International Journal of Advertising

The Review of Marketing Communications

ISSN: (Print) (Online) Journal homepage: www.tandfonline.com/journals/rina20

Diversity in the digital age: how consumers

respond to diverse virtual influencers

Carla Ferraro, Sean Sands, Nives Zubcevic-Basic & Colin Campbell

To cite this article: Carla Ferraro, Sean Sands, Nives Zubcevic-Basic & Colin Campbell (2024)
Diversity in the digital age: how consumers respond to diverse virtual influencers, International
Journal of Advertising, 43:8, 1342-1365, DOI: 10.1080/02650487.2023.2300927

To link to this article: https://doi.org/10.1080/02650487.2023.2300927

© 2024 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 30 Jan 2024.

Submit your article to this journal

Article views: 7024

View related articles

View Crossmark data

Citing articles: 11 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rina20
International Journal of Advertising
2024, VOL. 43, NO. 8, 1342–1365
https://doi.org/10.1080/02650487.2023.2300927

Diversity in the digital age: how consumers respond to

diverse virtual influencers
Carla Ferraroa, Sean Sandsa , Nives Zubcevic-Basica and Colin Campbellb
a
Department of Management and Marketing, Swinburne University of Technology, Hawthorn, VIC,
Australia; bDepartment of Marketing, University of San Diego School of Business, San Diego, CA, USA

ABSTRACT ARTICLE HISTORY

In recent years the popularity of social media influencers has Received 4 January 2023
grown exponentially, in part because influencers tend to not be Accepted 22 December
seen as advertising and they enable brands to reach engaged 2023
audiences. As such, influencer marketing is perceived as an effec- KEYWORDS
tive and cost-effective marketing tool. However, like all areas of Virtual influencer;
marketing there are broad-based shifts that are having significant influencer marketing;
effects on the domain of influencer marketing. First, there are calls social media; diversity;
for greater levels of diversity and broader representation of diverse generative AI
communities. Second, digital innovation is having a profound
effect on the field of influencer marketing, leading to the advent
of virtual influencers that are artificially created, often through
generative AI. These shifts raise questions about the effect of diver-
sity in influencer marketing – with implications for brands that
engage virtual influencers. In this paper, we present two experi-
ments that investigate the effect of diversity representation in the
context of virtual influencers. Results show that when brands
engage virtual influencers to represent diverse subgroups, novelty
and likeability act to mediate positive effects for the influencer
(word-of-mouth and follow intentions) and the brand (purchase
intentions). Theoretical and practical implications are discussed.

Introduction
Firms are increasing selecting and incentivizing online influencers to engage their
followers on social media to promote firm offerings (Campbell and Farrell 2020; Leung,
Gu, and Palmatier 2022). The subsequent field of influencer marketing has thrived in
recent years, being an effective means whereby influential opinion leaders foster
positive attitudinal and behavioral responses in their followers, spreading messages
about new products, popularizing new trends, and ultimately driving sales (Farrell,
Campbell, and Sands 2022; Jin, Muqaddam, and Ryu 2019; Martínez-López et al. 2020).
While debate exists in relation to the degree of effectiveness of influencers in an
advertising context (Taylor 2020), deploying influencers has become a defining

CONTACT Sean Sands ssands@swin.edu.au Department of Management and Marketing, Swinburne University
of Technology, Hawthorn, VIC, Australia.
© 2024 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives
License (http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction
in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. The terms
on which this article has been published allow the posting of the Accepted Manuscript in a repository by the author(s) or with
their consent.
 International Journal of Advertising 1343

element of social media marketing (De Veirman, Cauberghe, and Hudders 2017; Jin,
Muqaddam, and Ryu 2019). Brand spend on influencer marketing has grown signifi-
cantly in recent years, from $1.7 billion in 2016 to $16.4 billion in 2022 (Celestino 2023).
Within the context of influencer marketing, technological advancements have given
rise to virtual influencers, or social media entities that mimic realistic characteristics,
features, and personalities of human influencers (Sands et al. 2022; Thomas and Fowler
2021). Virtual influencers are created by AI and visually presented as an interactive,
real-time rendered entity in a digital environment (Sands et al. 2022). They can be
managed through AI algorithms and human guidance. Importantly, research suggests
that just as consumers can develop intimate relationships with a human influencer
(De Jans et al. 2020), there is similarly emerging evidence that relationships can be
developed between social media users and virtual influencers (Arsenyan and Mirowska
2021; Mirowska and Arsenyan 2023). Many brands, including Red Bull, Prada, Dior,
Versace, Adidas, and Ikea among others, are embracing virtual influencers as they
offer relatively lower risk and expense yet can command strong followings and effects
(Fitzpatrick and Rothaus 2022). The most well-known virtual influencer is Lil Miquela
– a ‘19-year-old American’ who debuted on Instagram in April 2016 and now has over
2.5 million followers, being recognized as one of Time’s most influential people on
the Internet (Nguyen 2023).
As virtual influencers strive to stand out and connect with target audiences, there
has been a move toward diverse influencers that can enhance perceptions of inclusion
(Hiort 2022). While related, diversity and inclusion are distinct concepts. In the context
of advertising, diversity refers to the portrayal of people with distinct attributes while
inclusion refers to consumer valuation of their presence and perspectives (Eisend,
Muldrow, and Rosengren 2023). There are numerous examples of diverse virtual
influencers, including those that represent diverse gender identities (i.e. @bangkok-
naughtyboo), ethnic groups (i.e. @shudu.gram), and minority groups including people
with Down Syndrome (i.e. @itskamisworld). Indeed, diversity, or the real or perceived
physical or socio-cultural differences among individuals (Arsel, Crockett, and Scott
2022), is an increasingly important consideration across all facets of marketing.
However, limited research exists on diversity in the context of influencer marketing,
and none to the authors’ knowledge, addresses diversity in relation to virtual
influencers.
With this research, we seek better understanding of the role diversity represen-
tation plays in the context of virtual influencer marketing. Our aims are twofold:
first, to investigate how consumers respond to diverse representation in the context
of virtual influencers, and second, to understand the potential mediating effects of
novelty and likability in driving consumer response. Consumer response to novel
stimuli has been shown to positively affect consumer response (Kim, Yoo, and Jeon
2021; Wang et al. 2022) and be an important variable driving advertising response
(Schmidt and Eisend 2015). Similarly, likability has been shown to be an important
factor in determining consumer response to social media influencers (De Veirman,
Cauberghe, and Hudders 2017). The remainder of this paper is organized as follows.
First, we present a review of key literature and develop testable hypotheses. We
next present two studies which test our hypotheses across two different diversity
and brand contexts. We conclude with a general discussion of the findings,
1344 C. FERRARO ET AL.

implications for theory and practice, as well as limitations and suggestions for future
research.

Key literature and hypothesis development

The rise of virtual influencers
The increased popularity of social media and the advancement of computer-generated
imagery have led to the rise of virtual influencers in recent years (Ham et al. 2023;
Sands et al. 2022; Thomas and Fowler 2021). A virtual influencer can be defined as
an entity—humanlike or not—that is created by AI and is visually presented as an
interactive, real-time rendered entity in a digital environment (Sands et al. 2022). They
are typically managed by a team of artists, designers, and animators, often created
using generative AI, and are increasingly being used to promote products, brands,
and services. Recent research suggests that there are over 200 active virtual influencers
on social media (Hiort 2022), with some of the most popular – such as Lil Miquela,
Noonoouri, and Shudu Gram – having millions of followers, and representing brands,
including Nike, Calvin Klein, and Samsung (Sands et al. 2022). Virtual influencers are
a promising asset for brands because they are always available, reliable, and immune
to scandal (Thomas and Fowler 2021). Further, they can provide brands with a con-
sistent stream of content, and they are less likely to get caught up in controversy
than human influencers (Sands et al. 2022). This makes them a safe and reliable choice
for brands that want to avoid negative publicity.

The effect of diversity and inclusion on consumers

Consumers today are more demanding than ever before, and they are looking to
brands, marketers, and social media influencers to do more to embrace diversity and
inclusion (Ferraro, Hemsley, and Sands 2022; Harrison, Thomas, and Cross 2017).
Diversity and inclusion are related, but distinct, concepts that play an important role
in advertising research (Eisend, Muldrow, and Rosengren 2023). Eisend, Muldrow, and
Rosengren (2023) define diversity as the portrayal of people with distinct attributes,
whereas inclusion refers to consumers’ valuation of the presence and perspective of
diverse groups. In recent years, social movements and diversity, equity, and inclusion
(DEI) initiatives have placed mounting pressure on brands to transform their practices
and better reflect the diversity of their audiences (Demsar et al. 2022). As a conse-
quence, there has been an increased focus on diversity in the context of advertising
(Eisend, Muldrow, and Rosengren 2023; Falchetti, Ponchio, and Botelho 2016; Mason
and Pavia 2006; Raun and Christensen-Strynø 2022; Södergren and Vallström 2022).
In the context of social media influencers, diversity reflects the representation of
people from different backgrounds, including race, ethnicity, gender, sexual orientation,
and socioeconomic status (Eisend, Muldrow, and Rosengren 2023). For brands, engag-
ing diverse influencers can have numerous benefits: they can create a more inclusive
and representative online space, help brands reach wider audiences, and build trust
and credibility with consumers. However, the literature remains somewhat limited in
understanding diversity in the context of influencer marketing (Södergren and Vallström
 International Journal of Advertising 1345

2022). While it has been shown that ads featuring diverse models can have mixed
effects, we expect the positive effects to outweigh the negative, given representations
of diversity have been shown to have positive effects (e.g. Geena Davis Institute 2017;
Häfner and Trampe 2009; Micu, Coulter, and Price 2009), such as increased attention
in the context of advertising (e.g. Baxter, Kulczynski, and Ilicic 2016), positive effects
on attitudes toward the model, ad, brand, and product (e.g. Micu, Coulter, and Price
2009) and brand and product associations (e.g. Liljedal, Berg, and Dahlen 2020).

Influencer and brand-oriented outcomes

Prior literature suggests that social media influencers engage in two primary forms
of promotion, self-image building and brand promotion (Li and Peng 2021). Self-image
promotion is an influencer-oriented outcome that is important in order to build suf-
ficiently large, engaged and trusting communities of followers (Hair, Clark, and Shapiro
2010; Hall 2015) in order to attract the interest of relevant brand marketers. For a
social media influencer, two important variables that can build or enhance an influ-
encer’s profile are word-of-mouth and intention to follow (Erkan and Evans 2016;
Gupta, Dash, and Mahajan 2022). In addition to self-image promotion, social media
influencers also engage in brand promotion. For brands influencer are attractive to
marketers who wish to capitalize on opportunities to influence consumers purchase
intention (Weismueller et al. 2020). Purchase intention refers to a consumer’s conscious
effort to make a purchase from a brand (Spears and Singh 2004, p. 56). Importantly,
for social media influencers to be effective, they must engage in both influencer and
brand oriented goals.
In sum, we expect that for a virtual influencer there will be a net positive effect
of diversity, such that when a virtual influencer represents a diverse group there will
be positive effects on both influencer and brand-oriented outcomes. We hypothesize
this formally as:
Hypothesis 1: Diverse (vs. non-diverse) virtual influencers will result in higher influencer
(H1a word-of-mouth, H1b intention to follow) and brand-oriented (H1c purchase intention)
outcomes.

The mediating effect of novelty

When a consumer is confronted with an encounter that is unexpected or unfamiliar
it can be perceived as novel (Wang et al. 2022; Wittmann et al. 2008). Indeed, novelty
is a central component of advertising, typically referred to as fresh, unique, and
unexpected (Ang et al. 2014; Batra, Myers, and Aaker 1996), distinctive, unusual, and
surprising (Pieters, Warlop, and Wedel 2002), or new, different, and interesting (Wang
et al. 2022; Wittmann et al. 2008).
Virtual influencers, particularly those of diverse backgrounds, represent a relatively
new and uncommon phenomenon in digital media. Hence, we expect that a diverse
virtual influencer represents an unexpected or unfamiliar object that is seen as novel
(i.e. new, different, and interesting). When consumers encounter diverse virtual
1346 C. FERRARO ET AL.

influencers, such novel experiences are likely to challenge their pre-existing stereo-
types (Baxter, Kulczynski, and Ilicic 2016; Chu, Lee, and Kim 2019). Specifically, we
expect that exposure to diverse virtual influencers will increase perceptions of novelty
among consumers (Moustakas et al. 2020), particularly given ‘human’ social media
influencers are flooding digital channels, virtual influencers might serve to stand out
from the crowded influencer landscape (Yap 2018).
We anticipate that the novelty effect will positively affect both the influencer and
brand outcomes. Prior research indicates a variety of beneficial impacts associated
with novel stimuli (i.e. Demsar et al. 2022; Hopp and Gangadharbatla 2016) including
consumer response to ads and advertised brands (Sheinin, Varki, and Ashley 2011).
In general, consumers respond to novel stimuli though intensified deliberation (Ajzen,
2002), with novel stimuli thought to be comparatively more difficult to cognitively
categorize (Cox and Locander 1987). This deliberation leads to increased arousal and
exploratory (approach) behavior (Nye, Roth, and Shimp 2008). Novel stimuli have been
shown to positively impact a range of attitudinal and behavioral outcomes (i.e.
word-of-mouth, purchase intention, recall) (Edwards and Gangadharbatla 2001; Sheinin,
Varki, and Ashley 2011; Yim, Cicchirillo, and Drumwright 2012).
We predict that exposure to diverse virtual influencer will be a new, different, and
interesting stimuli that will heighten novelty and that novelty will lead to positive
assessments for the influencer, in terms of word-of-mouth and intention to follow.
We also expect these effects to spill over from the influencer to the brand, in terms
of purchase intention. Specifically, we hypothesize that novelty will act as a mediator
between the presence of a diverse virtual influencer and the associated influencer
outcomes (word-of-mouth and intention to follow) and brand outcomes (purchase
intention). We formally state H2 as follows:
Hypothesis 2: Novelty mediates the effect of a virtual influencer’s diversity on influencer
(H2a word-of-mouth, H2b intention to follow) and brand (H2c purchase intention)
outcomes.

The mediating effect of likability

Likability, or how much a consumer likes or dislikes a stimulus, is an important mea-
sure in advertising (Bergkvist and Rossiter 2008; Smit, Van Meurs, and Neijens 2006),
often affecting persuasion (Reysen 2005). Likeability is a psychological factor that
influences consumers’ reactions to a source (Reysen 2005), is important in breaking
through advertising clutter (Sacharin 2004), and is particularly pertinent in the context
of social media influencers (Breves et al. 2019). Importantly, characteristics of the
influencer can impact consumers perceptions of influencer likeability (De Veirman,
Cauberghe, and Hudders 2017).
In the context of social media influencers, prior research has shown that diversity
can have positive effects on consumer perceptions. For instance, Li (2022) shows that
an Instagram influencer from the marginalized LGBTQ social group makes the audience
as a perceive the influencer more favorably. Nontraditional stereotypes can enhance
the likeability and credibility of characters in ads (Sharma and Bumb 2022). Further,
numerous studies have shown positive effects of diverse representation, with diversity
 International Journal of Advertising 1347

shown to evoke positive emotions and likability (Thomas and Wise 1999; Volante
et al. 2016). We propose that a diverse virtual influencer will result in a positive impact
on likeability for consumers.
When a stimulus is deemed likeable, there are subsequent positive downstream
impacts. For instance, research has shown likability influences consumer response to
the influencer, such as through parasocial interaction which explains part of the
persuasive appeal of social media influencers (Sokolova and Kefi 2020). Likeability
also impacts subsequent consumers’ behavioral intentions (De Veirman, Cauberghe,
and Hudders 2017; Evans et al. 2017), including purchase intentions (Taillon et al.
2020) via a spillover effect between the celebrity/influencer and brand (i.e.
McCracken 1989).
We expect that a diverse virtual influencer will elevate consumer likeability and that
heightened levels of likeability will positively impact influencer and brand outcomes.
Specifically, we predict that likability will act to mediate the relationship between virtual
influencer diversity representation and subsequent influencer (word-of-mouth, follow inten-
tion) and brand (purchase intention) outcomes. This prediction of mediation is in line with
research which shows brands stand to benefit from embracing diversity, which especially
occurs among those in diverse community groups (Eisend, Muldrow, and Rosengren 2023;
Rößner, Gvili, and Eisend 2021). Specifically, we hypothesize H3 as follows:
Hypothesis 3: Likability mediates the effect of a virtual influencer’s diversity on influencer (H3a
word-of-mouth, H3b intention to follow) and brand (H3c purchase intention) outcomes.

Overview of studies
We test our hypotheses across two studies, assessing the relationship between
virtual influencer diversity representation and influencer (word-of-mouth, follow
intention) and brand (purchase intention) outcomes. In addition, we also inves-
tigate the mediating effect of novelty and likability on these relationships. Diverse
representation of a virtual influencers is manipulated via different forms of diver-
sity in each study. In Study 1 we manipulate diversity representation of a virtual
influencer, presenting an influencer with Down Syndrome (diversity condition) or
not (non-diversity condition). For Study 1 we do this in the context of an unknown
brand in order to remove any familiarity effects (Wedel et al. 1998). For Study 2
we provide an intra-study replication (Easley, Madden, and Dunn 2000), whereby
we manipulate an alternative form of diversity (a group of virtual influencers that
vary in terms of age, race, and gender) and test our theory in the context of a
real brand to assess if the effects are consistent. Figure 1 provides an overview
of our conceptual model.

Study 1
Design and procedure
In Study 1 we manipulate a virtual influencer’s diversity representation by manipu-
lating a virtual influencer with Down Syndrome versus a virtual influencer without
1348 C. FERRARO ET AL.

Figure 1. Conceptual model.

Down Syndrome. We manipulate diversity representation using a written description

and visual stimulus (detailed in Appendix 1). The scenario introduces Tammy, a virtual
influencer, and describes her as being an influencer with Down Syndrome (diverse
condition) or not (non-diverse condition). We selected an influencer with Down
Syndrome as individuals with Down Syndrome are a marginalized group on social
media platforms (Dias de Faria and Moreira Casotti 2019) and recent developments
have seen a virtual influencer with Down Syndrome portrayed on social media. The
scenario described Tammy as promoting a local coffee shop. We choose coffee as it
has been shown to be a gender-neutral product category (Crawford et al. 2004) and
we opted for a fictitious brand to remove any potential for known brand associations
(Bergkvist and Rossiter 2008). Respondents were randomly allocated into one of the
conditions, presented with the scenario, and then answered questions pertaining to
the variables in the model (see Appendix 2), with the outcome variables presented
first (Geuens and De Pelsmacker 2017).

Sample, analysis and results

A sample of 200 US respondents were obtained via the Prolific panel, with 11 (5.5%)
removed for failing an attention check item, leaving a final sample of 189 (48.7%
female, Mage = 37.7 years, S.D. = 12.7). This sample size is above the minimum sample
of 80 required for mediation analysis conducted via bootstrapping with medium effect
size (Sim, Kim, and Suh 2022).

Manipulation check. . To check comprehension of the diversity manipulation,

respondents were asked if ‘Tammy is a diverse influencer’, anchored at 1 = strongly
disagree to 7 = strongly agree. Results confirm the manipulation of diversity in
the direction expected (Mnon-diverse = 2.95, S.D. = 1.57, Mdiverse = 5.43, S.D. = 1.60,
F(1, 186) = 114.42, p < .001). Respondents were also asked if the influencer
described was human (1) or AI-generated (7). Results were in the expected
 International Journal of Advertising 1349

direction with respondents in both conditions perceiving Tammy as AI-generated

(Mnon-diverse = 6.28, S.D. = 1.59, Mdiverse = 6.30, S.D. = 1.20, F(1, 186) = 0.47 p > .10).

Effect of diversity representation. We use Hayes (2013) PROCESS macros to test

our hypotheses. Specifically, we test main (H1) and mediating (H2, H3) effects.
We code diversity representation as 0 (non-diverse virtual influencer) or 1
(diverse virtual influencer) and run a parallel mediation model (model 4) to test
the effect of multiple mediators. We run multiple mediation analyses separately
for each dependent variable as suggested by Hayes (2013). All coefficients
reported are bias-corrected, boot-strapped 95% CIs, computed with 5,000
resamples using the PROCESS macro (Hayes 2013, model 4). Results for Study
1 are presented in Table 1.
We first investigate the effect of virtual influencer diversity representation on
influencer (H 1a: word-of-mouth and H1b: intention to follow) and brand outcomes
(H 1c: purchase intention). Results show significant effects for word-of-mouth (H1a:
b = 1.46, SE = 0.20, 95% CI [1.10; 1.87]), intention to follow (H1b: b = 1.42, SE = 0.20,
95% CI [1.04; 1.82]), and purchase intention (H1c: b = 1.27, SE = 0.18, 95% CI [0.93;
1.62]). Figure 2 displays mean values for conditions across the influencer and
brand outcome variables. Taken together, these results support H1. Relative to
those in the non-diverse virtual influencer condition, consumers who were ran-
domly assigned to the diverse virtual influencer condition reported significantly
higher influencer (word-of-mouth and follow intention) and brand (purchase
intention) outcomes.

Mediating effects of novelty and likability. For H2 we investigate the mediating

effect of novelty between diversity representation and influencer (WOM and
intention to follow) and brand (purchase intention) outcomes for virtual influencers.
Specifically, we predict a mediating effect of novelty (M1) for virtual influencer
diversity representation (X) on WOM intention (Y1, H2a), follow intention (Y2, H2b),
and purchase intention (Y3, H2c). Results reveal a significant mediating effect of
novelty between virtual influencer diversity representation on WOM (H2a: b = 1.46,
SE = 0.20, 95% CI [0.44; 1.15]), follow intention (H2b: b = 0.49, SE = 0.13, 95% CI [0.26;
0.77]), and purchase intention (H2c: b = 0.43, SE = 0.12, 95% CI [0.21; 0.70]). Consistent

Table 1. Main and mediation effects for Study 1.

Paths Effect SE LLCI ULCI
Main effects (c-path, total effects)
H1a VI diversity representation → word-of-mouth 1.46 0.20 1.10 1.87
H1b VI diversity representation → follow intention 1.42 0.20 1.04 1.82
H1c VI diversity representation → purchase intention 1.27 0.18 0.93 1.62
Mediation effects (ab-path)
H2a VI diversity representation → novelty → word-of-mouth 1.46 0.20 0.44 1.15
H2b VI diversity representation → novelty → follow intention 0.49 0.13 0.26 0.77
H2c VI diversity representation → novelty → purchase intention 0.43 0.12 0.21 0.70
H3a VI diversity representation → likability → word-of-mouth 0.69 0.18 0.36 1.08
H3b VI diversity representation → likability → follow intention 0.92 0.15 0.65 1.24
H3c VI diversity representation → likability → purchase intention 0.84 0.14 0.57 1.12
Note: VI = virtual influencer; SE = standard error; LLCI = lower limit confidence interval; ULCI = upper limit confidence
interval.
1350 C. FERRARO ET AL.

Figure 2. Significant main effects of virtual influencer diversity representation on influencer (WOM
intention, follow intention) and brand (purchase intention) outcomes in Study 1.

with our theory and supporting H2, consumers exposed to a diverse virtual
influencer reported significantly higher perceived novelty, which in turn was
associated with significantly higher WOM intention, intention to follow, and
purchase intention.
For H3 we investigate the mediating effect of likability between diversity represen-
tation and influencer (WOM and intention to follow) and brand (purchase intention)
outcomes for virtual influencers. Specifically, we predict a mediating effect of likability
(M2) for virtual influencer diversity representation (X) on WOM intention (Y1, H3a),
follow intention (Y2, H3b), and purchase intention (Y3, H3c). Results reveal a significant
mediating effect of novelty between virtual influencer diversity representation on
WOM (H3a: b = 0.69, SE = 0.18, 95% CI [0.36; 1.08]), follow intention (H3b: b = 0.92,
SE = 0.15, 95% CI [0.65; 1.24]), and purchase intention (H3c: b = 0.84, SE = 0.14, 95% CI
[0.57; 1.12]). Consistent with our theory and supporting H3, consumers exposed to a
diverse virtual influencer reported significantly higher perceived likability, which in
turn was associated with significantly higher WOM intention, intention to follow, and
purchase intention.

Study discussion
With Study 1 we investigate the effect of diversity representation within the
context of a virtual influencer with Down Syndrome. Our results show that in
this context virtual influencer diversity representation positively impacts influencer
(word-of-mouth, intention to follow) and brand (purchase intention) outcomes.
We further show that these effects are mediated by novelty and likability. Together,
these results show a significant diversity representation effect for virtual influ-
encers, such that a virtual influencer presented with Down Syndrome is regarded
more positively compared to a virtual influencer that is not. These findings build
on prior research showing the positive effects of diversity representation (Li 2022),
and suggest these effects spill over to virtual contexts.
 International Journal of Advertising 1351

Study 2
Design and procedure
In Study 2 we investigate the same conceptual model presented in Figure 1, in an
alternative diversity context and with a real rather than fictitious brand. In this study
we change the context of diversity representation from a single virtual influencer
representing a diverse group, to a group of virtual influencers that represent a diverse
set of ages, genders, and races. This allows for the effects of diversity representation
to be tested in an alternative diversity context. Further, we change the scenario (see
Appendix 3) from a fictitious brand to a known brand to test if the effects hold when
participants have pre-existing brand associations (Bergkvist and Rossiter 2008).
The scenario describes how Starbucks Coffee is launching a new fall coffee blend
and has engaged four virtual influencers to promote the roast and share it with their
followers on Instagram over the next few months. Our diversity manipulation describes
the four virtual influencers as either similar (non-diverse condition) or different (diverse
condition) in gender, age, and race. We employ the same experimental approach as
in Study 1 with virtual influencers described as either similar (non-diverse condition)
or different (diverse condition) in gender, age, and race.
The same measures as Study 1 are used to assess novelty, likability, purchase
intention, word-of-mouth, and follow intention (see Appendix 2). After providing
consent, respondents were randomly allocated to read one of the two scenarios and
then asked to write a brief description of how they would ‘describe the social media
influencers chosen to represent Starbucks’ new fall product launch’. Written responses
were assessed and used as an attention check, with respondents providing irrelevant
or nonsensical written responses screened out.

Sample, analysis and results

A sample of 235 US respondents were obtained via the Prolific panel, 11 (4.6%) were
removed for failing the attention check, leaving a final sample of 224 (43.3% female,
Mage = 36.74 years, S.D. = 12.11). Again, our sample size is above the minimum sample
required of 80 for mediation analysis conducted via bootstrapping with medium effect
size (Sim, Kim, and Suh 2022).

Manipulation check. To check respondent comprehension of the diversity

manipulation respondents assessed three-items, anchored at 1 = strongly disagree
to 7 = strongly agree, pertaining to diversity of the group of virtual influencers:
they represent a range of cultures (Mnon-diverse = 2.04, S.D. = 1.54, Mdiverse = 5.51,
S.D. = 1.49, F(1, 185) = 241.26, p < .001), they are a diverse group (Mnon-diverse =
1.91, S.D. = 1.46, Mdiverse = 5.60, S.D. = 1.47, F(1, 185) = 293.03, p < .001), and
they represent a range of different age brackets (Mnon-diverse = 1.93, S.D. = 1.48,
Mdiverse = 4.90, S.D. = 1.67, F(1, 185) = 164.81, p < .001). All effects were significant
and in the expected direction. Respondents were also asked if the influencers
were human (1) or AI-generated (7), and results were in the expected direction
with both conditions rating as AI generated (M non-diverse = 6.18, S.D. = 1.59,
Mdiverse = 6.29, S.D. = 1.20, F(1, 185) = 0.37, p > .10).
1352 C. FERRARO ET AL.

Effect of diversity representation (main effects). We use Hayes (2013) PROCESS

macros to test the hypotheses with results for Study 2 and present the results
in Table 2.
We again investigate the effect of virtual influencer diversity representation on
influencer (H1a: word-of-mouth and H1b: intention to follow) and brand outcomes (H1c:
purchase intention). Results are presented in Table 2 and show significant effects for
word-of-mouth (H1a: b = 0.63, SE = 0.18, 95% CI [0.29; 1.00]) and intention to follow
(H1b: b = 0.63, SE = 0.18, 95% CI [0.24; 0.98]). However, there is no significant main
effect for purchase intention (H1c: b = −0.11, SE = 0.19, 95% CI [-0.49; 0.27]). Figure 3
charts mean values for conditions across the outcome variables. Our results from
Study 2 provide mixed support to H1. We find that consumers who were randomly
assigned to the diverse virtual influencer condition reported significantly higher
influencer (word-of-mouth and follow intention). However, we do not find this effect
for the brand outcome of purchase intention.

Mediating effects of novelty and likability. Again, for H2 we investigate the

mediating effect of novelty between diversity representation and influencer (WOM
and intention to follow) and brand (purchase intention) outcomes for virtual
influencers. Results are presented in Table 2 and reveal a significant mediating
effect of novelty between virtual influencer diversity representation on WOM (H2a:
b = 0.50, SE = 0.15, 95% CI [0.24; 0.81]), follow intention (H2b: b = 0.43, SE = 0.12,
95% CI [0.21; 0.68]), and purchase intention (H2c: b = 0.19, SE = 0.08, 95% CI [0.04;
0.37]). Consistent with our theory and supporting H2, Study 2 provides additional
support that consumers exposed to a diverse virtual influencer reported significantly
higher perceived novelty, which in turn was associated with significantly higher
WOM intention, intention to follow, and purchase intention.
For H3 we again investigate the mediating effect of likability between diversity
representation and influencer (WOM and intention to follow) and brand (purchase
intention) outcomes for virtual influencers. Results are presented in Table 2 and reveal
significant mediating effect of novelty between virtual influencer diversity represen-
tation on WOM (H3a: b = 0.13, SE = 0.08, 95% CI [0.01; 0.30]), follow intention (H3b:
b = 0.20, SE = 0.10, 95% CI [0.01; 0.41]), and purchase intention (H3c: b = 0.28, SE = 0.14,
95% CI [0.03; 0.56]). Study 2 provides additional support with our theory and support

Table 2. Main and mediation effects for Study 2.

Paths Effect SE LLCI ULCI
Main effects (c-path, total effects)
H1a VI diversity representation → word-of-mouth 0.63 0.18 0.29 1.00
H1b VI diversity representation → follow intention 0.63 0.18 0.24 0.98
H1c VI diversity representation → purchase intention 0.46 0.17 0.14 0.80
Mediation effects (ab-path)
H2a VI diversity representation → novelty → word-of-mouth 0.50 0.15 0.24 0.81
H2b VI diversity representation → novelty → follow intention .043 0.12 0.21 0.68
H2c VI diversity representation → novelty → purchase intention 0.19 0.08 0.04 0.37
H3a VI diversity representation → likability → word-of-mouth 0.13 0.08 0.01 0.30
H3b VI diversity representation → likability → follow intention 0.20 0.10 0.01 0.41
H3c VI diversity representation → likability → purchase intention 0.28 0.14 0.03 0.56
Note: VI = virtual influencer; SE = standard error; LLCI = lower limit confidence interval; ULCI = upper limit confidence
interval.
 International Journal of Advertising 1353

Figure 3. Significant main effects of virtual influencer diversity representation on influencer (WOM
intention, follow intention) and brand (purchase intention) outcomes in Study 2.

for H3 that consumers exposed to a diverse virtual influencer reported significantly

higher perceived likability, which in turn was associated with significantly higher WOM
intention, intention to follow, and purchase intention.

Study discussion
With Study 2 we conduct an intra-study replication, extending Study 1 in two ways.
First, we investigate a different, and more overt, diversity context whereby a group
of virtual influencers are presented as diverse in age, gender, and race (or not). Second,
we extend Study 2 to a known brand to test if the effects hold in the context of
existing memory traces (Wedel et al. 1998). Our results provide additional support
for the positive effects of diversity in influencer marketing, beyond effects previously
shown such as attention (e.g. Baxter, Kulczynski, and Ilicic 2016), attitudes (e.g. Micu,
Coulter, and Price 2009) and brand associations (e.g. Liljedal, Berg, and Dahlen 2020).
Specifically, our results show virtual influencer diversity representation positively
impacts the influencer outcomes of word-of-mouth and intention to follow. While we
do not find an effect for purchase intention, it is possible that existing brand asso-
ciations may have played a role (Bergkvist and Rossiter 2008). Consistent with our
findings in Study 1 we also show that these effects are mediated by novelty and
likability. Taken together, our results are in line with prior research that shows robots
can be imbued with perceived traits, such as diversity, that lead to positive outcomes.
For instance, emotional contagion can occur not only when humans smile, but also
when a robot smiles (Hennig-Thurau et al. 2006).

General discussion
We show that diversity representation can have positive effects in the context of
virtual influencers. Our results are novel in two ways: first, we show the effect of
diversity representation in virtual settings, in contrast to prior research conducted in
the physical world (i.e. Park, Voss, and Voss 2023). Second, we extend the notion of
1354 C. FERRARO ET AL.

diversity from traditional representation of age, race, and gender to include other
important minority groups (Södergren, Hietanen, and Vallström 2023; Södergren and
Vallström 2022), in our case an influencer with Down Syndrome. We test our hypoth-
eses across two studies. First, testing the direct effect of virtual influencer diversity
representation on influencer (word-of-mouth, follower intentions) and brand (purchase
intention) outcomes. We find consistent support for these predicted effects for the
influencer outcomes of word-of-mouth and follow intention. However, we find mixed
support for the brand outcome of purchase intention across our two studies.
Specifically, in Study 2 we suspect that we did not find a main effect for purchase
intention because of the use of a known brand, which comes with a number of
implicit associations in a consumer’s mind (Bergkvist and Rossiter 2008). We did,
however, find support across both studies for the predicted mediating effect of novelty
between diversity representation and both the influencer and brand outcomes. Further,
we find support for the predicted mediating effect of likability. These effects hold
across our two studies despite them varying in terms of the presentation of diversity
and whether an unknown (with no existing associations) or known brand (which
comes with a number of implicit associations in a consumer’s mind, i.e. Bergkvist and
Rossiter 2008). Together, our findings enrich advertising knowledge by providing
evidence that diversity representation can increase consumer evaluations of an influ-
encer in a virtual, or artificial context.

Contributions to theory
With this research, we make three key contributions to theory. First, we advance the
emerging body of research considering diversity in advertising (Eisend, Muldrow, and
Rosengren 2023; Falchetti, Ponchio, and Botelho 2016; Södergren, Hietanen, and
Vallström 2023) and more specifically in the domain of influencer marketing (Ellis
et al. 2021; Raun and Christensen-Strynø 2022). While prior research has shown that
diverse ads tend to be more salient and appealing for individuals in the target diverse
group (i.e. Perkins, Thomas, and Taylor 2000; Thomas and Wise 1999), with this research
we demonstrate these positive effects with a broad sample of consumers, beyond
those within a specific diverse group. These effects may be explained with the novel
stimulus concept (Langer et al. 1976), whereby schema-incongruent information can
impact viewer response to the influencer and their content. In essence, diversity
representation could act as a means of schema-incongruency (Wyer and Gordon 1982)
in influencer marketing and lead to positive outcomes.
Second, this research extends the focus of diversity in advertising from traditional
media to online media (Eisend, Muldrow, and Rosengren 2023). Research has only
recently begun to investigate diversity in online ad contexts, for example, investigating
consumer reactions to advertising campaigns (e.g. Feng, Chen, and He 2019) and
brand collaborations with influencers who take a stand on diversity of similar causes
(Yang, Chuenterawong, and Pugdeethosapol 2021). However, many more questions
remain unanswered. For instance, there is a need to better understand how positive
and negative reactions are likely to multiply via comments and sharing of advertising
content social media channels.
 International Journal of Advertising 1355

Third, our findings extend prior research in the context of human-computer inter-
actions, which has shown that artificial agents with human likeness (looks or acts)
can engender positive responses from users (Siegel, Breazeal, and Norton 2009).
Specifically, we find the positive effects of diversity representation are present for
virtual influencers as well as humans. This means that despite consumers being aware
that a virtual influencer is created by AI, the positive halo effect of diversity repre-
sentation spills-over to a virtual entity. This contribution addresses calls to investigate
and understand differences between virtual and human influencers (Thomas and
Fowler 2021). Such knowledge is important to enable marketers to understand con-
sumer perceptions of virtual entities –which will likely be an increasingly common
phenomenon in a range of settings, including social media, services, and the metaverse.

Implications for practice

Our findings have important implications for marketers, brand managers and adver-
tisers. First, we show that diversity representation is an important consideration for
engaging a virtual influencer and show that a virtual influencer that represents a
diverse group can act to gain audience attention. This is important given that prior
research highlights that consumers tend to have a lack advertising focus in online
environments (Cho and Cheon 2004, Danaher and Mullarkey 2003). Taken together,
our findings suggest diversity representation in social media and brand campaigns
may be an effective way to connect with audiences and influencer purchase intentions.
It will be important for advertisers to consider how intersectionality could be incor-
porated into social media contexts (Gopaldas and DeRoy 2015). Second, our research
provides valuable insights for brands and advertisers considering the use of virtual
influencers. While prior research highlights opportunities and challenges for imple-
menting virtual influencers in social platforms (Sands et al. 2022), our findings reveal
a more nuanced understanding. Specifically showing that a virtual influencer can be
imbued with other human traits – in our case diversity representation, which can
result in more positive perception. Finally, our findings have broader implications for
advertising practitioners considering the use of synthetic advertising (Campbell et al.
2022), or deepfakes in ads. Our findings suggest that consumers might attribute
similar positive effects to virtual – or deepfake models – in advertising contexts.
Advertisers interested in synthetic advertising should be encouraged by these results,
albeit noting that they may be context, category, or brand dependent.

Limitations and future research directions

Our research, like all, is subject to limitations which lead to opportunities for future
research. First, our research should be understood in the context of its use of hypo-
thetical experiments. While experimental research is common in marketing and psy-
chology research, and our manipulation checks confirm consumer interpretation of
our scenarios, we encourage researchers to extend this research to more realistic
contexts. For instance, as AI technology becomes more readily available, we expect
1356 C. FERRARO ET AL.

that researchers may be able to create AI influencers and insert them in social media
channels in order to assess their efficacy in real world contexts. Our experimental
stimuli also have shortcomings. In Study 1 our stimuli are not perfectly equivalent.
However, arguably it is near impossible to achieve equivalence when comparing two
different model types based on a diversity element such as an influencer with down
syndrome. Further, Study 1 presents a potential confounding factor, as Downs
Syndrome could be contextualized as representing the special needs category perhaps
even better than it does diversity. Perhaps otherwise, the influencer would represent
the cultural default and thus not as diverse as someone else with special needs may
be. Study 2 also has potential confounding factors we must acknowledge. For instance,
in Study 2, we use images of real people and prime respondents that they are virtual
influencers. The decision to use real human images was influenced by two consider-
ations. Firstly, using recognizable virtual influencers might have introduced bias, as
respondents could have pre-existing perceptions about these figures. Secondly, finding
a diverse range of virtual influencers that matched our study criteria proved to be
challenging. However, future research might seek to develop custom virtual influencers
in experiments. Across both our experiments, we also acknowledge that it is difficult
to capture diversity with a single individual. Thus, the findings from our studies need
to be framed against these limitations.
Second, our study is limited in terms of the set of outcomes and mediators we
investigate. No doubt diversity in social media advertising almost certainly has effects
that go beyond our focus. Future work should examine additional moderators and
mediators that may influence the relationship between diversity representation and
even extend to investigate brand associations and outcomes in the context of virtual
influencer marketing. Future research might also consider the variables we examine
in different ways. For instance, while we study the effect of diverse virtual influencers
on novelty, there are important considerations that remain. Consumer reactions to
novel stimuli can be short-lived, with repeated exposure leading to adaptation and
decreased effects (Helson 1964). Hence, the short-term reactions to novel stimuli – in
our case diverse virtual influencers – may lose strength in the long-run (i.e. Brüggen,
Foubert, and Gremler 2011).
Third, our study investigates the effect of diverse virtual influencers on a general
population of consumers. However, prior research suggests ad response differs between
general consumers and specific subgroup (Descubes, McNamara, and Bryson 2018;
Eisend and Hermann, 2019). For instance, lesbians are known to respond more favor-
ably to lesbian imagery, but less favorably to imagery of gay men (Descubes,
McNamara, and Bryson 2018). This suggests that LGBTQIA + consumers are not always
homogenous in reaction to advertising that targets them. Future research should
investigate the specific effects of diverse representation on the subgroup being
represented.
Finally, there are limitations in terms of brand effects studied. In our studies we
tested effects for an unknown brand (Study 1) and a known (Study 2) brand, being
Starbucks. While research suggests that most consumers (92%) in the US are aware
of the Starbuck’s brand (Kunst, 2023), we do not control for brand awareness or any
brand preference. It is possible that these, or other factors, may act as confounds
and may be a reason why we did not find an effect for purchase intention.
 International Journal of Advertising 1357

Disclosure statement
To the best of the authors’ knowledge, no conflicts of interest exist.

ORCID
Sean Sands http://orcid.org/0000-0001-9192-3676
Colin Campbell http://orcid.org/0000-0002-6218-0866

Data availability statement

The data that support the findings of this study are available from the corresponding author,
upon reasonable request.

References
Ajzen, I. 2002. Perceived behavioral control, selfefficacy, locus of control, and the theory of
planned behavior 1. Journal of applied social psychology 32, no. 4: 665–683.
Ang, S.H., S.M. Leong, Y.H. Lee, and S.L. Lou. 2014. Necessary but not sufficient: Beyond nov-
elty in advertising creativity. Journal of Marketing Communications 20, no. 3: 214–30.
Arsel, Z., D. Crockett, and M.L. Scott. 2022. Diversity, equity, and inclusion (DEI) in the journal
of consumer research: A curation and research agenda. Journal of Consumer Research 48, no.
5: 920–33.
Arsenyan, J., and A. Mirowska. 2021. Almost human? A comparative case study on the social
media presence of virtual influencers. International Journal of Human-Computer Studies 155:
102694.
Batra, R., J.G. Myers, and D.A. Aaker. 1996. Advertising management. 5th ed. Englewood Cliffs,
NJ: Prentice-Hall.
Baxter, S.M., A. Kulczynski, and J. Ilicic. 2016. Ads aimed at dads: Exploring consumers’ reactions
towards advertising that conforms and challenges traditional gender role ideologies.
International Journal of Advertising 35, no. 6: 970–82.
Bergkvist, L., and J.R. Rossiter. 2008. The role of ad likability in predicting an ad’s campaign
performance. Journal of Advertising 37, no. 2: 85–98.
Breves, P.L., N. Liebers, M. Abt, and A. Kunze. 2019. The perceived fit between Instagram influ-
encers and the endorsed brand: How influencer–brand fit affects source credibility and
persuasive effectiveness. Journal of Advertising Research 59, no. 4: 440–54.
Brüggen, E.C., B. Foubert, and D.D. Gremler. 2011. Extreme makeover: Short-and long-term
effects of a remodeled servicescape. Journal of Marketing 75, no. 5: 71–87.
Campbell, C., and J.R. Farrell. 2020. More than meets the eye: The functional components
underlying influencer marketing. Business Horizons 63, no. 4: 469–79.
Campbell, C., K. Plangger, S. Sands, and J. Kietzmann. 2022. Preparing for an era of deepfakes
and AI-generated ads: A framework for understanding responses to manipulated advertising.
Journal of Advertising 51, no. 1: 22–38.
Celestino, A. 2023. Influencer marketing and FTC regulations. Forbes. https://www.forbes.com/sites/
forbesagencycouncil/2020/12/02/influencer-marketing-and-ftc-regulations/?sh=42a3f7ed1566
Chaiken, S. 1980. Heuristic versus systematic information processing and the use of source ver-
sus message cues in persuasion. Journal of Personality and Social Psychology 39, no. 5: 752–66.
Cho, C.-H, and Cheon HJ. 2004. Why do people avoid advertising on the internet? Journal of
Advertising 33, no. 4: 89–97.
Chu, K., D.H. Lee, and J.Y. Kim. 2019. The effect of verbal brand personification on consumer
evaluation in advertising: Internal and external personification. Journal of Business Research
99: 472–80.
1358 C. FERRARO ET AL.

Cox, D.S, and W.B. Locander. 1987. Product novelty: Does it moderate the relationship between
ad attitudes and brand attitudes? Journal of Advertising 16, no. 3: 39–44.
Crawford, J.T., P.A. Leynes, C.B. Mayhorn, and M.L. Bink. 2004. Champagne, beer, or coffee? A
corpus of gender-related and neutral words. Behavior Research Methods, Instruments, &
Computers: A Journal of the Psychonomic Society, Inc 36, no. 3: 444–58.
Danaher, P.J, and G.W. Mullarkey. 2003. Factors affecting online advertising recall: A study of
students. Journal of Advertising Research 43, no. 3: 252–67.
De Jans, S., D. Van de Sompel, M. De Veirman, and L. Hudders. 2020. #Sponsored! how the
recognition of sponsoring on Instagram posts affects adolescents’ brand evaluations through
source evaluations. Computers in Human Behavior 109: 106342.
De Veirman, M., V. Cauberghe, and L. Hudders. 2017. Marketing through Instagram influencers:
The impact of number of followers and product divergence on brand attitude. International
Journal of Advertising 36, no. 5: 798–828.
Demsar, V., S. Sands, S. Rosengren, and C. Campbell. 2022. Ad creativity in a negative context:
How a thanking message frame enhances purchase intention in times of crisis. Journal of
Retailing and Consumer Services 64: 102825.
Descubes, I., T. McNamara, and D. Bryson. 2018. Lesbians’ assessments of gay advertising in
France: Not necessarily a case of ‘La vie en rose? Journal of Marketing Management 34, no.
7–8: 639–63.
Dias de Faria, M., and L. Moreira Casotti. 2019. “Welcome to holland!” people with down syn-
drome as vulnerable consumers. European Journal of Marketing 53, no. 11: 2245–67.
Easley, R.W., C.S. Madden, and M.G. Dunn. 2000. Conducting marketing science: The role of
replication in the research process. Journal of Business Research 48, no. 1: 83–92.
Edwards, S.M, and H. Gangadharbatla. 2001. The novelty of 3D product presentations online.
Journal of Interactive Advertising 2, no. 1: 10–8.
Eisend, M., and E. Hermann. 2019. Consumer responses to homosexual imagery in advertising:
A meta-analysis. Journal of Advertising 48, no.4: 380–400.
Eisend, M., A.F. Muldrow, and S. Rosengren. 2023. Diversity and inclusion in advertising research.
International Journal of Advertising 42, no. 1: 52–9.
Ellis, K., K. Hall, B. Haller, J. Loebner, C. Mallon, and E. Timke. 2021. Roundtable on disability
and advertising, part I. Advertising & Society Quarterly 22, no. 1. https://muse.jhu.edu/
article/788616
Erkan, I., and C. Evans. 2016. The influence of eWOM in social media on consumers’ purchase
intentions: An extended approach to information adoption. Computers in Human Behavior
61: 47–55.
Evans, N.J., J. Phua, J. Lim, and H. Jun. 2017. Disclosing Instagram influencer advertising: The
effects of disclosure language on advertising recognition, attitudes, and behavioral intent.
Journal of Interactive Advertising 17, no. 2: 138–49.
Falchetti, C., M.C. Ponchio, and N.L.P. Botelho. 2016. Understanding the vulnerability of blind
consumers: Adaptation in the marketplace, personal traits and coping strategies. Journal of
Marketing Management 32, no. 3–4: 313–34.
Farrell, J.R., C. Campbell, and S. Sands. 2022. What drives consumers to engage with influenc-
ers? Segmenting consumer response to influencers: Insights for managing social-media re-
lationships. Journal of Advertising Research 62, no. 1: 35–48.
Feng, Y., H. Chen, and L. He. 2019. Consumer responses to femvertising: A data-mining case
of dove’s “campaign for real beauty” on YouTube. Journal of Advertising 48, no. 3: 292–301.
Ferraro, C., A. Hemsley, and S. Sands. 2022. Embracing diversity, equity, and inclusion (DEI):
considerations and opportunities for brand managers. Business Horizons 66, no. 4: 463–79.
Fitzpatrick, A., and S. Rothaus. 2022. Avoiding Metaverse Mayhem: Regulating Virtual Influencers
as Nearly Human—PRNEWS. PR News. https://www.prnewsonline.com/virtual-influencer
s-metaverse-regulation/
Geena Davis Institute. 2017. Gender Bias in Advertising. Geena Davis Institute on Gender in
Media. https://seejane.org/research-informs-empowers/gender-bias-advertising/
 International Journal of Advertising 1359

Geuens, M., and P. De Pelsmacker. 2017. Planning and conducting experimental advertising
research and questionnaire design. Journal of Advertising 46, no. 1: 83–100.
Gopaldas, A., and G. DeRoy. 2015. An intersectional approach to diversity research. Consumption
Markets & Culture 18, no. 4: 333–64.
Gratch, J., N. Wang, J. Gerten, E. Fast, and R. Duffy. 2007. Creating rapport with virtual agents.
In Intelligent virtual agents, ed. C.Pelachaud, J.-C. Martin, E. André, G. Chollet, K. Karpouzis,
and D. Pelé, 7th International Conference, IVA 2007 Paris, France, September 17-19, Proceedings.
(pp. 125-138). Springer Berlin Heidelberg.
Gupta, S., S.B. Dash, and R. Mahajan. 2022. The role of social influencers for effective public
health communication. Online Information Review 46, no. 5: 974–92.
Häfner, M., and D. Trampe. 2009. When thinking is beneficial and when it is not: The effects
of thin and round advertising models. Journal of Consumer Psychology 19, no. 4: 619–28.
Hair, N., M. Clark, and M. Shapiro. 2010. Toward a classification system of relational activity in
consumer electronic communities: The moderators’ tale. Journal of Relationship Marketing 9,
no. 1: 54–65.
Hall, J. 2015. Build authentic audience experiences through influencer marketing. https://www.
forbes.com/sites/johnhall/2015/12/17/build-authentic-audience-experiences-through-influencer-
marketing/#1cef3d684ff2.
Ham, J., S. Li, P. Shah, and M.S. Eastin. 2023. The “mixed” reality of virtual brand endorsers:
Understanding the effect of brand engagement and social cues on technological perceptions
and advertising effectiveness. Journal of Interactive Advertising 23, no. 2: 98–113.
Harrison, R.L., K.D. Thomas, and S.N. Cross. 2017. Restricted visions of multiracial identity in
advertising. Journal of Advertising 46, no. 4: 503–20.
Hayes, A.F. 2013. Mediation, moderation, and conditional process analysis. Introduction to
Mediation, Moderation, and Conditional Process Analysis: A Regression-Based Approach 1: 20.
Helson, H. 1964. Current trends and issues in adaptation-level theory. American Psychologist 19,
no. 1: 26–38.
Hennig-Thurau, T., M. Groth, M. Paul, and D.D. Gremler. 2006. Are all smiles created equal? How
emotional contagion and emotional labor affect service relationships. Journal of Marketing
70, no. 3: 58–73.
Hiort, A. 2022. Why Diversity and Inclusivity in Virtual Influencers Matter. VirtualHumans.Org.
https://www.virtualhumans.org/article/why-diversity-and-inclusivity-in-virtual-influencers-matter
Hopp, T., and H. Gangadharbatla. 2016. Novelty effects in augmented reality advertising envi-
ronments: The influence of exposure time and self-efficacy. Journal of Current Issues & Research
in Advertising 37, no. 2: 113–30.
Jin, S.V., A. Muqaddam, and E. Ryu. 2019. Instafamous and social media influencer marketing.
Marketing Intelligence & Planning 37, no. 5: 567–79.
Kim, S.H., S.R. Yoo, and H.M. Jeon. 2021. The role of experiential value, novelty, and satisfaction
in robot barista coffee shop in South Korea: COVID-19 crisis and beyond. Service Business 16,
no. 3: 771–90.
Kunst, A. 2023. Starbucks brand profile in the United States 2022. Statista. Accessed: 15
November 2023, Available at: https://www.statista.com/forecasts/1335789/starbucks-restauran
t-chains-brand-profile-in-theunited-states
Langer, E.J., S. Fiske, S.E. Taylor, and B. Chanowitz. 1976. Stigma, staring, and discomfort: A
novel-stimulus hypothesis. Journal of Experimental Social Psychology 12, no. 5: 451–63.
Lee, J.E, and B. Watkins. 2016. YouTube vloggers’ influence on consumer luxury brand percep-
tions and intentions. Journal of Business Research 69, no. 12: 5753–60.
Leung, F.F., F.F. Gu, and R.W. Palmatier. 2022. Online influencer marketing. Journal of the Academy
of Marketing Science 50, no. 2: 226–51.
Li, M. 2022. Influence for social good: Exploring the roles of influencer identity and comment
section in Instagram-based LGBTQ-centric corporate social responsibility advertising.
International Journal of Advertising 41, no. 3: 462–99.
Li, Y., and Y. Peng. 2021. Influencer marketing: Purchase intention and its antecedents. Marketing
Intelligence & Planning 39, no. 7: 960–78.
1360 C. FERRARO ET AL.

Liljedal, K.T., H. Berg, and M. Dahlen. 2020. Effects of nonstereotyped occupational gender role
portrayal in advertising: How showing women in Male-Stereotyped job roles sends positive
signals about brands. Journal of Advertising Research 60, no. 2: 179–96.
Mai, E.S., D.L. Haytko, and B.J. Taillon. 2022. How advertisements mixing black and white actors
affect consumer intent: Perceived authenticity can strengthen responses to interracial adver-
tising. Journal of Advertising Research 62, no. 3: 252–70.
Martínez-López, F.J., R. Anaya-Sánchez, M. Fernández Giordano, and D. Lopez-Lopez. 2020.
Behind influencer marketing: Key marketing decisions and their effects on followers’ re-
sponses. Journal of Marketing Management 36, no. 7–8: 579–607.
Mason, M., and T. Pavia. 2006. When the family system includes disability: Adaptation in
the marketplace, roles and identity. Journal of Marketing Management 22, no. 9–10:
1009–30.
McCracken, G. 1989. Who is the celebrity endorser? Cultural foundations of the endorsement
process. Journal of Consumer Research 16, no. 3: 310–21.
Micu, C.C., R.A. Coulter, and L.L. Price. 2009. How product trial alters the effects of model at-
tractiveness. Journal of Advertising 38, no. 2: 69–82.
Mirowska, A., and J. Arsenyan. 2023. Sweet escape: The role of empathy in social media en-
gagement with human versus virtual influencers. International Journal of Human-Computer
Studies 174: 103008.
Moustakas, E., N. Lamba, D. Mahmoud, and C. Ranganathan. 2020. Blurring lines between fiction
and reality: Perspectives of experts on marketing effectiveness of virtual influencers. 2020
International Conference on Cyber Security and Protection of Digital Services (Cyber Security), 1–6. IEEE.
Nguyen, M. 2023. Meet the virtual influencers trying to sell you stuff. InDaily. https://indaily.
com.au/opinion/2023/09/20/meet-the -vir tual-influencers-tr ying-to-sell-you-
stuff/#:~:text=Lil%20Miquela%20has%20been%20named,and%20marketers%2C%20writes%20
Mai%20Nguyen.
Nye, C.W., M.S. Roth, and T.A. Shimp. 2008. Comparative advertising in markets where brands
and comparative advertising are novel. Journal of International Business Studies 39, no. 5:
851–63.
Park, Y.W., G.B. Voss, and Z.G. Voss. 2023. Advancing customer diversity, equity, and inclusion:
Measurement, stakeholder influence, and the role of marketing. Journal of the Academy of
Marketing Science 51, no. 1: 174–97.
Perkins, L.A., K.M. Thomas, and G.A. Taylor. 2000. Advertising and recruitment: Marketing to
minorities. Psychology & Marketing 17, no. 3: 235–55.
Pieters, R., L. Warlop, and M. Wedel. 2002. Breaking through the clutter: Benefits of advertise-
ment originality and familiarity for brand attention and memory. Management Science 48,
no. 6: 765–81.
Raun, T., and M.B. Christensen-Strynø. 2022. ‘We belong to something beautiful’: Julie vu’s and
madeline stuart’s use of minority identity as a popular feminist self-branding strategy on
Instagram. Information, Communication & Society 25, no. 12: 1790–807.
Reysen, S. 2005. Construction of a new scale: The Reysen likability scale. Social Behavior and
Personality: An International Journal 33, no. 2: 201–8.
Rößner, A., Y. Gvili, and M. Eisend. 2021. Explaining consumer responses to ethnic and religious
minorities in advertising: The case of Israel and Germany. Journal of Advertising 50, no. 4: 391–407.
Sacharin, K. 2004. Attention!: How to interrupt, yell, whisper, and touch consumers. John Wiley &
Sons.
Sands, S., C. Ferraro, V. Demsar, and G. Chandler. 2022. False idols: Unpacking the opportunities and
challenges of falsity in the context of virtual influencers. Business Horizons 65, no. 6: 777–88.
Schmidt, S., and M. Eisend. 2015. Advertising repetition: A meta-analysis on effective frequen-
cy in advertising. Journal of Advertising 44, no. 4: 415–28.
Sharma, S., and A. Bumb. 2022. Femluencing: Integration of femvertising and influencer mar-
keting on social media. Journal of Interactive Advertising 22, no. 1: 95–111.
Sheinin, D.A., S. Varki, and C. Ashley. 2011. The differential effect of ad novelty and message
usefulness on brand judgments. Journal of Advertising 40, no. 3: 5–18.
 International Journal of Advertising 1361

Siegel, M., C. Breazeal, and M.I. Norton. 2009. Persuasive Robotics: The influence of robot gen-
der on human behavior. 2009 IEEE/RSJ International Conference on Intelligent Robots and
Systems, 2563–2568.
Sim, M., S.Y. Kim, and Y. Suh. 2022. Sample size requirements for simple and complex mediation
models. Educational and Psychological Measurement 82, no. 1: 76–106.
Smit, E.G., L. Van Meurs, and P.C. Neijens. 2006. Effects of advertising likeability: A 10-year
perspective. Journal of Advertising Research 46, no. 1: 73–83.
Södergren, J., J. Hietanen, and N. Vallström. 2023. Tales from the crypt: A psychoanalytic ap-
proach to disability representation in advertising. Journal of Consumer Culture 23, no. 4:
747–768.
Södergren, J., and N. Vallström. 2022. Disability in influencer marketing: A complex model of
disability representation. Journal of Marketing Management 39, no. 11–12: 1012–42.
Sokolova, K., and H. Kefi. 2020. Instagram and YouTube bloggers promote it, why should I buy?
How credibility and parasocial interaction influence purchase intentions. Journal of Retailing
and Consumer Services 53: 101742.
Spears, N., and S. Singh. 2004. Measuring attitude toward the brand and purchase intentions.
Journal of Current Issues & Research in Advertising 26, no. 2: 53–66.
Statista. Statista. 2019. Global adaptive apparel market size 2019-2025. https://www.statista.
com/statistics/875613/global-adaptive-apparel-market-size/
Taillon, B.J., S.M. Mueller, C.M. Kowalczyk, and D.N. Jones. 2020. Understanding the relationships
between social media influencers and their followers: The moderating role of closeness.
Journal of Product & Brand Management 29, no. 6: 767–82.
Taylor, C.R. 2020. The urgent need for more research on influencer marketing. International
Journal of Advertising 39, no. 7: 889–91.
Thomas, K.M, and P.G. Wise. 1999. Organizational attractiveness and individual differences: Are diverse
applicants attracted by different factors? Journal of Business and Psychology 13, no. 3: 375–90.
Thomas, V.L, and K. Fowler. 2021. Close encounters of the AI kind: Use of AI influencers as
brand endorsers. Journal of Advertising 50, no. 1: 11–25.
Volante, M., S.V. Babu, H. Chaturvedi, N. Newsome, E. Ebrahimi, T. Roy, S.B. Daily, and T. Fasolino.
2016. Effects of virtual human appearance fidelity on emotion contagion in affective
inter-personal simulations. IEEE Transactions on Visualization and Computer Graphics 22, no.
4: 1326–35.
Wang, Y., Q. Kang, S. Zhou, Y. Dong, and J. Liu. 2022. The impact of service robots in retail:
Exploring the effect of novelty priming on consumer behavior. Journal of Retailing and
Consumer Services 68: 103002.
Wittmann, B.C., N.D. Daw, B. Seymour, and R.J. Dolan. 2008. Striatal activity underlies novelty-based
choice in humans. Neuron 58, no. 6: 967–73.
Wedel, M., M. Vriens, T.H. Bijmolt, W. Krijnen, and P.S. Leeflang. 1998. Assessing the effects of
abstract attributes and brand familiarity in conjoint choice experiments. International Journal
of Research in Marketing 15, no. 1: 71–8.
Weismueller, J., P. Harrigan, S. Wang, and G.N. Soutar. 2020. Influencer endorsements: How
advertising disclosure and source credibility affect consumer purchase intention on social
media. Australasian Marketing Journal 28, no. 4: 160–70.
Wyer, R.S, and S.E. Gordon. 1982. The recall of information about persons and groups. Journal
of Experimental Social Psychology 18, no. 2: 128–64.
Yang, J., P. Chuenterawong, and K. Pugdeethosapol. 2021. Speaking up on black lives matter:
A comparative study of consumer reactions toward brand and influencer-generated corporate
social responsibility messages. Journal of Advertising 50, no. 5: 565–83.
Yap, L. 2018. The case for virtual influencers. Luxury Society. https://www.luxurysociety.com/
en/articles/2018/09/case-virtual-influencers.
Yim, M.Y.C., V.J. Cicchirillo, and M.E. Drumwright. 2012. The impact of stereoscopic
three-dimensional (3-D) advertising. Journal of Advertising 41, no. 2: 113–28.
1362 C. FERRARO ET AL.

Appendix 1. Study 1 scenario manipulation of virtual influencer

diversity representation
Condition Written description Visual stimuli
Non-diverse [Page 1] Please read this information [Page 2] Below is a typical
about a social media influencer posts from Tammy.
called Tammy. Tammy has been a
social media influencer for 5 years and
has a relatively large following (a few
hundred thousand followers). As an
influencer, Tammy promotes a range of
products and works with many top
brands. In exchange she is offered a lot
of free products and experiences to
share with followers. Recently she has
taken on an influencer engagement for
a local coffee shop, St Ali Coffee. St Ali
Coffee are launching a new blend of
coffee roast and Tammy will be creating
a series of posts about the new blend
to share with her followers over the
next few months. What’s unique about
Tammy is that she is a virtual influencer,
or a virtual persona/ model that is a
computer-generated fictional character.
On the next page you will see a
typical post from Tammy.
Diverse [Page 1] Please read this information [Page 2] Below is a typical
about a social media influencer posts from Tammy.
called Tammy. Tammy has been a
social media influencer for 5 years and
has a relatively large following (a few
hundred thousand followers). As an
influencer, Tammy promotes a range of
products and works with many top
brands. In exchange she is offered a lot
of free products and experiences to
share with followers. Recently she has
taken on an influencer engagement
opportunity with a local coffee shop, St
Ali Coffee. St Ali Coffee are launching a
new blend of coffee roast and Tammy
will be creating a series of posts about
the new blend to share with her
followers over the next few months.
One of the unique aspects about
Tammy is that she has Down Syndrome.
In being an influencer with Down
Syndrome, Tammy hopes to create
greater inclusion in social media. Tammy
is also a virtual influencer, or a virtual
persona/ model that is a
computer-generated fictional character.
On the next page you will see a
typical post from Tammy.
 International Journal of Advertising 1363

Appendix 2. Scale items and Cronbach’s alpha

Cronbach’s alpha
Construct Scale items Study 1 Study 2
Purchase intention1 (Mai, If you could, how likely would .98 .98
Haytko, and Taillon 2022) you be to purchase from the
brand?
Never (1) vs definitely (7)
Definitely do not intend (1) vs
definitely intend (7)
Very low interest (1) vs high
interest (7)
Definitely not (1) vs definitely
(7)
Probably not (1) vs probably (7)
Likability (Chaiken 1980)
1
Would you find the influencer: .96 .95
Cold (1) vs warm (7)
Unlikable (1) vs likable (7)
Unfriendly (1) vs friendly (7)
Novelty (Sheinin, Varki, and
2
Based on what I know about .92 .92
Ashley 2011) the social media influencer, I
imagine they would be…
Original
Different form my expectations
about other influencers
Memorable
Interesting
Different
Word-of-mouth3 If you could, how likely would – –
you be to tell others about
this influencer?
Follow intention3 If you could, how likely would – –
you be to follow this
influencer on social media?
Notes: 1Measured on a 7-point bipolar scale; 2Measured on a 7-point Likert scale, anchored at 1 = ‘strongly disagree’
vs 7 = ‘strongly agree’; 3Measured on a 7-point Likert scale, anchored at 1 = ‘extremely unlikely’ vs 7 = ‘extremely
likely’.
1364 C. FERRARO ET AL.

Appendix 3. Study 2 scenario manipulation of virtual influencer

diversity representation.
Condition Written description Visual stimuli
Non-diverse [Page 1] Please read the below [Page 2] Below are some typical posts
information about Starbucks and their from each of these virtual influencers.
plans to engage social media
influencers to represent the brand on
Instagram. Starbucks is the world’s largest
coffeehouse, operating a multinational chain
of coffeehouses and roastery reserves.
Starbucks serves hot and cold drinks,
whole-bean coffee, micro-ground instant
coffee, espresso, caffe latte, full and
loose-leaf teas, juices, Frappuccino
beverages, pastries, and snacks. Starbucks is
about to launch a new fall coffee blend and
have engaged four social media influencers
to promote the roast and share it with their
followers on Instagram over the next few
months: Michelle (24-year-old white female),
Mary (22-year-old white female), Linda
(24-year-old white female), and Sue
(22-year-old white female).
Each influencer Starbucks has chosen to
represent the brand has been a social media
influencer for a few years and have relatively
large followings (a few hundred thousand
followers). They each promote a range of
products and work with many top brands. In
exchange, they are offered free products and
experiences to share with followers (in
addition to being paid).
What’s unique about these influencers is that
they are all virtual influencers, or virtual
personas/ models. They are
computer-generated (CGI) ‘people’ that are
created and managed via artificial
intelligence (AI). They represent a variety of
brands but are each are managed by
separate companies that created and
manage them on Instagram.
On the next page you will see images of
these virtual influencers.

(Continued)
 International Journal of Advertising 1365

Appendix 3. Continued.
Condition Written description Visual stimuli
Diverse [Page 1] Please read the below [Page 2] Below are some typical posts
information about Starbucks and their from each of these virtual influencers.
plans to engage social media
influencers to represent the brand on
Instagram. Starbucks is the world’s largest
coffeehouse, operating a multinational chain
of coffeehouses and roastery reserves.
Starbucks serves hot and cold drinks,
whole-bean coffee, micro-ground instant
coffee, espresso, caffe latte, full and
loose-leaf teas, juices, Frappuccino
beverages, pastries, and snacks. Starbucks is
about to launch a new fall coffee blend and
have engaged four social media influencers
to promote the roast and share it with their
followers on
Instagram over the next few months: Michael
(44-year-old black male), Mary (29-year-old
Asian female), Linda (34-year-old black
female), and John (32-year-old Hispanic male
with a disability).
Each influencer Starbucks has chosen to
represent the brand has been a social media
influencer for a few years and have relatively
large followings (a few hundred thousand
followers). They each promote a range of
products and work with many top brands. In
exchange, they are offered free products and
experiences to share with followers (in
addition to being paid).
What’s unique about these influencers is that
they are all virtual influencers, or virtual
personas/ models. They are
computer-generated (CGI) ‘people’ that are
created and managed via artificial
intelligence (AI). They represent a variety of
brands but are each are managed by
separate companies that created and
manage them on Instagram.
On the next page you will see images of
these virtual influencers.