Exploring the effectiveness of digital manipulation disclosures for Instagram posts on source credibility and authenticity of social media influencers

Abstract

There is an ongoing discussion on the necessity of implementing disclosures for digitally manipulated pictures within the branded content of social media influencers (SMIs). To address this, we conducted two online survey-based between-subjects experiments to assess the effectiveness of digital manipulation disclosures used on Instagram by an SMI. In the first study (N = 99), we examined how a visual prominent disclosure placed directly on the sponsored posts of an SMI, mediated by the recognition of digital manipulation and the perceived authenticity of the SMI, influenced participants’ intentions toward the SMI, their brand attitude, and their likelihood in engaging in electronic Word-of-Mouth. In the second study (N = 155), we manipulated three conditions: the absence of digital manipulation, the presence of digital manipulation without disclosure, and the presence of digital manipulation with disclosure. We investigated the effects of digital manipulation on the perceived credibility of the SMI and on brand related factors, mediated by the recognition of digital manipulation. Moreover, we explored the moderating effects of product involvement on the relationship between digital manipulation recognition and source credibility. Recognizing digital manipulation has no effects on the perceived authenticity of the SMI. Among social media users with high levels of product involvement who can recognize the digital manipulation, the perceived credibility of the SMI tends to be lower. The research holds both theoretical and practical implications for the field of influencer marketing and disclosure practices.

Keywords:

• Authenticity
• credibility
• digital manipulation disclosure
• product involvement
• social media influencers

Introduction

Digital manipulation of visual content is a widespread practice on social media. Instagram is a platform where aesthetics plays a relevant role (Hund 2017), and therefore, users often enhance their visual appearance by using accessible tools (Fardouly, Willburger, and Vartanian 2018; Vendemia and De Andrea 2018). Scholars consider disclosing digitally enhanced pictures as an option to warn consumers about possible deceiving practices (Campbell et al. 2022).

An ongoing discussion centers around the necessity to regulate incorporating disclosures for digitally manipulated pictures showcased in advertising (e.g. in the United Kingdom and Germany). With the goal of educating consumers about this practice, policymakers in numerous European countries have already enacted legislative frameworks. As a result, in 2017, France passed a law regarding labeling retouched photos used for commercial purposes (Daldorph 2017) and in 2021, Norway legislated mandatory labeling of digitally enhanced pictures posted by social media influencers (SMIs) when publishing sponsored content (Gray 2021). However, the effectiveness of disclosing digitally enhanced pictures remains a topic of debate.

Previous studies stressed the negative impact on subjective well-being as a result of the exposure to digitally manipulated pictures that often depict an unrealistic body image (e.g. Brown and Tiggemann 2020; Coutoure and Harrison 2020; Fardouly and Holland 2018). However, to this date, only few studies, mostly in traditional media settings, explored the effects of photo edit disclosures on advertising outcomes showing a lack of effectiveness (e.g. Harmon and Rudd 2016; Kwan et al. 2018; Slater et al. 2012). Considering that social media influencers (SMIs) have become, in recent years, relevant advertising actors to promote businesses (Ye, Hudders, De Jans, and De Veirman 2021) research on the impact of disclosing digitally manipulated images showcased in social media influencer (SMI) sponsored content is needed.

Although SMIs receive compensation for their content (Campbell and Grimm 2019), they wield significant persuasive power within their follower communities and are often perceived as honest product reviewers (De Veirman, Cauberghe, and Hudders 2017). Hence, SMI’s perceived credibility (Hudders, De Jans, and De Veirman 2021; Ki and Kim 2019; Lou and Yuan 2019; Munnukka, Uusitalo, and Toivonen 2016; Nafees et al. 2021) and authenticity (Lee and Eastin 2021) are crucial elements of the persuasive mechanism behind influencer advertising. However, there is intense competition among content creators for users’ attention on social media (Campbell and Farrell 2020). Therefore, it is crucial to look at consumers’ involvement with the promoted products that increases their attention and cognitive processing (Celsi and Olson 1998). Hence, when exploring the effects of digital disclosures we considered product involvement, which refers to the relevance of the product endorsed by a SMI to their audience.

Our research specifically addresses the impact of digitally manipulated pictures used by SMIs in branded content, covering both influencer-related outcomes and advertising outcomes, such as the intention toward the influencer, brand attitude, and electronic Word-of-Mouth (eWOM). Thus, we look at the effectiveness of a visual prevalent clearly worded disclosure label placed directly on the branded posts in recognizing digital manipulation. Furthermore, we aim to investigate how recognizing digital manipulation affects SMI’s perceived credibility and authenticity. The focus on the moderating effect of product involvement when investigating the effects of recognizing digital disclosures on SMIs’ credibility is central to our investigation of SMI advertising.

Literature review and hypotheses

Disclosures and digital manipulation recognition

The practice of picture manipulation is on the rise and gaining popularity within the advertising industry (Campbell et al. 2022). While enhancing visual appearances is not a new practice and has been undertaken in analog settings, such as print magazines or outdoor advertising, the extensive array of digital manipulation tools and the scale of this phenomenon represent a new dimension in this practice. The aim of disclosures is to enable audiences to recognize digitally manipulated pictures. Disclosures must come in a form that effectively help consumers recognize digitally enhanced pictures as such (Naderer, Peter, and Karsay 2021). The effectiveness of disclosures, therefore, hinges on their ability to command attention and convey clarity to the audience (van Reijmersdal et al. 2023). Revealing digital manipulation and commercial connections in SMI advertising play distinct roles in providing transparency to audiences. Digital manipulation disclosures are designed to inform viewers about any alterations made to visual content through photo editing tools (Coutoure and Harrison 2020; Fardouly and Holland 2018). On the other hand, sponsorship disclosures aim to reveal the material connection or financial arrangement between the SMI and a brand or product mentioned in the content, focusing on business affiliations and potential financial interests (Boerman, van Reijmersdal, and Rozendaal 2023; Campbell and Grimm 2019). These two types of disclosures serve different purposes and have a different history of implementation.

The Persuasion Knowledge Model (Friestad and Wright 1994) often served as a theoretical framework for understanding the mechanism behind sponsorship disclosures. Thus, ad disclosures help recipients to activate and retrieve the general knowledge and beliefs they have accumulated over time about advertising (Boerman and van Reijmersdal 2019; Eisend et al. 2020), which relates to the cognitive dimension of persuasion knowledge (PK) (Boerman et al. 2018). Every new encounter with advertising content, every persuasive attempt adds to the conceptual persuasion knowledge of media recipients. The concept of advertising literacy builds on PK (Rozendaal et al. 2011; Zarouali et al. 2019). It primarily pertains to advertising as a form of persuasive communication. Disclosures activate advertising literacy (Boerman, Rozendaal, and van Reijmersdal 2023). While advertising disclosures in SMIs’ branded content are mandatory in many countries, digital manipulation disclosures are relatively new to the public.

Previous existing scholarship on sponsorship disclosure in SMI advertising can contribute to the research on the efficacy of digital manipulation disclosures. Hence, studies focusing on sponsorship transparency highlighted that disclosures should be obvious to consumers, straightforward, and noticeable (Campbell and Grimm 2019; Wojdynski and Evans 2016) and that they must be understood by the target audiences to which they are presented (Naderer, Peter, and Karsay 2021), suggesting that subtlety or placement within a cluster of hashtags may hinder their effectiveness. Campbell and Grimm (2019) emphasize that advertising disclosures should be straightforward and conspicuous, while Wojdynski and Evans (2016) found that placing disclosures among other hashtags does not contribute significantly to advertising recognition. These findings collectively suggest that the visual salience and clarity of disclosures play a pivotal role in their efficacy.

The few studies focusing on photo edit disclosures in social media settings have consistently employed disclaimers placed as captions (Brown and Tiggemann 2020; Coutoure and Harrison 2020; Fardouly and Holland 2018; Livingston, Holland, and Fardouly 2020; McComb, Gobin, and Mills 2021). However, these studies did not find a significant effect from either form of caption on women’s body satisfaction or mood.

It is noteworthy that these studies employed various types of captions, ranging from self-disclaimers in the form of body positive captions or hashtags (Brown and Tiggemann 2020) to captions either idealizing or criticizing the retouched images (Coutoure and Harrison 2020). Other variations included generic self-disclaimers of a more general nature or specific self-disclaimers detailing the altered body parts (McComb, Gobin, and Mills 2021). Conversely, Naderer, Peter, and Karsay (2021) have created standard disclaimers placed as captions, testing their efficacy on the perceived realism of the presented images. Intriguingly, neither disclaimer type, whether placed as a caption or as a hashtag, demonstrated a significant effect when compared to a no-disclaimer condition. However, drawing from the above-mentioned studies on advertising disclosure (e.g. Campbell and Grimm 2019; Wojdynski and Evans 2016) we argue that disclosing digitally manipulated SMI’s branded posts by placing a clearly worded label directly on the branded posts can contribute to the recognition of such digital manipulation and therefore we posit:

H1. The presence of a digital manipulation disclosure enhances the recognition of digitally manipulated pictures, compared to the absence of a disclosure on a digitally manipulated picture.

The effects of digital manipulation disclosure on SMI’s perceived credibility and authenticity

In addition to the cognitive aspects related to recognizing the persuasive intent, the PK framework also includes an evaluative dimension. This dimension involves the affective component of persuasion knowledge, encompassing critical attitudes or feelings, such as skepticism or disliking towards the persuasive message (Boerman et al. 2018; Boerman, van Reijmersdal, and Neijens 2012; van Reijmersdal et al. 2016).

Persuasion knowledge is related to source credibility that refers to the users’ perception of the communication source (Hovland and Weiss 1951; Metzger and Flanagin 2015) and it is a relevant asset for SMIs (Balaban and Mustățea 2019; Djafarova and Rushworth 2017). Consistent with the rationale of the PK model, individuals perceive and evaluate the credibility of SMIs considering the recognition of the advertising intent. If they are aware of the persuasion tactics of the SMIs they might scrutinize the credibility of the SMI closely.

Various studies (De Jans et al. 2020; Grigsby 2020; van Reijmersdal and van Dam 2020) collectively suggest that advertising disclosures lead to advertising recognition, which can have negative effects on brand attitude and influencer perceptions, such as the intention toward the SMI. The intention toward the SMI refers to the audience’s intent to follow the influencer and engage with their content (Liljander, Gummerus, and Söderlund 2015). van Reijmersdal and van Dam (2020) identified a negative influence of advertising recognition on both brand attitudes and attitudes toward the influencer, attributing this outcome to heightened attitudinal persuasion. Grigsby (2020) further revealed that ad recognition led to perceptions of pushiness, thereby negatively impacting both ad and brand attitudes. Additionally, De Jans et al. (2020) highlighted the mediating roles of influencers’ assessments, particularly in terms of credibility, where sponsorship disclosures led to advertising recognition, triggering skepticism, and reducing the influencer’s credibility, resulting in lower brand attitudes and brand recognition.

However, there are also studies indicating that consumers perceive advertisement disclosures as a sign of transparency displayed by influencers (Balaban, Mucundorfeanu, and Naderer 2022; Evans, Wojdynski, and Hoy 2019; van Reijmersdal et al. 2023; Wojdynski, Evans, and Hoy 2018). The persuasive message is conveyed openly, and users do not feel deceived (Abendroth and Heyman 2013). Consequently, this positively affects the credibility of the advertiser (Hwang and Jeong 2016), the credibility of the blogger, measured as trust (Colliander and Erlandsson 2015), and the perceived trustworthiness of an SMI (Audrezet, de Kerviler, and Moulard 2020). Trustworthiness and likeability are elements of the source credibility of actors engaged in embedded advertising, such as SMIs and bloggers (Audrezet, de Kerviler, and Moulard 2020; Colliander and Erlandsson 2015; Lou and Yuan 2019).

Overall, existing literature (van Reijmersdal et al. 2023) indicates that the impact of disclosures may depend on their efficacy in informing consumers about their intended purpose (van Reijmersdal et al. 2023). Transparent advertising disclosures play a crucial role in mitigating negative effects associated with heightened persuasion knowledge, ultimately fostering positive perceptions of SMIs, and increasing engagements with posts (Giuffredi-Kähr, Petrova, and Malär 2022; Karagür et al. 2022). Unlike traditional advertising disclosures, photo edit disclosure transparently communicating image alterations without direct reference to persuasive intent, have the potential to positively affect audience perceptions, aligning with consumer preferences for transparency in advertising (Evans, Wojdynski, and Hoy 2019). Consequently, we hypothesized:

H2. Digital manipulation disclosure has a positive impact on SMI’s perceived credibility.

H3. The recognition of the digital manipulation has a negative impact on SMI’s credibility.

Authenticity is a context-related concept that was defined in relation to SMI’s communication strategies as posting content in line with the ideals and values of the influencer (Pöyry et al. 2019). According to Lee and Eastin (2021), sponsorship transparency is a crucial component of SMI authenticity, reflected through various factors, such as content, visibility, transparency, and imaginative brand endorsement (Balaban and Szambolics 2022). Abidin and Ots (2016) describe the outcomes of publishing unedited content by SMIs as ‘porous authenticity’, emphasizing the strategic efforts of SMIs in presenting authentic personas online. Lee and Eastin (2021) stress sincerity as a dimension of SMIs’ perceived authenticity, while Audrezet, de Kerviler, and Moulard (2020) emphasize that posting unedited content enhances authenticity perception. These findings underscore the importance of transparency and genuine self-presentation in shaping how followers perceive the authenticity of SMI on social media platforms.

In the context of our study, we anticipate that recognizing digital alterations in pictures, which deviate from the unedited content that enhances authenticity perception, may have a negative impact on authenticity. The transparency introduced by the disclosure could disrupt the seamless presentation of unedited content, potentially leading followers to question the authenticity of the SMI. We thus, assume the following:

H4. The recognition of the digital manipulation has a negative impact on SMI’s perceived authenticity.

The moderating role of product involvement

Regardless of the type of advertising, source credibility is a strong predictor for advertising outcomes and has substantial effects when combined with emotions and cognitions (Eisend and Tarrahi 2016). Product involvement defined as the personal relevance of a product for the consumer (O’Cass 2000) affects consumer attention and the comprehension process (Krugman 1966). Previous literature on advertising disclosures (Eisend et al. 2020) highlighted that product characteristics influence the impact of disclosures. Thus, high product involvement can lead to increased processing of advertising content (Krugman 1966) and tends to prompt recipients to engage in a more thorough and careful evaluation of advertising messages (Celsi and Olson 1998). Therefore, when product involvement is high in the context of recognizing digital manipulation, consumers are likely to engage in a more critical assessment of the influencer’s credibility. As a result, we anticipate significant interaction effects between product involvement and the recognition of digital manipulation on the perceived credibility of the SMI. Thus, we hypothesized:

H5. Product involvement moderates the relationship between the recognition of digital manipulation and the perceived credibility of the SMI, in the sense that individuals with high levels of product involvement who recognize digital manipulation will rate the SMI’s perceived credibility lower.

The impact of SMI’s perceived credibility and authenticity on advertising outcomes

Robust scholarship stressed that credibility plays a crucial role in the persuasive processes for all types of endorsers (Eisend and Tarrahi 2016). In the context of SMI advertising, previous research has highlighted the favorable impact of SMI’s credibility on outcomes related to both the influencer and the brand. Hence, increased credibility of the SMI has favorable implications for both the intention towards the SMI (Liljander, Gummerus, and Söderlund 2015) and the endorsed brand (Balaban, Mucundorfeanu, and Mureșan 2022; Lou and Yuan 2019; Munnukka, Uusitalo, and Toivonen 2016). This aspect is crucial for SMIs as strategic communicators (Enke and Borchers 2019) given that SMIs aspire to grow their community of followers (Campbell and Farrell 2020).

In addition, users actively engage in electronic-Word-of-Mouth (eWOM) by sharing the SMI’s posts within their own follower communities. eWOM holds significance for both brands and SMIs since it results in the sponsored content gaining exposure beyond the SMI’s immediate audience. When followers engage with and share content, it reaches users who may not be directly connected to the SMI (Evans et al. 2017). This sharing mechanism widens the content’s visibility, making it accessible to potential audiences who might not have been exposed to the brand or influencer otherwise. In essence, eWOM creates a ripple effect turning followers into brand advocates who organically introduce the content to new and diverse audiences, enhancing the overall reach and exposure of the sponsored material (Balaban, Szambolics, and Chirică 2022).

Previous research highlighted the relevance of SMI’s perceived authenticity for resulting in various advertising outcomes. These outcomes encompass brand-related aspects, such as brand attitude and purchase intention, as well as factors pertaining to the content creator as an advertising figure, including the intention toward the SMI (Fritz, Schoenmueller, and Bruhn 2017; Kühn and Riesmeyer 2021; Pöyry et al. 2019; Schallehn, Burmann, and Riley 2014). Hence, followers favor SMIs that are thought to be authentic, and this preference results in affective and behavioral outcomes (Jin 2018). The intention toward the SMI is crucial given that the goal of SMIs is to grow their community of followers (Campbell and Farrell 2020). According to previous research, users are more likely to choose the endorsed brand if they perceive the SMI as authentic (Ki et al. 2020; Kowalczyk and Pounders 2016). Hence, we posit:

H6. The perceived credibility of the SMI has a positive impact on (a) the intention toward the SMI, (b) brand attitude, and (c) eWOM.

H7. The perceived authenticity of the SMI has a positive impact on (a) the intention toward the SMI, (b) brand attitude, and (c) eWOM.

The conceptual model is presented in Figure 1.

Figure 1. Conceptual model.

Method

We conducted two studies to examine the effects of a digital manipulation disclosure used in sponsored SMI posts. We examined the impact of a manipulation disclosure placed directly on the pictures as a stamp used in sponsored SMI posts, a disclosure type that is applicable and mandatory in Norway. This labeling standard represents the singular method currently employed for disclosing edited pictures on social media platforms. It entails directly integrating the photo editing disclosure onto the image itself.

In the first study, we looked at the impact of a digital manipulation disclosure mediated by the digital manipulation recognition and perceived authenticity of the SMI on the intention toward the influencer, brand attitude, and eWOM. Building on the first study, in Study 2 we investigated the effectiveness of the digital manipulation disclosure mediated by the digital manipulation recognition and perceived credibility of the SMI on the intention toward the influencer, brand attitude, and eWOM. In both studies, only branded posts were subjected to digital manipulation and disclosure. Branded posts were digitally enhanced to increase the SMI’s facial symmetry, enlarge her lips and eyes, and make her body appear very fit.

Study 1

Method

We conducted an online survey-based between-subjects experiment by manipulating the absence (n₁ = 73) vs. the presence (n₂ = 80) of a photo manipulation disclosure. We examined different mediators, including the digital manipulation recognition and the perceived authenticity of the SMI. The disclosure tested was worded as ‘Retouched person’ (original wording in German: ‘Retuschierte Person’) and placed as a stamp directly on the digitally modified photos of an SMI, in compliance with the new regulations in countries like Norway (Gray 2021). The label was positioned in the upper left-hand corner of the pictures, contrasting the background, and adhering to the imposed size and fonts. Our stimulus materials depicted a fictitious female SMI’s Instagram account (33,000 followers), endorsing the brand Adidas.

Design and stimulus materials

We created a fictitious Instagram page for the influencer by sourcing the images from a public Instagram account from a different country than the country where our research was conducted. This approach is consistent with established practices in previous studies involving manipulated pictures, where images were drawn from public Instagram accounts (e.g. Livingston, Holland, and Fardouly 2020, McComb, Gobin, and Mills 2021, Tiggemann and Anderberg 2020). We have given the influencer a new name, manipulated the follower count, created captions, and included likes for all posts. Moreover, we asked participants if they were familiar with the SMI and received no positive answers. The three branded pictures were further enhanced for the study using Adobe Photoshop 24.1.

Participants in both groups used their mobile phones to complete a survey distributed via a Qualtrics link, which consisted of three parts. In the first part, after providing informed consent, respondents were asked about their socio-demographics and social media usage. In the second part, they were shown the account information of our fictitious SMI and then presented with six Instagram posts from the SMI. Among these posts, three were branded content featuring the influencer promoting sports clothing from the brand Adidas.

In the three branded posts that were disclosed, we deliberately enhanced specific features of the SMI’s physical appearance. The advertising disclosure worded ‘Paid partnership with Adidas’ was positioned above the digitally modified photos of the branded posts, that being standardized disclosure commonly used on Instagram. The digital manipulation disclosure was placed as a stamp on the pictures showing brand endorsement.

The first group viewed the posts without any digital manipulation disclosure, while the second group saw the exact same posts but with the digital manipulation disclosure included. Advertising disclosure, a mandatory practice in Germany, was not a factor in this experiment and was shown to both groups. Examples of the posts considered as stimulus materials are presented in Appendix A.

In the third part of the survey, we asked participants to assess our mediators and dependent variables. Furthermore, to draw attention to the potential adverse effects of digitally enhanced images, as emphasized in previous research, our participants were presented with the following disclaimer after completing the questionnaire: ‘Instagram users frequently digitally edit their photos. Digitally manipulated photographs do not represent an attainable standard of beauty. Each individual possesses their own unique and inherent beauty, and we encourage you to discover and embrace your inner and outer beauty beyond the beauty ideals promoted on social media and in the media.’

Participants

For this study, initially, 153 participants were recruited through the crowdsourcing platform Prolific, known for providing high data quality and being a relevant tool for conducting online experiment-based surveys (Palan and Schitter 2018). Participants, fluent in German and Instagram users, were rewarded ∼5 euros for viewing the stimulus material and completing the questionnaire on their mobile phones. Half of the participants in the disclosure group (n = 40) reported seeing the disclosure ‘Retouched person.’ Moreover, 14 respondents (19.17%) from the no-disclosure reported also seeing the disclosure ‘Retouched person.’ Therefore, the manipulation check was successful (χ²(1) = 15.88, p < .001, Φ = .32). Aligned with previous research (McComb, Gobin, and Mills 2021) we excluded for further analysis participants who failed to pass the manipulation check. Hence, our final sample consists of N = 99, 50.5% female, 46.5% male young adults, and 3% other gender, aged 18–33 (M = 22.63, SD = 4.07).

Measures

Respondents were asked to rate our mediators, first the manipulation recognition with a single question on a scale from 1 = totally disagree to 7 = totally agree (M = 4.83, SD = 1.54) and their perceived authenticity of the SMI was measured using 14 questions on a 7-point Likert scale (e.g. ‘The influencer gives very honest reviews on brands’ from 1 = ‘not at all’ to 5 = ‘very much’; α = .92, M = 3.88, SD = .87; Lee and Eastin 2021). The dependent variable the intention toward the SMI rated on a 7-point Likert scale using four questions (e.g. ‘I would like to know more about this influencer’, from 1 = ‘totally disagree’ to 7 = ‘totally agree’; α = .93, M = 1.92, SD = 1.07; Liljander, Gummerus, and Söderlund 2015), and brand attitude on a 7-point Likert point semantic differential scale with the attributes ‘unattractive’-’attractive’, ‘negative’-’positive’, ‘boring’-’interesting’, and ‘unlikeable’-’likeable’ (α = .92, M = 4.56, SD = 1.35; Matthes and Naderer 2016). Moreover, eWOM was assessed by participants on a 7-point Likert scale using four questions (e.g. ‘I would share these posts and/or stories with my Instagram friends’, from 1 = ‘totally disagree’ to 7 = ‘totally agree’; α = .94, M = 1.72, SD = 1.05; Sohn 2009; Evans et al. 2017).

For the randomization checks, we considered self-photo manipulation measured using four questions on a 5-point Likert scale (e.g. ‘I often make specific parts of my body look larger or look smaller’ from 1 = ‘never’ to 5 = ‘always’; α = .72, M = 2.01, SD = .82; Gioia et al. 2021) and body appreciation that was measured using six questions also on a 5-point Likert scale (e.g. ‘I take a positive attitude towards my body’, from 1 = ‘never’ to 5 = ‘always’; α = .87, M = 3.81, SD = .68; Slater, Varsani, and Diedrichs 2017).

Results

Randomization checks for age [t (97) = −0.19, p = .78], gender [χ²(2) = .99, p = .61], body appreciation [t (97) = −1.10, p = .97], and self-photo manipulation [t (97) = 1.04, p = .49] showed that the differences between outcome variables are not a result of inherent differences between conditions. Descriptive statistics per condition and between-subjects differences are presented in Table 1, while bivariate correlations for mediators and dependent variables are presented in Table 2.

Table 1. Descriptive analysis and between-subjects differences for Study 1.

	No disclosure	Disclosure	Mean differences
	M (SD)	M (SD)	t-Tests and p-values
Recognition of the digital manipulation	4.49 (1.18)	5.33 (1.87)	t(97) = 2.71, p = .01
Perceived SMI’s authenticity	3.91 (.87)	3.83 (.87)	t(97) = .45, p = .78
Intention toward the SMI	2.03 (1.11)	1.76 (.98)	t(97) = 1.27, p = .57
Brand attitude	4.66 (1.28)	4.41 (1.47)	t(97) = .89, p = .22
eWOM	1.89 (1.24)	1.47 (.62)	t(97) = 1.98, p = .004

Note: N = 99.

Table 2. Bivariate correlations Study 1.

	1.	2.	3.	4.	5.
1. Digital manipulation recognition	1
2. Perceived SMI’s authenticity	.002	1
3. Intention toward the SMI	−0.051	.325**	1
4. Brand attitude	−0.011	.344**	.159	1
5. eWOM	−0.140	.314**	.561**	.195	1

Notes: N = 99; **p < .01.

In Study 1 we tested the effects of the disclosure on the digital manipulation recognition (H1) and the impact of the recognition on the perceived authenticity of the SMI (H4). Furthermore, we tested how the perceived authenticity of the SMI affects (a) the intention toward the SMI (H7a), (b) brand attitude (H7b), and (c) eWOM (H7c). We conducted three serial mediation analyses for each dependent variable, intention toward the SMI, brand attitude, and eWOM with digital manipulation recognition and the perceived authenticity of the SMI as mediators. We used Model 6, PROCESS macro 3 in SPSS (Hayes 2022) employing 5000 bootstrap samples. The no-disclosure group was used as a reference group.

We posited that the presence of a disclosure contributes to digital manipulation recognition (H1). Disclosure has a direct significant impact on the recognition of digitally manipulated pictures (b = .83, SE = .31, 95% CI = [.22, 1.44], p = .008). Therefore, we found evidence to support H1. We hypothesized that the digital manipulation recognition has a negative impact on SMI’s authenticity (H4). The recognition of the use of digitally manipulated pictures had no direct impact on the perceived authenticity of the SMI (b = .009, SE = .06, 95% CI = [−0.11, .13], p = .87), hence H4 was not supported. Furthermore, the digital manipulation disclosure has no significant direct impact on the perceived authenticity of the SMI (b = −0.09, SE = .19, 95% CI = [−0.46, .28], p = .64). We posited that the perceived authenticity of the SMI has a positive impact on (a) the intention toward the SMI (H7a), brand attitude (H7b), and eWOM (H7c). Our findings showed that the perceived authenticity of the SMI has a significant positive impact on the intention toward the SMI (b = .39, SE = .12, 95% CI = [.16, .63], p < .001), brand attitude (b = .53, SE = .15, 95% CI = [.23, .83], p < .001), and eWOM (b = .37, SE = .11, 95% CI = [.14, .60], p < .001). Thus, results are in favor of H7a, H7b, and H7c. Moreover, the digital manipulation disclosure has no significant direct impact on the intention toward the SMI (b = −0.23, SE = .21, 95% CI = [−0.69, .19], p = .29), brand attitude (b = −0.21, SE = .27, 95% CI = [−0.76, .33], p = .44), and eWOM (b = −0.34, SE = .21, 95% CI = [−0.75, .08], p = .32). Coefficients of direct effects are shown in Figure 2.

Figure 2. Coefficients of serial mediation Study 1. Notes: N = 99; Unstandardized effects are presented N = 99;**p < .01, ***p < .001.

Placing the disclosure directly on the pictures of the branded posts contributed to digital manipulation recognition. However, the recognition of digital manipulation did not yield the anticipated impact on SMI’s authenticity. Moreover, as shown in Table 3 where the results of hypotheses testing are presented, there is no indirect effect of the digital manipulation disclosures on the SMI and on brand-related outcomes.

Table 3. Hypotheses testing Study 1.

	Pathway
Hypothesis	Direct effects	b (SE)	95% CI
H1.	Digital manipulation disclosure > Recognition of the digital manipulation	.83 (.31)**	[22, .1.44]
H4.	Recognition of the digital manipulation > Perceived SMI’s authenticity	.009 (.06)	[−0.11, .13]
	Digital manipulation disclosure > Perceived SMI’s authenticity	−0.09 (.19)	[−0.46, .28]
H7a.	Perceived SMI’s authenticity > Intention toward the SMI	.39 (.12)***	[.16, .63]
	Digital manipulation disclosure > Intention toward the SMI	−0.23 (.21)	[−0.69, .19]
	Recognition of the digital manipulation > Intention toward the SMI	−0.02 (.07)	[−0.15, .12]
H7b.	Perceived SMI’s authenticity > Brand attitude	.53 (.15)***	[.23, .83]
	Digital manipulation disclosure > Brand attitude	−0.21 (27)	[−0.76, .33]
	Recognition of the digital manipulation > Brand attitude	.008 (.09)	[−0.17, .18]
H7c.	Perceived SMI’s authenticity > eWOM	.37 (.11)***	[.14, .60]
	Digital manipulation disclosure > eWOM	−0.34 (.21)	[−0.75, .08]
	Recognition of the digital manipulation > eWOM	−0.07 (.07)	[−0.20, .07]
	Indirect effects	Effect (BootSE)	95% BootCI
	Digital manipulation disclosure > Recognition of the digital manipulation > Perceived SMI’s authenticity > Intention toward the SMI	−0.04 (.09)	[−0.22, .14]
	Digital manipulation disclosure > Recognition of the digital manipulation > Perceived SMI’s authenticity > Brand attitude	−0.05 (.13)	[−0.33, .18]
	Digital manipulation disclosure > Recognition of the digital manipulation > Perceived SMI’s authenticity > eWOM	−04 (.09)	[−0.22, .14]

CI: bias corrected 5000 bootstrap confidence interval.

Notes: N = 99; Unstandardized b coefficients (with boot SE between parentheses). *p < .05; **p < .01; ***p < .001.

Study 2

Method

In the second study, our objective was to evaluate the effects of SMIs’ advertising posts, specifically comparing those containing unretouched images to digitally retouched ones, both with and without a disclosure. To achieve this, we digitally altered the pictures we had captured to employ them as experimental stimuli, and we included a visual disclosure as a stamp to them. We decided to test the impact of digitally enhanced pictures in SMI advertising for skincare products, considering the importance for consumers of a genuine depiction of how skin care products work. Moreover, in this study, we explored the moderating role of product involvement in the relationship between recognizing digital manipulation and the perceived credibility of the SMI.

Participants

Initially, we recruited 177 female participants from the United Kingdom (UK), via the crowdsourcing platform Prolific, and randomly assigned them to one of the three conditions. Given the relevance of the ongoing public discussion in this country regarding the introduction of a mandatory disclosure for digitally manipulated pictures on SMI advertising posts, we conducted the study with British participants. Fluency in English and having an active Instagram account were prerequisites, and participants were compensated with ∼5 USD for their involvement. We opted for a female sample because our stimulus materials featured a fictitious female SMI. Previous research stressed the relevance of matching the SMI’s gender with the gender of participants to avoid confounding effects (De Jans et al. 2020; De Veirman, Cauberghe, and Hudders 2017). Furthermore, women perceived higher levels of similarity with a female SMI than men, and same-gender endorsement positively affects persuasion (Hudders and De Jans 2022).

Stimulus materials and study design

Participants across the three groups were instructed to view the SMI account information on their smartphones. The SMI was showcased as a lifestyle digital content creator. Following this, they were shown a total of six Instagram posts, comprising three branded posts promoting skincare products from the CeraVe brand and three organic posts that did not feature any brands. The branded posts included photos showcasing the SMI endorsing skin care products as close-up shots focusing on the SMI’s face, while the other three pictures included in the organic posts showed the SMI from a distance. We included pictures of a fictitious female SMI (41,000 followers) in our stimulus material after closely examining similar posts from genuine SMIs promoting skincare products, thus enhancing the external validity of our experiment. The stimulus materials are shown in Appendix B. Figure 3 illustrates an example of a branded post without digital manipulation, a branded post featuring a digitally manipulated image without disclosure, and one with a disclosure included.

Figure 3. Exampled of a branded post used as stimulus material for each condition I Study 2.

Similar to previous studies (McComb, Gobin, and Mills 2021), a manipulation check was performed by asking participants to answer the following question: ‘Did any of the Instagram posts contain one of the following statements or a similar statement?’ Participants were required to select one out of four possible answers. For the third group the correct statement was ‘The person in the photos has been retouched’, while for the other groups, the correct choice was ‘none of the above’. Five participants from each of the two groups failed to indicate the correct answer. In the third group, which viewed digitally manipulated branded pictures with a disclosure, n = 49 (∼80%) participants indicated that they saw the message about the retouched person in the posts. Therefore, the manipulation check was successful [χ²(1) = 92.5, p < .001, Φ = .72]. We eliminated a total number of 22 cases from the original sample and we considered for the analysis a sample of N = 155 participants aged 19–36 years (M = 28.81, SD = 4.56).

The differences between the three groups consist of the absence vs. the presence of digitally manipulated pictures with or without a disclosure in the branded posts. Hence, the first group (n₁ = 52) viewed branded posts that featured unaltered pictures of the SMI. The second group (n₂ = 54) saw identical branded posts but with digitally manipulated images. The third group (n₃ = 49) viewed branded posts with digitally manipulated images, accompanied by a digital manipulation disclosure. Like in Study 1, the disclosure was placed directly on the pictures as a stamp with the wording ‘retouched person’.

Considering existing legislation in the UK that requires mandatory disclosure of SMI advertising, the paid partnership was disclosed for all conditions in the standardized disclosure form used on Instagram. ‘Paid partnership’ placed above each of the branded posts. Thus, similar to Study 1, the advertising disclosure was not considered a factor and was shown in the stimulus materials for all groups.

The survey was distributed as a Qualtrics link and included three parts. After the informed consent, in the first part, participants answered socio-demographic questions. They rated their Instagram usage, satisfaction with their physical appearance, how often they digitally manipulated their pictures, and how important skin care products were to them. In the second part, we instructed the participants to view the stimulus material in the same manner as they typically engage with social media influencers’ posts. Moving on to the third part, we requested participants to provide ratings for our moderators and dependent variables. Similar to Study 1, at the end of the survey, we included a disclaimer aimed at increasing awareness of the potential drawbacks associated with viewing digitally enhanced pictures.

Pre-test

We chose a fictitious SMI and used pictures of her that we took for this study, thereby enhancing the internal validity of our study. Aligned with practices commonly observed among social media influencers on platforms like Instagram, we employed Adobe Photoshop 24.1 to retouch the SMI’s appearance. This involved the removal of skin imperfections and selective enhancements of specific features of the SMI’s physical appearance.

Before conducting our experiment, we ran a pre-test involving three groups of 20 female Instagram users each, totaling N = 60 participants. Participants assessed the SMI-branded posts, which included pictures that were unaltered, compared to digitally manipulated pictures with and without a disclosure. They were asked to rate the extent of digital manipulation in these Instagram posts on a scale from 1, ‘not at all manipulated’ to 7, ‘strongly manipulated.’ The results indicated that posts with unaltered pictures received a lower average rating, M = 1.7, whereas digitally manipulated pictures without disclosure received an average rating of M = 3.35, and those containing a disclosure scored the highest, M = 5.8. These findings confirmed that participants accurately perceived the level of image manipulation as intended for our study’s stimuli. Therefore, we proceeded to use these materials in the study.

Measures

The first mediator, namely digital manipulation recognition, was measured with a single question using a 7-point Likert scale (from 1 = ‘totally disagree’ to 7 = ‘totally agree’; M = 4.10, SD = 1.31). The second mediator, the perceived credibility of the SMI was rated by participants also on a 7-point Likert scale using eight questions (‘The influencer is attractive/good looking/sincere/honest/trustworthy/competent/professional/experienced’ from 1 = ‘totally disagree’ to 7 = ‘totally agree’ (α = .87, M = 4.65, SD = .77; Ohanian 1990; Russell and Rasolofoarison 2017).

Our dependent variables, the Intention toward the SMI (α = .98, M = 2.94, SD = 1.62; Liljander, Gummerus, and Söderlund 2015), brand attitude (α = .89, M = 5.16, SD = 1.01; Matthes and Naderer 2016), and eWOM (α = .94, M = 2.19, SD = 1.46; Sohn 2009; Evans et al. 2017) were measured with the same instruments as in Study 1.

Furthermore, participants rated how critical skin care was to them, measured as product involvement, considered as a moderator with five questions on a 7-point Likert scale (e.g. ‘Skin care plays an important role in my daily life’; from 1 = ‘totally disagree’ to 7 = ‘totally agree’; α = .95, M = 5.20, SD = 1.44; Choo et al. 2014).

Moreover, participants’ body appreciation (α = .91, M = 3.18, SD = .78; Slater, Varsani, and Diedrichs 2017) was also measured in the same way as for Study 1. Advertising recognition was measured with one question on a 7-point Likert scale: ‘Did you notice advertising while watching the Instagram posts?’ (M = 6.66, SD = .89; De Jans et al. 2020). Finally, participants’ familiarity with the brand CeraVe shown in our stimulus materials was rated with a yes or no question.

Results

As randomization checks, we run a series of one-way ANOVA and chi-square tests to ensure that outcome differences between groups result from our stimuli and not from other variables we controlled for. Thus, age [F (2) = .49, p = .61], and participants’ body appreciation [F (2) = .35, p = .71] were successful. More than 89% (n = 138) of the participants were familiar with the brand CeraVe. However, familiarity with the brand was not significantly different among conditions [χ²(2) = 7.51, p = .06].

Descriptive statistics per condition are shown in Table 4. A series of one-way ANOVA were conducted for the mediators and dependent variables. Results showed significant differences between conditions regarding the recognition of the digital manipulation [F(2,155) = 24.48, p < .001] and brand attitude [F(2,155) = 3.12, p = .04]. There are no significant differences regarding the perceived SMI’s credibility [F(2,155) = .73, p = .48], intention towards the SMI [F(2,155) = .002, p = .99], and eWOM [F(2,155) = 2.08, p = .12].

Table 4. Descriptive analysis and the results of the one-way ANOVA tests Study 2.

	No digital manipulation	Digital manipulation without disclosure	Digital manipulation with disclosure	Mean differences
	M (SD)	M (SD)	M (SD)	p-Values
Recognition of the digital manipulation	3.33(1.09)^b,c	3.92 (1.10)^a,c	4.97 (1.21)^a,b	pab < .001, pac < .001, pbc < .001
Perceived SMI’s credibility	4.93 (.64)	4.88 (.77)	4.74 (.90)	pab = 1, pac = .73, pbc = 1
Intention toward the SMI	2.95 (1.56)	2.93 (1.54)	2.95 (1.79)	pab = 1, pac = 1, pbc = 1
Brand attitude	5.43 (1.09)^b	4.96 (.98)^a	5.10 (.91)	pab < .048, pac = .28, pbc = 1
eWOM	2.46 (1.61)	1.89 (1.12)	2.22 (1.59)	pab = .13, pac = 1, pbc = .73

Notes: N = 155; ^a,b,cGroup differences, a represents the no digital manipulation group, b represents the digital manipulation with no disclosure group, and c represents the digital manipulation with disclosure group; p-values from the Bonferroni post-hoc tests are shown.

Our dependent variables, as well as SMI’s perceived credibility, positively correlated. Digital manipulation recognition negatively correlates with the perceived credibility of the SMI. Advertising recognition positively correlates with digital manipulation recognition (r = .23, p = .03) and there are no differences between conditions concerning the recognition of advertising [F(2) = 2.33, p = .10, η_p² = .30]. The results of the bivariate correlations are shown in Table 5.

Table 5. Bivariate correlations Study 2.

	1.	2.	3.	4.	5.
1. Recognition of the digital manipulation	1
2. SMI’s perceived credibility	−0.205*	1
3. Intention toward the SMI	−0.126	.471**	1
4. Brand attitude	−0.079	.353**	.318**	1
5. eWOM	−0.052	.235**	.476**	.335**	1

Notes: N = 155; *p < .05, **p < .01.

We posited that using a digital manipulation disclosure will contribute to digital manipulation recognition (H1). We performed an ANOVA test with the experimental conditions as the fix factor and digital manipulation recognition as a dependent variable. We observed a significant effect of the digital manipulation [F(2) = 24.48, p < .001, η_p² = .26]. The Bonferroni post hoc test showed that digital manipulation recognition is significantly different in our three conditions, the participants in the group that saw branded posts with digitally manipulated pictures scored higher (M = 4.81, SD = 1.21) in recognizing the digital manipulation compared to participants in the group that saw the post with no digital manipulation (M = 3.85, SD = 1.42, p < .001). Moreover, the participants who saw the posts containing digitally manipulated pictures with a disclosure (M = 5.88, SD = 1.39) had significantly higher recognition values than those in the first and second group (p < .001). Hence, our findings support H1.

We posited that disclosing digital manipulation has a positive effect on the SMI’s perceived credibility (H2). To test the second hypothesis, we performed an ANOVA test with the experimental conditions as the fix factor and the SMI’s perceived credibility as a dependent variable. We observed no significant effect of the digital manipulation disclosure [F(2) = .73, p = .48, η_p² = .01]. Hence, we found no evidence to support H2.

To test the proposed hypothesized model of serial moderated mediation regarding digital manipulation recognition via SMI’s credibility on the intention toward the SMI, brand attitude, and eWOM as dependent variables, we applied Model 91, PROCESS macro 3 in SPSS (Hayes 2022), employing 5000 bootstrap samples. Our dependent variable is multicategorical, and the group exposed to the unretouched images was used as a reference group. Product involvement was considered as a moderator. We applied separate analyses for each dependent variable.

Our analysis showed that digital manipulation recognition is higher in the disclosure condition (b = 1.66, SE = .24, 95% CI = [1.17, 2.12], p < .001) and in the no disclosure condition (b = .59, SE = .24, 95% CI = [.12, 1.06], p = .01) compared to the condition without a digital manipulation.

We posited that recognizing the digital manipulation of pictures will negatively affect the SMI’s perceived credibility (H3). Our analysis showed that recognizing digital manipulation has no significant direct effect on SMI’s credibility (b = .31, SE = .19, 95% CI = [−0.07, .69], p = .11). Thus, H3 was not supported.

We posited that product involvement will have a moderating role in the relationship between the recognition of digital manipulation and the perceived credibility of the SMI (H5). Findings showed that product involvement positively affects SMI’s perceived credibility (b = .30, SE = .14, 95% CI = [.02, .58], p = .03). The interaction between the recognition of the digital manipulation and credibility (b = −0.08, SE = .03, 95% CI = [−0.15, −0.02], p = .02) had a negative significant effect on SMI’s credibility. Hence, our mediation hypothesis (H5) was confirmed. Our findings indicate the following: at low levels of product involvement = 3.73 (Effect = −0.002, SE = .08, 95% BootCI = [−0.16, .15], p = .98), at medium levels of product involvement = 5.19 (Effect = −0.12, SE = .06, 95% BootCI = [−0.24, −0.009], p = .03), and at high levels of product involvement = 6.65 (Effect = −0.24, SE = .07, 95% BootCI = [−0.39, −0.09], p = .001). Interaction effects are depicted in Figure 4.

Figure 4. Interaction effects of the recognition of the digital manipulation moderated by product involvement on the SMI’s perceived credibility.

We posited that the SMI’s perceived credibility will have a positive effect on the intention toward the SMI (H6a), brand attitude (H6b), and eWOM (H6c). The analysis showed that the SMI’s perceived credibility positively affects the intention toward the brand (b = .99, SE = .15, 95% CI = [.69, 1.29], p < .001), brand attitude (b = .43, SE = .10, 95% CI = [.23, .63], p < .001), and eWOM (b = .43, SE = .15, 95% CI = [.12, .73], p = .006). Thus, H6a, H6b, and H6c were supported. In the case of the intention toward the influencer as the dependent variable, the index of the moderated mediation is significant for the digitally manipulated pictures with no disclosure (Index = −0.05 BootSE = . 03, BootCI [−0.11, −0.005]), and with disclosure conditions (Index = −0.14, BootSE = . 06, BootCI [−0.26, −0.03]). In the case of brand attitude as the dependent variable, the index of the moderated mediation is significant for the digitally manipulated pictures with no disclosure (Index = −0.02, BootSE = . 01, BootCI [−0.06, −0.01]), and with disclosure conditions (Index = −0.06, BootSE = . 03, BootCI [−0.12, −0.01]). Finally, also in the case of eWOM as the dependent variable, the index of the moderated mediation is significant for the digitally manipulated pictures with no disclosure (Index = −0.02, BootSE = . 02, BootCI [−0.06, −0.0004]), and with disclosure conditions (Index = −0.06, BootSE = .04, BootCI [−0.14, −0.005]). Table 6 shows hypotheses testing for Study 2. The coefficients of direct effects are shown in Figure 5.

Figure 5. Coefficients of serial mediation Study2. Notes: N = 155; Unstandardized effects are presented N = 99; *p < .05, **p < .01, **p < .001.

Table 6. Hypotheses testing study.

Hypothesis	Pathway	b (SE)	95% CI
	Direct effects
H1.	Digital manipulation disclosure without disclosure > Recognition of the digital manipulation	.59 (.24)*	[.12, .1.06]
	Digital manipulation disclosure with disclosure > Recognition of the digital manipulation	1.66 (.24)**	[1.17, 2.12]
H2.	Digital manipulation disclosure with disclosure > Perceived SMI’s credibility	.07 (.19)	[−0.29, .45]
	Digital manipulation disclosure without disclosure > Perceived SMI’s credibility	.03 (.19)	[−0.29, .25]
H3.	Recognition of the digital manipulation > Perceived SMI’s credibility	.31 (.19)	[−0.17, .69]
H5.	Product involvement > Perceived SMI’s credibility	.30 (.14)*	[.02, .58]
H5.	Interaction (recognition of the digital manipulation × product involvement)	−0.08 (.03)*	[−0.15, .02]
H6a.	Perceived SMI’s credibility > Intention toward the SMI	.99 (.15)***	[.69, 1.29]
	Digital manipulation disclosure without disclosure > Intention toward the SMI	−0.04 (.29)	[−0.63, .54]
	Digital manipulation disclosure with disclosure > Intention toward the SMI	.19 (.34)	[−0.49, .87]
	Recognition of the Digital manipulation > Intention toward the SMI	−0.07 (.11)	[−0.28, .14]
H6b.	Perceived SMI’s credibility > Brand attitude	.43 (.10)***	[.23, .63]
	Digital manipulation disclosure without disclosure > Brand attitude	−0.54 (19)**	[−0.93, −0.15]
	Digital manipulation disclosure with disclosure > Brand attitude	−0.39 (23)	[−0.84, .05]
	Recognition of the digital manipulation > Brand attitude	.04 (.07)	[−0.09, .18]
H6c.	Perceived SMI’s credibility > eWOM	.43 (.15)**	[.12, .73]
	Digital manipulation disclosure without disclosure > eWOM	−0.64 (.30)*	[−1.23, −0.05]
	Digital manipulation disclosure with disclosure > eWOM	−0.26 (.35)	[−0.94, .42]
	Recognition of the digital manipulation > eWOM	.01 (.10)	[−0.19, .22]

CI: bias corrected 5000 bootstrap confidence interval.

Notes: N = 155. Unstandardized b coefficients (with boot SE between parentheses). *p < .05; **p < .01; ***p < .001.

Discussion

The main objective of our research was to explore the effectiveness of disclosing digitally enhanced pictures on SMIs’ branded posts. The discussion on the effectiveness of digital disclosures begins with participants’ ability to recall having seen the disclosure. Our studies have revealed that social media users face challenges in remembering encounters with digital disclosures. It is worth emphasizing that in Study 1, even though the disclosure was visibly integrated as a stamp on the branded posts with the wording ‘retouched person’, only approximately half of the participants were able to remember encountering the disclosure. This underscores the issue of recall and recognition surrounding digital disclosures used on social media. Even though the topic of mandatory disclosures for digitally enhanced pictures on SMI’s branded posts has been a subject for public debate in Germany, the country where Study 1 was conducted, such measures have not yet been implemented. Consequently, consumers are not as familiar with these types of disclosures as they are with advertising disclosures, which could explain why they struggle to remember encountering them.

Moreover, according to eye-tracking studies investigating sponsored news articles, significant proportions of users do not even look at sponsorship disclosures. Among those who do, only a fraction accurately identify that the disclosure denotes the content they are viewing as a paid message (Wojdynski and Evans 2016; Wojdynski et al. 2017). Despite the general acknowledgment of image editing and the likelihood that many advertising images have been digitally altered (Harrison and Hefner 2014), our findings indicate that the ability to detect photo manipulation solely through visual inspection is somewhat limited. Additionally, users are not accustomed to these types of disclosures as they are with advertising disclosures and do not remember seeing them.

Furthermore, if the disclosure was noticed it contributed to the recognition of digital manipulation. However, the disclosure had no significant direct or indirect impact on the perceived authenticity of the SMI nor on the SMI and brand-related outcomes. The subjective character of authenticity and social media culture’s evolving standards and ideals may explain these findings. On social media, digital manipulation has spread, with many users viewing it as a creative tool to express themselves and improve the visual attractiveness of their content. Social media users may perceive a SMI who uses digital manipulation as authentic because the altered images showcase the person’s unique aesthetic tastes and vision (Balaban and Szambolics 2022). Furthermore, social media users could place more value on aesthetics than strict adherence to unaltered reality, leading them to believe that digitally altered photographs accurately portray the SMI’s online persona.

We know from traditional media literacy literature that although users may be aware of certain negative aspects stemming from media content, such media content might still instinctively be appealing to them (Schreurs and Vandenbosch 2021). The findings indicating that an SMI is perceived as authentic, despite digitally altering her appearance and disclosing this, suggests that image processing is deeply ingrained in the consciousness of users, and it is not perceived as artificial or unrealistic.

Study 2 was conducted on UK female participants and the SMI endorsed skincare products. The disclosure in the form of a ‘retouched person’ stamp placed on the enhanced images, in accordance with the Norwegian legislation, was recalled by a higher percentage of participants than in the first study. This can be a result of a more prominent discussion on disclosing digital enhancements in the UK than in Germany, where Study 1 was conducted. However, we found evidence in both studies that disclosing digitally enhanced pictures in SMIs’ branded posts contribute to the recognition of the digital manipulation for the participants who remembered the wording of the disclosure. Therefore, visual disclosures of digitally manipulated branded SMIs’ posts are effective to a certain extent when noticed. Hence, our results contradict previous research on this topic (McComb, Gobin, and Mills 2021; Naderer, Peter, and Karsay 2021).

In Study 2, the high mean values for recognizing digital manipulation were also observed in the group where the pictures were not retouched. Although we carefully selected our stimulus materials and pre-tested the pictures. We included posts that showcased close-up images of the SMI with visible skin imperfections in the condition without digital manipulation and without a disclosure. Nevertheless, participants still assessed these pictures as digitally manipulated. These findings can be interpreted as an indication that social media users often believe that SMIs, in their role as advertising figures, commonly engage in the manipulation of their pictures.

In Study 2, we observed that both disclosing digital manipulation and recognizing digital manipulation do not directly impact SMI’s perceived credibility. However, we found evidence that product involvement has a moderating role in the relationship between recognizing digitally manipulated SMIs’ branded posts and the perceived credibility of the SMI. This finding is the main takeaway of the present research. SMIs are perceived as experts who usually create and post content on social media in a particular area of interest for the audiences (Campbell and Farrell 2020). Previous literature has shown that topic interest plays a crucial role when choosing to engage with an SMI’s generated content (Coco and Eckert 2020). Our results showed that social media users with high product involvement perceived the influencer as more credible. However, when users recognize digital alteration and experience high product involvement, the perceived credibility of the SMIs decreases. Previous research stressed that the awareness of the falsity of advertising manipulation negatively affects advertising outcomes (Campbell et al. 2022). Our findings showed that product involvement is a boundary condition to the negative SMI and brand-related advertising outcomes.

Our research confirmed the positive effects of SMI’s perceived credibility (Ki and Kim 2019; Lou and Yuan 2019; Munnukka, Uusitalo, and Toivonen 2016) and authenticity (Lee and Eastin 2021; SOURCE 1) on SMI and brand-related advertising outcomes. Thus, SMIs that are perceived as trustworthy and authentic are worthy of being followed. Furthermore, as users who perceive SMIs as authentic engage in eWOM, their persuasive power will grow.

Overall, our research contributes to a better understanding of the mechanism behind the effectiveness of disclosures used for digitally enhanced pictures. Additionally, we evaluate the impact of such disclosures on a cognitive pathway that encompasses the recognition of the disclosure itself and the perceived credibility and authenticity of SMIs.

Limitations and future research

It is essential to consider the findings of the present studies while acknowledging several limitations. First, both studies are single-exposure studies. Therefore, a possible limit is the brief exposure to disclaimer labels affixed to advertisements within one brief session. Moreover, we did not control for the time our participants spent watching the stimulus materials. Future research should consider repeated exposure to disclaimer labels over different sessions and the use of eye-tracking devices. Given that consumers are increasingly adept at seeing and possibly tolerating manipulated advertisements as their exposure to them increases (Campbell et al. 2022), longitudinal studies are needed.

Second, in our studies, we used well-known brands, such as Adidas as stimulus material for the first study, and CeraVe for the second study. Thus, preexisting attitudes towards the brands Adidas and CeraVe might have interfered with the objective assessment of the endorser, advertisement, and intentions. To further enhance the study’s scope, future research could explore the use of different brands not familiar to participants as stimulus material and examine how photo manipulation may impact consumer perceptions and attitudes towards them and the advertising agent.

Given that disclosing advertising is a mandatory practice in the countries where we conducted our research, we included an advertising disclosure on all SMI’s advertising posts. Looking at the interaction between sponsorship disclosure and digital manipulation disclosure could contribute to refining ethical standards in influencer marketing, ensuring that both forms of transparency effectively serve their intended purposes. As social media users are accustomed with advertising disclosures, research (Giuffredi-Kähr, Petrova, and Malär 2022; van Reijmersdal et al. 2023) showed that sponsorship transparency is not only appreciated but also contributes to positive advertising outcomes. Future research perspectives should investigate the effects associated with transparency concerning digital manipulation disclosures.

Third, while our first study was conducted on a sample including both males and females, the second study was conducted on an exclusively female sample. Therefore, the findings need to be interpreted under this limitation. Future studies must consider gender diverse samples.

Fourth, in the case of the branded posts that were digitally manipulated, surroundings might have influenced the results of our studies. Investigating the effects of the presence vs. the absence of surroundings in digitally manipulated pictures is a factor to be considered for future research.

Fifth, the participants originate from Germany and the UK, countries where public discussion regarding the mandatory disclosure of digitally enhanced images is ongoing, and therefore, they were not yet familiar with this practice. There is a need for future research to be conducted in countries where such disclosures are mandatory.

Sixth, the current research focuses on users’ perspectives. However, studying SMI’s views on ethical practices and transparency through photo edit disclosures in sponsored content should be a compelling opportunity for scholars to shape this emerging field.

Implications

Our research carries both theoretical and practical implications. Theoretically, these studies offer insights into the effectiveness of disclosures designed to signal the presence of retouched content in SMI posts. Both our studies showed that if noticed, a visually prevalent clearly worded label placed directly on the branded posts significantly contributes to the recognition of digital enhancement. Besides, our findings contribute to a better understanding of how digital manipulation disclosures on SMI’s branded posts relate to the perceived credibility and authenticity of SMIs. Hence, while the perceived authenticity of the SMI is not affected by the recognition of the digital manipulation disclosure, the SMI’s perceived credibility is affected in a negative way only in the case of high levels of product involvement.

This contributes to the broader digital manipulation framework (Campbell et al. 2022) and adds depth to the source credibility literature in the context of influencer marketing. Furthermore, our research contributes to the discussion surrounding digital manipulation disclosures within the framework of the PKM model by identifying a nuanced mechanism of conditional effects. When examining digital manipulation disclosures akin to how previous studies have analyzed advertising disclosures, we were not able to identify two paths, the positive impact of transparency and the negative effects of PK (Giuffredi-Kähr, Petrova, and Malär 2022; Karagür et al. 2022; van Reijmersdal et al. 2023). Thus, contrary to our expectations, we did not find any positive impact from transparently disclosing digital manipulation. Moreover, we highlighted that recognizing digital manipulation has a negative impact on the perceived credibility of SMIs, under the conditional effect of product involvement.

Our findings carry significance for policymakers, marketers, and SMIs. Despite our research revealing the limited effectiveness of disclosures, we recommend that SMIs incorporate them into their practices. Our research underscores the dual role of photo edit disclosures as a sign of transparency. We argue that these disclosures serve as a positive step toward openness in content creation, providing audiences with a clearer understanding of the editing process involved.

Moreover, policymakers should reevaluate the effectiveness of the specific disclosures examined in regulating digitally manipulated photos in SMIs’ posts and prioritize the development of more comprehensive disclosure strategies. For marketers and SMIs, our results demonstrate that transparency does indeed yield benefits for influencers, as it contributes to maintaining the perceived credibility and authenticity of their followers, as these factors play a pivotal role in influencing advertising outcomes.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The fictitious ads and the datasets in this article are available from the corresponding author upon reasonable request.

Additional information

Funding

This article was supported by the research funds of the Babeș-Bolyai University.

Appendix A.

Stimulus materials for study 1

Table

Digital manipulation without disclosure condition	Digital manipulation with disclosure condition (disclosure placed as a stamp on the picture, with the wording ‘Retuschierte Person’)

Appendix B.

Stimulus materials for study 2

Table

No digital manipulation condition	Digital manipulation without disclosure condition	Digital manipulation with disclosure condition (disclosure placed as a stamp on the picture, with the wording ‘Retouched Person’)