresses the effects of product involvement on the perceived
tions. Nevertheless, as Marriott and Williams (2018) point out, risk and authenticity of the sponsored message as factors mitigating the
trust are closely linked, and a one-dimensional examination provides perceived risk in influencer recommendations. In doing so, it puts for
only a partial understanding of their effects. ward a valuable approach for gaining a better understanding of the
Our fifth finding is about the connections between product presence of general risk associated with influencer endorsements.
involvement, measured through fashion consciousness (in the context of Therefore, this study contributes to both the knowledge base on social
this type of product), perceived message authenticity (H8), and attitude changes that influence new consumer behaviours and to the existing
towards recommendations made by influencers (H9). With regard to literature on influencer marketing (Enke and Borchers, 2019; Hudders
involvement, we find a significant and positive influence on attitude, in et al., 2021). The proposed model can be applied to any consumer
12
product recommended by influencers; for example, by adapting the 2012; Singh et al., 2021; Xu et al., 2004; Zaichkowsky, 1985). While it
product involvement construct (ecological, technological categories, should be noted that the reliability and validity of the measures were
etc.). On the other hand, it extends the theory of planned behaviour found to be satisfactory, future research should seek to build on the
(Ajzen, 1991; Meng and Choi, 2016) and combines it with observational knowledge produced here by validating new instruments specifically
learning theory (Bandura, 1977; Bikhchandani et al., 1998), using three designed for research into influencer marketing. The second is linked to
metrics considered strategic in the field of marketing and social net the concept of risk used, measured here as a general perception (Bar
works (Zafar et al., 2021). celos et al., 2018). Previous literature suggests we should seek out more
accurate antecedents for both risk and trust (Biswas et al., 2006).
5.2. Management implications Moving forward, this will make it possible to evaluate said antecedents
of risk, and identify those with the greatest impact on influencers' au
The study raises specific practical implications, especially for mar diences. Third, this cross-sectional quantitative research is conducted
keters and brands interested in using influencers. Firstly, a larger using a relatively large convenience sample, in a specific context
number of followers (Pittman and Abell, 2021) or interactions (such as following the Covid-19 lockdowns, and for a specific product (fashion).
“likes”) does not always have positive associations for the buyer. De Accordingly, the model should be verified for other types of products,
Vries (2019) indicates that “likes” from unwanted users pose a risk to the and in a context in which possible changes in the buyer's behaviour as a
sponsor, as do paid-for interactions (an excessive number of them); result of lockdown have less impact on the results. Lastly, the conve
followers observe not only the number, but also who “shares”, “com nience sample consists of followers of influencers; this is a novel feature
ments” or “likes”. Indeed, in our analysis, the only quantitative indicator with respect to previous studies, but it raises the question of the extent to
that turned out to be significant, albeit in the opposite direction to the which the results can be generalized. Consequently, future analyses
expected relationship, was the number of comments. This could mean should examine the general public that occasionally sees or follows a
that an influencer's followers view a large number of comments as sponsored recommendation.
indicating risks associated with the sponsored product, which would be
more easily observable (De Vries, 2019). This result has important im Data availability
plications for marketing executives as they will therefore need to focus
their efforts on targeting the comments generated by influencers' fol Data will be made available on request.
lowers, paying particular attention to the content, rather than the
quantity received. References
Secondly, the perceived authenticity of the influencer's message is
found to be the most effective tool for mitigating possible risks stemming Ajitha, S., Sivakumar, V.J., 2019. The moderating role of age and gender on the attitude
towards new luxury fashion brands. J. Fash. Mark. Manag. 23 (4), 440–465. https://
from the recommendations. Therefore, the choice of the right influencer doi.org/10.1108/JFMM-05-2018-0074.
for a particular campaign should be based on a study of the credibility of Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50
their messages. (2), 179–211.
Alalwan, A.A., Dwivedi, Y.K., Rana, N.P., Algharabat, R., 2018. Examining factors
Thirdly, the multigroup analysis showed some heterogeneity in the influencing Jordanian customers’ intentions and adoption of internet banking:
effects, although only for the age variable in terms of the relationship extending UTAUT2 with risk, 2018/01/01/ J. Retail. Consum. Serv. 40, 125–138.
between involvement and attitude towards recommendations. In the https://doi.org/10.1016/j.jretconser.2017.08.026.
AlSharji, A., Ahmad, S.Z., Abu Bakar, A.R., 2018. Understanding social media adoption
field of fashion, it is not uncommon to come across such findings; for in SMEs. J. Entrep. Emerg. Econ. 10 (2), 302–328. https://doi.org/10.1108/JEEE-
example, in the studies by Ajitha and Sivakumar (2019) or Schade et al. 08-2017-0058.
(2016). On this occasion, we found greater involvement with fashion in Arora, A., Bansal, S., Kandpal, C., Aswani, R., Dwivedi, Y., 2019. Measuring social media
influencer index- insights from facebook, Twitter and Instagram [Article]. Jul
the younger age group, which points to the inclusion of variables related
J. Retail. Consum. Serv. 49, 86–101. https://doi.org/10.1016/j.
to product involvement in the segmentation of influencer audiences. In jretconser.2019.03.012.
contrast, gender did not cause heterogeneity in the model, which sug Bandura, A., 1977. Social Learning Theory, Vol. 1.
Barcelos, R.H., Dantas, D.C., Sénécal, S., 2018. Watch your tone: how a brand's tone of
gests that the perception of risk and authenticity of the message, as well
voice on social media influences consumer responses, 2018/02/01/ J. Interact.
as its impact on the attitude or purchase intention of fashion products, Mark. 41, 60–80. https://doi.org/10.1016/j.intmar.2017.10.001.
does not have different connotations depending on the gender of an Bashari, B., Fazl-Ersi, E., 2020. Influential post identification on Instagram through
influencer's followers. Nonetheless, this result is limited to this type of caption and hashtag analysis. Meas. Control 53 (3–4), 409–415. https://doi.org/
10.1177/0020294019877489, 2020/03/01.
product, so it is essential to continue to consider gender as a possible Bikhchandani, S., Hirshleifer, D., Welch, I., 1998. Learning from the behavior of others:
control variable when analysing these models. conformity, fads, and informational cascades. J. Econ. Perspect. 12 (3), 151–170.
Finally, a very important practical aspect for the management of http://www.jstor.org/stable/2647037.
Biswas, D., Biswas, A., Das, N., 2006. The differential effects of celebrity and expert
social media marketing is the prior need to determine the product endorsements on consumer risk perceptions. The role of consumer knowledge,
involvement of the potential audience to be reached. In this connection, perceived congruency, and product technology orientation. J. Advert. 35 (2), 17–31.
the results of this study reflect the significant positive influence of the https://doi.org/10.1080/00913367.2006.10639231, 2006/05/01.
Boerman, S.C., 2020. The effects of the standardized Instagram disclosure for micro-and
user's fashion awareness on their perception of the authenticity of the meso-influencers [Article]. Feb Comput. Hum. Behav. 103, 199–207. https://doi.
message, as well as on their attitude towards recommendations. org/10.1016/j.chb.2019.09.015.
Consequently, it is necessary to direct marketing efforts to the segment Boerman, S.C., Willemsen, L.M., Van Der Aa, E.P., 2017. “This post is sponsored”: effects
of sponsorship disclosure on persuasion knowledge and electronic word of mouth in
of followers who have a higher degree of involvement with the partic
the context of Facebook, 2017/05/01/ J. Interact. Mark. 38, 82–92. https://doi.org/
ular product or service being recommended by the influencer. 10.1016/j.intmar.2016.12.002.
Borah, P., Xiao, X., 2018. The importance of ‘Likes’: the interplay of message framing,
source, and social endorsement on credibility perceptions of health information on
5.3. Limitations and future lines of research
Facebook. J. Health Commun. 23 (4), 399–411. https://doi.org/10.1080/
10810730.2018.1455770, 2018/04/03.
This research is subject to certain limitations, some of which open up Borchers, N.S., 2019. Social media influencers in strategic communication. Int. J. Strateg.
new opportunities for future work. The first concerns the exploratory Commun. 13 (4), 255–260. https://doi.org/10.1080/1553118X.2019.1634075,
2019/08/08.
nature of the model. The effects of the audience's product involvement Boulding, W., Kirmani, A., 1993. A consumer-side experimental examination of signaling
and the perceived risk in influencers' recommendations have received theory: do consumers perceive warranties as signals of quality? J. Consum. Res. 20
very little attention to date. From a methodological standpoint, this (1), 111–123. https://doi.org/10.1086/209337. %J Journal of Consumer Research.
required the adaptation and application of scales primarily drawn from

other areas (Alalwan et al., 2018; Lertwannawit and Mandhachitara,
13
Braojos-Gomez, J., Benitez-Amado, J., Javier Llorens-Montes, F., 2015. How do small intention more?, 2020/11/01/ Heliyon 6 (11), e05543. https://doi.org/10.1016/j.
firms learn to develop a social media competence? Int. J. Inf. Manag. 35 (4), heliyon.2020.e05543.
443–458. https://doi.org/10.1016/j.ijinfomgt.2015.04.003, 2015/08/01/. Enke, N., Borchers, N.S., 2019. Social media influencers in strategic communication: a
Bugshan, H., Attar, R.W., 2020. Social commerce information sharing and their impact conceptual framework for strategic social media influencer communication. Int. J.
on consumers [article]. Apr Technol. Forecast. Soc. Chang. 153 (8), 119875. https:// Strateg. Commun. 13 (4), 261–277. https://doi.org/10.1080/
doi.org/10.1016/j.techfore.2019.119875. 1553118X.2019.1620234, 2019/08/08.
Cabeza-Ramirez, L.J., Fuentes-Garcia, F.J., Munoz-Fernandez, G.A., 2021. Exploring the Faisal, C.M.N., Fernandez-Lanvin, D., De Andres, J., Gonzalez-Rodriguez, M., 2020.
emerging domain of research on video game live streaming in web of science: state Design quality in building behavioral intention through affective and cognitive
of the art, changes and trends [review]. Mar Int. J. Environ. Res. Public Health 18 involvement for e-learning on smartphones [Article]. Oct Internet Res. 30 (6),
(6), 27. https://doi.org/10.3390/ijerph18062917. Article 2917. 1631–1663. https://doi.org/10.1108/intr-05-2019-0217.
Campbell, C., Farrell, J.R., 2020. More than meets the eye: the functional components Folkvord, F., Bevelander, K.E., Rozendaal, E., Hermans, R., 2019. Children’s bonding
underlying influencer marketing. Bus. Horiz. 63 (4), 469–479. https://doi.org/ with popular YouTube vloggers and their attitudes toward brand and product
10.1016/j.bushor.2020.03.003, 2020/07/01/. endorsements in vlogs: an explorative study. Young Consum. 20 (2) https://doi.org/
Carmines, E.G., Zeller, R.A., 1979. Quantitative Applications in the Social Sciences: 10.1108/YC-12-2018-0896.
Reliability and Validity Assessment. S. Publications. https://doi.org/10.4135/ Fornell, C., Larcker, D.F., 1981. Evaluating structural equation models with unobservable
9781412985642. variables and measurement error. J. Mark. Res. 18 (1), 39–50. https://doi.org/
Casalo, L.V., Flavian, C., Ibanez-Sanchez, S., 2020. Sep). Influencers on instagram: 10.2307/3151312.
antecedents and consequences of opinion leadership [Article]. J. Bus. Res. 117, Forsythe, S.M., Shi, B., 2003. Consumer patronage and risk perceptions in internet
510–519. https://doi.org/10.1016/j.jbusres.2018.07.005. shopping. J. Bus. Res. 56 (11), 867–875. https://doi.org/10.1016/S0148-2963(01)
Chakraborty, U., Bhat, S., 2018. The effects of credible online reviews on brand equity 00273-9, 2003/11/01/.
dimensions and its consequence on consumer behavior. J. Promot. Manag. 24 (1), Francalanci, C., Hussain, A., 2017. Influence-based twitter browsing with NavigTweet,
57–82. https://doi.org/10.1080/10496491.2017.1346541, 2018/01/02. 2017/03/01/ Inf. Syst. 64, 119–131. https://doi.org/10.1016/j.is.2016.07.012.
Chen, Y., Wang, Q., Xie, J., 2011. Online social interactions: a natural experiment on Freberg, K., Graham, K., McGaughey, K., Freberg, L.A., 2011. Who are the social media
word of mouth versus observational learning. J. Mark. Res. 48 (2), 238–254. https:// influencers? A study of public perceptions of personality [Article]. Mar Public Relat.
doi.org/10.1509/jmkr.48.2.238, 2011/04/01. Rev. 37 (1), 90–92. https://doi.org/10.1016/j.pubrev.2010.11.001.
Chetioui, Y., Benlafqih, H., Lebdaoui, H., 2020. How fashion influencers contribute to Friedmann, E., Lowengart, O., 2019. Gender segmentation to increase brand preference?
consumers' purchase intention [Article], 1974/04/01 J. Fash. Mark. Manag. 24 (3), The role of product involvement. J. Prod. Brand Manag. 28 (3), 408–420. https://
361–380. https://doi.org/10.1108/jfmm-08-2019-0157. doi.org/10.1108/JPBM-06-2018-1917.
Childers, C.C., Lemon, L.L., Hoy, M.G., 2019. #Sponsored #Ad: agency perspective on Gaenssle, S., Budzinski, O., 2021. Stars in social media: new light through old windows?
influencer marketing campaigns. J. Curr. Issues Res. Advert. 40 (3), 258–274. [Article]. Apr J. Media Bus. Stud. 18 (2), 79–105. https://doi.org/10.1080/
https://doi.org/10.1080/10641734.2018.1521113, 2019/09/02. 16522354.2020.1738694.
Chin, W.W., 1998. The partial least squares approach for structural equation modeling. Gilmore, J.H., Pine, B.J., 2007. Authenticity: What Consumers Really Want. Harvard
In: Modern Methods for Business Research. Lawrence Erlbaum Associates Publishers, Business School Press Boston, MA.
pp. 295–336. Gold, A.H., Malhotra, A., Segars, A.H., 2001. Knowledge management: an organizational
Choi, D., Johnson, K.K.P., 2019. Influences of environmental and hedonic motivations on capabilities perspective. J. Manag. Inf. Syst. 18 (1), 185–214. https://doi.org/
intention to purchase green products: an extension of the theory of planned 10.1080/07421222.2001.11045669, 2001/05/31.
behavior. Sustain. Prod. Consum. 18, 145–155. https://doi.org/10.1016/j. Gutman, J., Mills, M.K., 1982. Fashion life-style, self-concept, shopping orientation, and
spc.2019.02.001, 2019/04/01/. store patronage-an integrative analysis. J. Retail. 58 (2), 64–86.
Choi, S.M., Rifon, N.J., 2012. It is a match: the impact of congruence between celebrity Hair, J.F., Sarstedt, M., Hopkins, L., Kuppelwieser, V.G., 2014. Partial least squares
image and consumer ideal self on endorsement effectiveness [doi:10.1002/ structural equation modeling (PLS-SEM) an emerging tool in business research
mar.20550]. Psychol. Mark. 29 (9), 639–650. https://doi.org/10.1002/mar.20550, [Review]. Eur. Bus. Rev. 26 (2) https://doi.org/10.1108/ebr-10-2013-0128, 106-+.
2012/09/01. Hair Jr., J.F., Matthews, L.M., Matthews, R.L., Sarstedt, M., 2017. PLS-SEM or CB-SEM:
Chopra, A., Avhad, V., Jaju, Sonali, 2020. Influencer marketing: an exploratory study to updated guidelines on which method to use. Int. J. Multivar. Data Anal. 1 (2),
identify antecedents of consumer behavior of millennial, 2021/01/01 Bus. Perspect. 107–123.
Res. 9 (1), 77–91. https://doi.org/10.1177/2278533720923486. Hair Jr., J.F., Hult Jr., G.T.M., Ringle Jr., C.M., Sarstedt Jr., M., 2021. A Primer on Partial
Chu, S.-C., Kamal, S., 2008. The effect of perceived blogger credibility and argument Least Squares Structural Equation Modeling (PLS-SEM). Sage publications.
quality on message elaboration and brand attitudes. J. Interact. Advert. 8 (2), 26–37. Hajli, N., 2015. Social commerce constructs and consumer's intention to buy [Article],
https://doi.org/10.1080/15252019.2008.10722140, 2008/03/01. 2017/03/01/ Int. J. Inf. Manag. 35 (2), 183–191. https://doi.org/10.1016/j.
Coates, A.E., Hardman, C.A., Halford, J.C.G., Christiansen, P., Boyland, E.J., 2019. Social ijinfomgt.2014.12.005.
media influencer marketing and children's food intake: a randomized trial [article]. Halim, E., Wardaya, P., Rianto, A., Hebrard, M., Ieee, 2020. The Impact of Marketing
Apr Pediatrics 143 (4), 9. https://doi.org/10.1542/peds.2018-2554. Article Influencer and Information Quality to Purchase Intention of Instagram Users, 2018/
e20182554. 05/01/. Ieee.
Coltman, T., Devinney, T.M., Midgley, D.F., Venaik, S., 2008. Formative versus reflective Han, H., Hyun, S.S., 2018. Role of motivations for luxury cruise traveling, satisfaction,
measurement models: two applications of formative measurement. J. Bus. Res. 61 and involvement in building traveler loyalty [Article]. Mar Int. J. Hosp. Manag. 70,
(12), 1250–1262. https://doi.org/10.1016/j.jbusres.2008.01.013, 2008/12/01/. 75–84. https://doi.org/10.1016/j.ijhm.2017.10.024.
Cox, D.F., Rich, S.U., 1964. Perceived risk and consumer decision-making—the case of He, W., Wang, F.-K., Chen, Y., Zha, S., 2017. An exploratory investigation of social media
telephone shopping. J. Mark. Res. 1 (4), 32–39. https://doi.org/10.1177/ adoption by small businesses. Inf. Technol. Manag. 18 (2), 149–160. https://doi.org/
002224376400100405, 1964/11/01. 10.1007/s10799-015-0243-3, 2017/06/01.
Das, G., Wiener, H.J.D., Kareklas, I., 2019. To emoji or not to emoji? Examining the Henseler, J., Ringle Christian, M., Sinkovics Rudolf, R., 2009. The use of partial least
influence of emoji on consumer reactions to advertising, 2019/03/01/ J. Bus. Res. squares path modeling in international marketing. In: Rudolf, R.S., Pervez, N.G.
96, 147–156. https://doi.org/10.1016/j.jbusres.2018.11.007. (Eds.), New Challenges to International Marketing, Vol. 20. Emerald Group
Dash, G., Paul, J., 2021. CB-SEM vs PLS-SEM methods for research in social sciences and Publishing Limited, pp. 277–319. https://doi.org/10.1108/S1474-7979(2009)
technology forecasting. Technol. Forecast. Soc. Chang. 173, 121092 https://doi.org/ 0000020014.
10.1016/j.techfore.2021.121092, 2021/12/01/. Henseler, J., Ringle, C.M., Sarstedt, M., 2016. Testing measurement invariance of
De Veirman, M., Hudders, L., Nelson, M.R., 2019. What is influencer marketing and how composites using partial least squares. Int. Mark. Rev. 33 (3), 405–431. https://doi.
does it target children? A review and direction for future research [review]. Dec org/10.1108/IMR-09-2014-0304.
Front. Psychol. 10 (16), 2685. https://doi.org/10.3389/fpsyg.2019.02685. Hilverda, F., Kuttschreuter, M., 2018. Online information sharing about risks: the case of
De Vries, E.L.E., 2019. When more likes is not better: the consequences of high and low organic food [Article]. Sep Risk Anal. 38 (9), 1904–1920. https://doi.org/10.1111/
likes-to-followers ratios for perceived account credibility and social media marketing risa.12980.
effectiveness, 2019/12/01 Mark. Lett. 30 (3), 275–291. https://doi.org/10.1007/ Hilverda, F., Kuttschreuter, M., Giebels, E., 2018. In: The Effect of Online Social Proof
s11002-019-09496-6. Regarding Organic Food: Comments and Likes on Facebook [Original Research], 3,
Dean, D.H., Biswas, A., 2001. Third-party organization endorsement of products: an p. 30. https://doi.org/10.3389/fcomm.2018.00030. <Go to ISI>://WOS:
advertising Cue affecting consumer prepurchase evaluation of goods and services. 000563242400010.
J. Advert. 30 (4), 41–57. https://doi.org/10.1080/00913367.2001.10673650, Huang, C.-Y., Chou, C.-J., Lin, P.-C., 2010. Involvement theory in constructing bloggers'
2001/12/01. intention to purchase travel products. Tour. Manag. 31 (4), 513–526. https://doi.
Deshbhag, R.R., Mohan, B.C., 2020. Study on influential role of celebrity credibility on org/10.1016/j.tourman.2009.06.003, 2010/08/01/.
consumer risk perceptions. J. Indian Bus. Res. 12 (1), 79–92. https://doi.org/ Hudders, L., De Jans, S., De Veirman, M., 2021. The commercialization of social media
10.1108/JIBR-09-2019-0264. stars: a literature review and conceptual framework on the strategic use of social
Djafarova, E., Rushworth, C., 2017. Exploring the credibility of online celebrities' media influencers [Review]. Int. J. Advert. 40 (3), 327–375. https://doi.org/
Instagram profiles in influencing the purchase decisions of young female users, 10.1080/02650487.2020.1836925.
2017/03/01/ Comput. Hum. Behav. 68, 1–7. https://doi.org/10.1016/j. Hussain, S., Ahmed, W., Jafar, R.M.S., Rabnawaz, A., Jianzhou, Y., 2017. eWOM source
chb.2016.11.009. credibility, perceived risk and food product customer's information adoption, 2017/
Duncan, O.D., 1966. In: Path Analysis: Sociological Examples, 72(1), pp. 1–16. https:// 01/01/ Comput. Hum. Behav. 66, 96–102. https://doi.org/10.1016/j.
doi.org/10.1086/224256. chb.2016.09.034.
Dwidienawati, D., Tjahjana, D., Abdinagoro, S.B., Gandasari, D., Munawaroh, 2020.
Customer review or influencer endorsement: which one influences purchase
14
Hwang, C., Kim, T.H., 2020. Muslim women’s purchasing behaviors toward modest Metzger, M.J., Flanagin, A.J., 2013. Credibility and trust of information in online
activewear in the United States, 2021/07/01 Cloth. Text. Res. J. 39 (3), 175–189. environments: the use of cognitive heuristics, 2013/12/01/ J. Pragmat. 59,
https://doi.org/10.1177/0887302X20926573. 210–220. https://doi.org/10.1016/j.pragma.2013.07.012.
InfluencerMarketingHub, 2021a. The state of influencer marketing 2021: benchmark Monsalve, A.R., 2017. Influencers and the Development of Fashion Trends: An Analysis
report. https://influencermarketinghub.com/influencer-marketing-benchmark-rep of Opinion Leaders in Social Networks. Univ Rosario. <Go to ISI>://WOS:
ort-2021/. 000563242400010.
InfluencerMarketingHub, 2021b. Top fashion influencers in Spain. https://influence Mou, J., Zhu, W., Benyoucef, M., 2020. Impact of product description and involvement
rmarketinghub.com/top-fashion-influencers-spain/. on purchase intention in cross-border e-commerce. Ind. Manag. Data Syst. 120 (3),
Jain, S., Khan, M.N., Mishra, S., 2017. Understanding consumer behavior regarding 567–586. https://doi.org/10.1108/IMDS-05-2019-0280.
luxury fashion goods in India based on the theory of planned behavior. J. Asia Bus. Nam, J., Hamlin, R., Gam, H.J., Kang, J.H., Kim, J., Kumphai, P., Starr, C., Richards, L.,
Stud. 11 (1), 4–21. https://doi.org/10.1108/JABS-08-2015-0118. 2007. In: The Fashion-conscious Behaviours of Mature Female Consumers, 31(1),
Jibril, A.B., Kwarteng, M.A., Chovancova, M., Pilik, M., 2019. The impact of social media pp. 102–108. https://doi.org/10.1111/j.1470-6431.2006.00497.x.
on consumer-brand loyalty: a mediating role of online based-brand community, Nkwocha, I., Bao, Y., Brotspies, H.V., Johnson, W.C., 2005. Product fit and consumer
2019/01/01 Cogent Bus. Manag. 6 (1), 1673640. https://doi.org/10.1080/ attitude toward brand extensions: the moderating role of product involvement.
23311975.2019.1673640. J. Mark. Theory Pract. 13 (3), 49–61. http://www.jstor.org/stable/40470227.
Kamalul Ariffin, S., Mohan, T., Goh, Y.-N., 2018. Influence of consumers’ perceived risk Nouri, M., 2018. The power of influence: traditional celebrity vs social media influencer.
on consumers’ online purchase intention. J. Res. Interact. Mark. 12 (3), 309–327. In: Pop Culture Intersections, p. 32. https://scholarcommons.scu.edu/engl_176/32.
https://doi.org/10.1108/JRIM-11-2017-0100. Nunnally, J.C., Bernstein, I.H., 1994. Psychometric Theory, 3rd ed. McGraw-Hill, New
Khan, G.F., Swar, B., Lee, S.K., 2014. Social media risks and benefits: a public sector York.
perspective, 2014/10/01 Soc. Sci. Comput. Rev. 32 (5), 606–627. https://doi.org/ Park, D.-H., Lee, J., 2008. eWOM overload and its effect on consumer behavioral
10.1177/0894439314524701. intention depending on consumer involvement. Electron. Commer. Res. Appl. 7 (4),
Kim, C., Yang, S.-U., 2017. Like, comment, and share on Facebook: how each behavior 386–398. https://doi.org/10.1016/j.elerap.2007.11.004, 2008/12/01/.
differs from the other. Public Relat. Rev. 43 (2), 441–449. https://doi.org/10.1016/ Parmar, Y., Mann, B.J.S., Ghuman, M.K., 2020. Impact of celebrity endorser as in-store
j.pubrev.2017.02.006, 2017/06/01/. stimuli on impulse buying. Int. Rev. Retail Distrib. Consum. Res. 30 (5), 576–595.
Kim, D.J., Ferrin, D.L., Rao, H.R., 2008. A trust-based consumer decision-making model https://doi.org/10.1080/09593969.2020.1781229, 2020/10/19.
in electronic commerce: the role of trust, perceived risk, and their antecedents. Petrescu, M., O’Leary, K., Goldring, D., Ben Mrad, S., 2018. Incentivized reviews:
Decis. Support. Syst. 44 (2), 544–564. https://doi.org/10.1016/j.dss.2007.07.001, promising the moon for a few stars, 2018/03/01/ J. Retail. Consum. Serv. 41,
2008/01/01/. 288–295. https://doi.org/10.1016/j.jretconser.2017.04.005.
Kim, D.Y., Kim, H.-Y., 2021. Trust me, trust me not: a nuanced view of influencer Pittman, M., Abell, A., 2021. More trust in fewer followers: diverging effects of
marketing on social media, 2021/09/01/ J. Bus. Res. 134, 223–232. https://doi.org/ popularity metrics and green orientation social media influencers, 2021/11/01/
10.1016/j.jbusres.2021.05.024. J. Interact. Mark. 56, 70–82. https://doi.org/10.1016/j.intmar.2021.05.002.
Kim, H., Karpova, E., 2009. Consumer attitudes toward fashion counterfeits: application Rahman, I., 2018. The interplay of product involvement and sustainable consumption: an
of the theory of planned behavior, 2010/04/01 Cloth. Text. Res. J. 28 (2), 79–94. empirical analysis of behavioral intentions related to green hotels, organic wines and
https://doi.org/10.1177/0887302X09332513. green cars [article]. Jul-Aug Sustain. Dev. 26 (4), 399–414. https://doi.org/
Kim, J., Lennon, S.J., 2013. Effects of reputation and website quality on online 10.1002/sd.1713.
consumers' emotion, perceived risk and purchase intention. J. Res. Interact. Mark. 7 Raun, T., 2018. Capitalizing intimacy: new subcultural forms of micro-celebrity
(1), 33–56. https://doi.org/10.1108/17505931311316734. strategies and affective labour on YouTube. Convergence 24 (1), 99–113. https://
Kock, N., 2015. Common method bias in PLS-SEM: a full collinearity assessment doi.org/10.1177/1354856517736983, 2018/02/01.
approach [Article]. Oct-Dec Int. J. E-Collaboration 11 (4), 1–10. https://doi.org/ Rehman, Z.U., Baharun, R., Salleh, N.Z.M., 2020. Antecedents, consequences, and
10.4018/ijec.2015100101. reducers of perceived risk in social media: a systematic literature review and
Kock, N., Lynn, G.S., 2012. Lateral collinearity and misleading results in variance-based directions for further research [doi:10.1002/mar.21281]. Psychol. Mark. 37 (1),
SEM: an illustration and recommendations [Article]. J. Assoc. Inf. Syst. 13 (7), 74–86. https://doi.org/10.1002/mar.21281, 2020/01/01.
546–580. https://doi.org/10.17705/1jais.00302. Ringle, C.M., Götz, O., Wetzels, M., Wilson, B., 2009. On the use of formative
Konstantopoulou, A., Rizomyliotis, I., Konstantoulaki, K., Badahdah, R., 2019. Improving measurement specifications in structural equation modeling: a Monte Carlo
SMEs’ competitiveness with the use of instagram influencer advertising and eWOM. simulation study to compare covariance-based and partial least squares model
Int. J. Organ. Anal. 27 (2), 308–321. https://doi.org/10.1108/IJOA-04-2018-1406. estimation methodologies. In: METEOR Research Memoranda.
Kotler, P., Zaltman, G., 1971. Social marketing: an approach to planned social change. Saima, Khan, M.A., 2021. Effect of social media influencer marketing on consumers’
J. Mark. 35 (3), 3–12. https://doi.org/10.2307/1249783. purchase intention and the mediating role of credibility. J. Promot. Manag. 27 (4),
Kowalczyk, C.M., Pounders, K.R., 2016. Transforming celebrities through social media: 503–523. https://doi.org/10.1080/10496491.2020.1851847.
the role of authenticity and emotional attachment. J. Prod. Brand Manag. 25 (4), Sánchez-Cañizares, S.M., Cabeza-Ramírez, L.J., Muñoz-Fernández, G., Fuentes-García, F.
345–356. https://doi.org/10.1108/JPBM-09-2015-0969. J., 2021. Impact of the perceived risk from Covid-19 on intention to travel. Curr.
Lehdonvirta, V., Oksanen, A., Räsänen, P., Blank, G., 2021. In: Social Media, Web, and Issue Tour. 24 (7), 970–984. https://doi.org/10.1080/13683500.2020.1829571,
Panel Surveys: Using Non-Probability Samples in Social and Policy Research, 13(1), 2021/04/03.
pp. 134–155. https://doi.org/10.1002/poi3.238. Schade, M., Hegner, S., Horstmann, F., Brinkmann, N., 2016. The impact of attitude
Lertwannawit, A., Mandhachitara, R., 2012. Interpersonal effects on fashion functions on luxury brand consumption: an age-based group comparison. J. Bus. Res.
consciousness and status consumption moderated by materialism in metropolitan 69 (1), 314–322. https://doi.org/10.1016/j.jbusres.2015.08.003, 2016/01/01/.
men, 2012/10/01/ J. Bus. Res. 65 (10), 1408–1416. https://doi.org/10.1016/j. Schreuder, H.T., Gregoire, T.G., Weyer, J.P., 2001. For what applications can probability
jbusres.2011.10.006. and non-probability sampling be used? Environ. Monit. Assess. 66 (3), 281–291.
Li, D., Walejko, G., 2008. Splogs and abandoned blogs: the perils of sampling bloggers https://doi.org/10.1023/A:1006316418865, 2001/02/01.
and their blogs. Inf. Commun. Soc. 11 (2), 279–296. https://doi.org/10.1080/ Schwemmer, C., Ziewiecki, S., 2018. Social media sellout: the increasing role of product
13691180801947976, 2008/03/01. promotion on YouTube. Soc. Media Soc. 4 (3) https://doi.org/10.1177/
Lin, H.-C., Swarna, H., Bruning, P.F., 2017. Taking a global view on brand post 2056305118786720, 2056305118786720.
popularity: six social media brand post practices for global markets. Bus. Horiz. 60 Seo, Y., Kim, J., Choi, Y.K., Li, X., 2019. In “likes” we trust: likes, disclosures and firm-
(5), 621–633. https://doi.org/10.1016/j.bushor.2017.05.006, 2017/09/01/. serving motives on social media. Eur. J. Mark. 53 (10), 2173–2192. https://doi.org/
Lou, C., Yuan, S., 2019. Influencer marketing: how message value and credibility affect 10.1108/EJM-11-2017-0883.
consumer trust of branded content on social media. J. Interact. Advert. 19 (1), Sherif, M., Cantril, H., 1947. The Psychology of Ego-involvements: Social Attitudes and
58–73. https://doi.org/10.1080/15252019.2018.1533501. Identifications [doi:10.1037/10840-000]. John Wiley & Sons Inc. https://doi.org/
Marriott, H.R., Williams, M.D., 2018. Exploring consumers perceived risk and trust for 10.1037/10840-000.
mobile shopping: a theoretical framework and empirical study, 2018/05/01/ Shim, S., Morris, N.J., Morgan, G.A., 1989. Attitudes toward imported and domestic
J. Retail. Consum. Serv. 42, 133–146. https://doi.org/10.1016/j. apparel among college students: the fishbein model and external variables. Cloth.
jretconser.2018.01.017. Text. Res. J. 7 (4), 8–18. https://doi.org/10.1177/0887302X8900700402, 1989/06/
Martensen, A., Brockenhuus-Schack, S., Zahid, A.L., 2018. How citizen influencers 01.
persuade their followers. J. Fash. Mark. Manag. 22 (3), 335–353. https://doi.org/ Shin, E., Lee, J.E., 2021. What makes consumers purchase apparel products through
10.1108/JFMM-09-2017-0095. social shopping services that social media fashion influencers have worn? [Article].
Martínez-López, F.J., Anaya-Sánchez, R., Esteban-Millat, I., Torrez-Meruvia, H., Aug J. Bus. Res. 132, 416–428. https://doi.org/10.1016/j.jbusres.2021.04.022.
D’Alessandro, S., Miles, M., 2020. Influencer marketing: brand control, commercial Silva, R., Sampaio, A., Rodrigues, P., 2020. The Role of Digital Influencers on Buying
orientation and post credibility. J. Mark. Manag. 36 (17–18), 1805–1831. https:// Intention. Marketing and Smart Technologies, Singapore, 2020//.
doi.org/10.1080/0267257X.2020.1806906, 2020/11/21. Singh, S., Singh, N., Kalinić, Z., Liébana-Cabanillas, F.J., 2021. Assessing determinants
Masuda, H., Han, S.H., Lee, J., 2022. Impacts of influencer attributes on purchase influencing continued use of live streaming services: an extended perceived value
intentions in social media influencer marketing: mediating roles of characterizations. theory of streaming addiction. Expert Syst. Appl. 168, 114241 https://doi.org/
Technol. Forecast. Soc. Chang. 174, 121246 https://doi.org/10.1016/j. 10.1016/j.eswa.2020.114241, 2021/04/15/.
techfore.2021.121246, 2022/01/01/. SmartInsights, 2021. Global social media statistics research summary 2022. https://www
Meng, B., Choi, K., 2016. Extending the theory of planned behaviour: testing the effects .smartinsights.com/social-media-marketing/social-media-strategy/new-global-soc
of authentic perception and environmental concerns on the slow-tourist decision- ial-media-research/.
making process. Curr. Issue Tour. 19 (6), 528–544. https://doi.org/10.1080/ Sokolova, K., Kefi, H., 2020. Instagram and YouTube bloggers promote it, why should I
13683500.2015.1020773, 2016/05/11. buy? How credibility and parasocial interaction influence purchase intentions.
15
J. Retail. Consum. Serv. 53, 101742 https://doi.org/10.1016/j. Xu, Y., Summers, T.A., Belleau, B.D., 2004. Who buys American alligator?: predicting
jretconser.2019.01.011, 2020/03/01/. purchase intention of a controversial product. J. Bus. Res. 57 (10), 1189–1198.
Sokolova, K., Perez, C., 2021. You follow fitness influencers on YouTube. But do you https://doi.org/10.1016/S0148-2963(02)00327-2, 2004/10/01/.
actually exercise? How parasocial relationships, and watching fitness influencers, Xue, F., Zhou, P., 2010. The effects of product involvement and prior experience on
relate to intentions to exercise [Article]. J. Retail. Consum. Serv. 58 (11), 102276 Chinese consumers’ responses to online word of mouth. J. Int. Consum. Mark. 23 (1),
https://doi.org/10.1016/j.jretconser.2020.102276. 45–58. https://doi.org/10.1080/08961530.2011.524576, 2010/11/30.
Stubb, C., Nyström, A.-G., Colliander, J., 2019. Influencer marketing: the impact of Yoon, D., Kim, Y.-K., 2016. Effects of self-congruity and source credibility on consumer
disclosing sponsorship compensation justification on sponsored content responses to coffeehouse advertising. J. Hosp. Mark. Manag. 25 (2), 167–196.
effectiveness. J. Commun. Manag. 23 (2), 109–122. https://doi.org/10.1108/JCOM- https://doi.org/10.1080/19368623.2014.1001932, 2016/02/17.
11-2018-0119. Yuan, C.L., Moon, H., Kim, K.H., Wang, S.M., 2021. The influence of parasocial
Tafesse, W., Wood, B.P., 2021. Followers' engagement with instagram influencers: the relationship in fashion web on customer equity [Article]. Jun J. Bus. Res. 130,
role of influencers’ content and engagement strategy. J. Retail. Consum. Serv. 58, 610–617. https://doi.org/10.1016/j.jbusres.2019.08.039.
102303 https://doi.org/10.1016/j.jretconser.2020.102303, 2021/01/01/. Zafar, A.U., Qiu, J.N., Li, Y., Wang, J.G., Shahzad, M., 2021. The impact of social media
Taylor, J.W., 1974. The role of risk in consumer behavior: a comprehensive and celebrities' posts and contextual interactions on impulse buying in social commerce
operational theory of risk taking in consumer behavior, 1974/04/01 J. Mark. 38 (2), [Article]. Feb Comput. Hum. Behav. 115 (14), 106178. https://doi.org/10.1016/j.
54–60. https://doi.org/10.1177/002224297403800211. chb.2019.106178.
TheSocialFlame, 2021. The All-in-One platform for the management of INFLUENCERS Zaichkowsky, J.L., 1985. Measuring the involvement construct. J. Consum. Res. 12 (3),
marketing campaigns. https://thesocialflame.com/en. 341–352. http://www.jstor.org/stable/254378.
Tolson, A., 2010. A new authenticity? Communicative practices on YouTube. Crit. Zhang, Z.Y., 2021. Infrastructuralization of Tik Tok: transformation, power relationships,
Discourse Stud. 7 (4), 277–289. https://doi.org/10.1080/17405904.2010.511834, and platformization of video entertainment in China [Article]. Media Cult. Soc. 43
2010/11/01. (2), 219–236. https://doi.org/10.1177/0163443720939452.
Trivedi, J., Sama, R., 2020. The effect of influencer marketing on consumers’ brand Zhu, W., Mou, J., Benyoucef, M., 2019. Exploring purchase intention in cross-border E-
admiration and online purchase intentions: an emerging market perspective. commerce: a three stage model, 2019/11/01/ J. Retail. Consum. Serv. 51, 320–330.
J. Internet Commer. 19 (1), 103–124. https://doi.org/10.1080/ https://doi.org/10.1016/j.jretconser.2019.07.004.
15332861.2019.1700741, 2020/01/02.
Tsiros, M., Heilman, C.M., 2005. The effect of expiration dates and perceived risk on
L. Javier Cabeza-Ramírez received the PhD in Business Administration from the Uni
purchasing behavior in grocery store perishable categories. J. Mark. 69 (2),
versity of Córdoba (UCO) and he became a member of the Business Direction and Applied
114–129. https://doi.org/10.1509/jmkg.69.2.114.60762, 2005/04/01.
Economics Research Group the same year. He holds two bachelor's degrees in business
Vrontis, D., Makrides, A., Christofi, M., Thrassou, A., 2021. In: Social Media Influencer
administration and in Market Research and Techniques, as well as a Master in Taxation
Marketing: A Systematic Review, Integrative Framework and Future Research
System. He is currently an Associate Professor in Business Organization. His main research
Agenda, 45(4), pp. 617–644. https://doi.org/10.1111/ijcs.12647.
interests are related to entrepreneurship, videogames, bibliometrics, innovation, family
Wang, Y., Yu, C., 2017. Social interaction-based consumer decision-making model in
business and Corporate Social Responsibility.
social commerce: the role of word of mouth and observational learning. Int. J. Inf.
Manag. 37 (3), 179–189. https://doi.org/10.1016/j.ijinfomgt.2015.11.005, 2017/
06/01/. Sandra M. Sánchez-Cañizares is the Head of the Business Direction and Applied Eco
WeAreSocial, 2020. Influencer marketing post covid-19: cómo ha evolucionado y cómo nomics Research Group. She received the PhD in Business Administration from the Uni
las marcas pueden adaptarse a los cambios. https://wearesocial.com/es/blog/20 versity of Córdoba (UCO). She holds two bachelor's degrees in business administration and
20/07/influencer-marketing-covid19-marcas-adaptarse-cambios. in Market Research and Techniques. She is currently a Professor in Business Organization
WeAreSocial, Hootsuite, 2021. Digital 2021 global overview report. https://datareportal. at University of Córdoba. Her main research interests are related to entrepreneurship,
com/reports/digital-2021-global-overview-report. tourism, organizational commitment and statistics and econometric techniques.
Wiedmann, K.P., Hennigs, N., Langner, S., 2010. Spreading the word of fashion:
identifying social influencers in fashion [Article]. J. Glob. Fash. Market. 1 (3),
Luna M. Santos-Roldán received the PhD in Business Administration from the University
142–153. https://doi.org/10.1080/20932685.2010.10593066.
of Córdoba UCO. She is currently a Senior Lecturer in Business Organization at University
Woobin, K.I.M., Jung, C.H., 2019. The effects of SNS fashion influencer authenticity on
of Córdoba. Her main research interests are related to entrepreneurship, internationali
follower behavior intention-focused on the mediation effect of fanship [Sns 패션 인
zation, marketing and social media.
플루언서 진정성이 팔로워 행동의도에 미치는 영향-팬쉽의 매개효과를 중심으로-]
[research-article]. J. Korean Soc. Cloth. Text. 43 (1), 17–32. https://doi.org/
10.5850/jksct.2019.43.1.17. Fernando J. Fuentes-García received the PhD in Business Administration in 1997 from
Xiao, M., Wang, R., Chan-Olmsted, S., 2018. Factors affecting YouTube influencer the University of Córdoba UCO. He is currently a Senior Lecturer in Business Organization
marketing credibility: a heuristic-systematic model. J. Media Bus. Stud. 15 (3), at University of Córdoba and the Director of the Chair Extenda for Internationalization as
188–213. https://doi.org/10.1080/16522354.2018.1501146, 2018/07/03. well as the Head of the Official University Master in Commerce and International Trade.
His main lines of research are related to entrepreneurship, business history, agri-food
business and business innovation and internationalization.
16

XXX Impact of The Perceived Risk

Uploaded by

Copyright:

Available Formats

XXX Impact of The Perceived Risk

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

XXX Impact of The Perceived Risk

Uploaded by

Copyright:

Available Formats

Technological Forecasting & Social Change 184 (2022) 121997

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

Impact of the perceived risk in influencers' product recommendations on

1. Introduction influencer marketing seeks to create direct, organic interactions through

Fig. 1. Research model.*

Age (years) 18–25 243 447 690 (72.8 %)

L S C PR PMA Inv A PBC SN Int

L 0.842 0.846 0.904 0.760 0.872

<25 years >25 years β1-β2 p-value Student-t p-value

INV ➔ A 0.259* 0.134* 0.125 0.05 1.985 0.048**

required the adaptation and application of scales primarily drawn from

You might also like