The impact of influencers' credibility and consumer involvement on attitudes and purchase intention: a comparison before and after the pandemic

The COVID-19 crisis was one of the most destructive events in recent times (Cruz-Cárdenas et al., 2021). Studies have reported that COVID-19 has reshaped consumer behaviour (Sheth, 2020; Zwanka and Buff, 2021), and consumers learnt new habits (Al-Geitany et al., 2023; Sheth, 2020). The extensive literature shows the impacts of the pandemic on consumer behaviour in two key ways. First, during the early stages of the COVID-19 pandemic and due to loss of control and fear (Koçak et al., 2021), consumers showed impulsive shopping (Wang et al., 2021) and unusual consumer behaviour such as hoarding toilet paper and food (Laato et al., 2020; Shojaei et al., 2024), purchasing essential commodities (Patil et al., 2022), and increasing the demand for health and hygiene products (Das et al., 2022). Sheth (2020) argued that hoarding, improvisation, and embracing digital technology were among the most noticeable immediate effects of the COVID-19 pandemic on consumption and consumer behaviour. Although impulsive shopping behaviour decreased after the initial stages of the pandemic, consumption of non-food categories dramatically decreased during COVID-19. According to the Global Consumer Insights Survey (2020), clothing and footwear sales experienced a significant decline of 51%, sales of sports equipment and outdoor gear were down by 46%, office equipment sales saw a decrease of 36%, and health and beauty product sales also experienced a decline of 35%. Within this context, the cosmetics and beauty industry faced significant challenges during the pandemic and lockdowns, and many customers reduced their beauty budget (Becker, 2021; Gerstell et al., 2020; Verma et al., 2022).

The shift towards online shopping was the second main change in consumers' shopping habits. Studies have found that the fear of the virus increased online shopping (Adibfar et al., 2022; Chopdar et al., 2022; Sheth, 2020). While in 2019, an estimated 53% of internet users engaged in online shopping, this number reached 60% in 2020 and 2021 (UNCTAD, 2022). Due to the rise of online shopping, 20–30% of businesses moved online during the pandemic's peak, and online grocery penetration settled at 9–12% at the end of 2020 (Aull et al., 2021). In the post-pandemic era, the number of online shoppers is still increasing. Tighe (2023) found that 68% of consumers worldwide shop online more often. In addition, the e-commerce market is expected to grow from a total of $6.3 trillion in 2023 to more than $8.1 trillion by 2026 (Baluch and Main, 2023).

Alongside the growth of online shopping, social media usage also increased during the pandemic, as people preferred to spend more time at home, their online activities increased (Adibfar et al., 2022). This resulted in increased consumption of social media (Aldrich, 2022; Turner and Ordonia, 2023) across all age groups (Aldrich, 2022). After the pandemic, consumers still actively seek information online, and social media platforms are a common source of information (Cruz-Cárdenas et al., 2021). With the increase in social media usage and online shopping, many brands decided to develop marketing activities through social media platforms (Lv et al., 2020). In particular, online retailers and services used social media influencers for their marketing strategies during the pandemic (Smith, 2022; Taylor, 2020b). Given their influence on their followers (Herrando and Martín-De Hoyos, 2022), brands increasingly use influencers in marketing strategies (Campbell and Farrell, 2020; Syrdal et al., 2023) to promote products or services (Eisend and Langner, 2010). Influencer marketing exceeded $21 billion globally in 2023 - more than triple its 2019 value (Statista, 2023a). Therefore, many firms allocate a significant portion of their marketing budgets to influencers for their participation in advertising campaigns (Hussain et al., 2024). As a result, research on influencer marketing has become an increasingly important and interesting topic in the field of marketing during and after the pandemic (Kim and Kim, 2022; Ruangwanit and Thongmak, 2025; Suri et al., 2023; Taylor, 2020a).

Despite the growing popularity of influencer marketing, a review of the effectiveness of influencers' endorsements by applying source credibility models (Hovland and Weiss, 1951; Ohanian, 1990, 1991) revealed that research in this area has presented mixed and inconclusive findings. For instance, some studies found a significant impact of attractiveness (Hani et al., 2018) and trustworthiness (Koay et al., 2022) on purchase intention, while other studies showed insignificant results regarding the effect of attractiveness (Lou and Yuan, 2019; Ohanian, 1991), trustworthiness (Magano et al., 2022; Ohanian, 1991) and expertise (Lou and Yuan, 2019) on purchase intention. Moreover, according to the Elaboration Likelihood Model (ELM), individuals may process persuasive messages through both central and peripheral routes simultaneously, particularly in high-involvement contexts such as web advertisements (Teng et al., 2014). Based on ELM, credibility can influence the likelihood of message processing via the central route. The model further suggests that influencer credibility plays a crucial role in individuals' decision-making processes, particularly at low levels of consumer involvement (Halder et al., 2021). However, there is limited empirical research applying ELM to investigate whether credible influencers can directly increase consumer involvement without the influence of intermediary variables.

In addition to the limited research on the effect of source credibility on consumer involvement, literature on consumer involvement itself highlights several gaps. First, most previous studies (Chavadi et al., 2021; Park and Keil, 2019) have primarily focused on the moderating role of consumer involvement(i.e. high vs. low-level product involvement), rather than its direct effects. Second, few studies have examined the direct relationship between consumer involvement and purchase intention; thus, the effect of consumer involvement on purchase intention needs to be clarified (Wu et al., 2022). Furthermore, although Ohanian (1990) (one of the significant researchers in the field of source credibility) suggested the need to study the mediating role of consumer involvement, a limited number of studies (Arora et al., 2020; McClure and Seock, 2020) have explored this mediating role, highlighting a critical gap in the existing literature.

On the other hand, attitude is one of the most critical factors in explaining people's behaviours. According to the Theory of Reasoned Action (Fishbein and Ajzen, 1975) and the Theory of Planned Behaviour (Ajzen, 1991), attitude is a significant predictor of consumer intention. This has been supported by numerous researchers across various fields such as tourism (Amaro and Duarte, 2015), technology usage (Silva et al., 2023), advertisements (Ho Nguyen et al., 2022; Izquierdo-Yusta et al., 2015), and eco-friendly products (Saepudin et al., 2023). The marketing literature shows that the studies on attitude have caught more academic attention than other concepts (Rizvi and Oney, 2018). Beliefs have the main influence on attitudes (Hill et al., 1977) in general, and more specifically, some evidence suggests that attitude is a prerequisite to behaviour (Rizvi and Oney, 2018). The present study focuses on two types of attitudes as follows: attitude towards the Ads on Instagram and attitude towards influencer-endorsed products. Although many studies explore consumer attitudes in influencer marketing, limited research distinguishes between these two specific attitudes. Moreover, there is a lack of comprehensive understanding of how the different dimensions of influencer credibility—expertise, trustworthiness, and attractiveness, as well as consumer involvement, directly and indirectly influence consumer attitudes. Understanding this distinction is crucial for better predicting purchase intentions and improving marketing strategies.

One of the main questions during the COVID-19 pandemic was “whether habits changed permanently.” This question was prevalent among society, the media (e.g. BBC, 2020; Euronews, 2020; Forbes, 2020), and academics (Broersma and Swart, 2022; Rodrigues et al., 2021). Particularly in the consumer behaviour domain, some researchers have claimed that “old habits die” (Kantur and Özcan, 2021; Sheth, 2020). Conversely, other researchers argued that the effect of the pandemic on online purchasing behaviour was temporary (Inoue and Todo, 2023). Despite the fact that numerous studies have examined consumer behaviour during COVID-19 (Chauhan et al., 2023; Yoon et al., 2023; Zwanka and Buff, 2021), understanding consumers' behaviour regarding beauty and cosmetic products in the post-pandemic world is still unaddressed by research. Along with doubt about changes in consumer habits, the current study investigates the direct and indirect impact of influencer credibility on consumer involvement towards attitudes and purchase intention. More specifically, this paper seeks to determine whether these relationships have changed after the COVID-19 pandemic. To achieve this, the study empirically compares these relationships before and after the pandemic.

The remainder of the article is organized as follows. The next section presents the literature review, which is followed by a description of the adopted methodology. After that, the research findings, conclusions, implications, and future directions are presented.

Understanding the effectiveness of endorsers is critical for both practitioners and academics (Till and Busler, 2000). Studies on the effectiveness of influencer endorsement have primarily evolved from four theoretical approaches, including, (1) the “source credibility model” (expertise and trustworthiness) (Hovland and Weiss, 1951), (2) the “match-up hypothesis/theory” (fit between endorser and product and consumer) (Kahle and Homer, 1985; Till and Busler, 1998), (3) the “source attractiveness model” (“familiarity”, “likeability”, “similarity”, and “attractiveness”) (McGuire, 1985), and (4) the “Meaning Transfer Model” (“stems from cultural meanings with which they are endowed” (McCracken, 1989). The primary model of “source credibility theory” (Hovland and Weiss, 1951) included the variables “expertness” and “trustworthiness”. Later, Ohanian (1990, 1991) expanded the model and introduced attractiveness as an additional dimension of source credibility. According to Ohanian (1990, 1991), “attractiveness”, “trustworthiness”, and “expertise” are key elements of endorsement effectiveness.

The concept of attractiveness has captured the attention of both practitioners and academics in the field of marketing. Many authors have sought to determine whether the attractiveness (or ineffectiveness) of influencers affects consumers' attitudes (Durau et al., 2022; Magano et al., 2022; Ruangwanit and Thongmak, 2025; Taillon et al., 2020). The literature indicated that using physically attractive influencers (or models) in advertising is effective and makes a positive impact on firm value (Kang et al., 2019), advertising evaluations (Baker and Churchill, 1977), and product evaluations (Joseph, 1982). Studies investigating the effect of attractiveness on purchase intentions have shown conflicting results. Many researchers have highlighted that attractiveness impacts purchase intention (Caballero and Solomon, 1984; Hani et al., 2018), but Lou and Yuan (2019) and Ohanian (1991) found insignificant results regarding the effect of the attractiveness of influencers on purchase intention.

In addition, previous studies found that trustworthiness is one important component of influencers' perceived credibility (Balaban and Mustăț;ea, 2019; Lou and Yuan, 2019; Munnukka et al., 2016). The influence of trustworthiness on the opinions and behaviours of the audience has been confirmed by several studies (Lou and Yuan, 2019; Ohanian, 1991; Osei-Frimpong et al., 2019). Although trustworthiness is considered essential in influencer endorsements, previous studies indicated that trustworthiness has an insignificant influence on purchase intention, as demonstrated by Magano et al. (2022) and Ohanian (1991), and also on customer attitudes (Al Mamun et al., 2023).

Expertise describes “the source's level of knowledge” (Wiedmann and von Mettenheim, 2020, p. 709). According to the “match-up hypothesis/theory” (Kahle and Homer, 1985; Till and Busler, 1998), endorser expertise plays a major role in shaping the compatibility between an endorser and a product. Within the context of consumer behaviour, empirical studies investigating the impact of influencer expertise have resulted in mixed findings. For example, a study conducted by Magano et al. (2022) showed that while influencer expertise has a positive influence on satisfaction, its impact on brand image and brand trust was non-significant. Similarly, Cheung et al. (2008) and Lou and Yuan (2019) did not find a significant impact of source expertise on information usefulness. Conversely, Ohanian (1991) found a significant effect of expertise on purchase intention.

The model was originally developed by Petty and Cacioppo (1979). ELM describes the process by which advertisements lead to the persuasion of individuals, leading to the development or modification of their attitudes or behaviour (Segev and Fernandes, 2023). According to ELM theory, two routes influence consumer decisions: the central and peripheral routes. The central route involves a high level of cognitive engagement, where consumers carefully evaluate the quality and strength of the message content, such as detailed product information or logical arguments. In contrast, the peripheral route depends on external cues like the influencer's attractiveness, credibility, or likability, without scrutinizing the message content (Hossain et al., 2025; Saad et al., 2025). ELM highlights that when information is presented to individuals (namely, consumers), they exhibit different reactions, which are influenced by the degree of cognitive effort they use in processing the given information (Yoo et al., 2017). This means that, for instance, when a person receives information in an unpleasant situation, the resulting reaction is adverse, and vice versa (Tampoli et al., 2021).

Several researchers have utilized ELM in the marketing field, including to study advertising (Navarro et al., 2009), social media (Putra et al., 2020), content analysis of viral advertising on social media (Segev and Fernandes, 2023), electronic word of mouth (Ismagilova et al., 2021; Li and See-To, 2024), and content credibility (Abbasi et al., 2023). Particularly, ELM has been applied as the theoretical foundation by researchers to predict customers' involvement (Luo et al., 2024), source credibility and involvement (El Hedhli et al., 2021; Li and See-To, 2024), and the impact of endorsements on consumers (Schaefer and Keillor, 1997).

Given this study's focus on social media, ELM is particularly relevant, as it explains how both message content and source characteristics influence consumer decision-making (Li and See-To, 2024). In addition, the integration of the Source Credibility theory and the Elaboration Likelihood Model provides a wider understanding of the impact of influencers on their followers' purchase intentions. While Source Credibility Theory focuses on the importance of the influencer's credibility-specifically “attractiveness”, “trustworthiness”, and “expertise” - in shaping consumer behaviour, ELM explains how this influence occurs through the consumer's message processing, with involvement serving as a key factor (Añaña and Barbosa, 2023). Within this context, Luo et al. (2024) stated that source credibility (e.g. expertise and trustworthiness), typically associated with the peripheral route, plays a key role in ELM. ELM focuses on how influencers' credibility (source credibility) affects the attitudes and behaviours of consumers (Chavadi et al., 2021). Considering behavioural changes due to the COVID-19 pandemic, consumers of beauty products are likely to carefully evaluate the information presented by their sources, including influencers, indicating that consumers may engage in central route processing when evaluating influencer content more carefully due to increased concern for accuracy and trust.

The term “involvement” emerged in the field of psychology to describe an individual's level of involvement, and it originated from the notion of “self-involvement” (Sherif and Cantril, 1947). Krugman (1965) introduced involvement theory into marketing research and studied the consumer decision process. Within this context, involvement has been acknowledged as one of the most important elements for achieving business success, and it has led to a new era in marketing (Bateman and Valentine, 2021; Neuhofer, 2016). The concept of “involvement” was applied in various categories such as high and low involvement purchases (Assael, 1981), “enduring involvement” (the link between a consumer's values with prior experience with a product) (Houston and Rothschild, 1978), and involvement types (i.e. interactions with a product, a brand, and its symbol) (Laurent and Kapferer, 1985), situational involvement (involvement forces on short-term which is depend on a particular situation) (Ferns and Walls, 2012), and Product class level involvement can endure over time and not dependent on a particular purchase situation (Bateman and Valentine, 2021; Beckman et al., 2020; Dholakia, 2001). To keep the scope of the research manageable, the study incorporated three key components of involvement, as outlined by Cass and O’Cass (2000): “purchase decision involvement, product involvement, and consumption involvement.”

According to Beatty et al. (1988), purchase involvement refers to “the level of concern for, or interest in, the purchase process triggered by the need to consider a particular purchase”. (Mittal and Lee, 1989) argued that symbolic, hedonic, and risk values are three key factors underpinned by purchase involvement. Customers with more purchasing involvement tend to gather and use information more frequently. According to Bateman and Valentine (2021), consumer purchase involvement is related to cognitive and emotional dimensions; it indicates the individual's cognitive involvement in the purchase decision process, which is influenced by their emotional involvement (Putrevu and Lord, 1994).

“Product involvement” refers to an individual's involvement with an item due to its inherent values, essentiality, and appeal (Zaichkowsky, 1985, 1994). Researchers have investigated how product involvement affects consumer choices and purchase intentions (Bian and Moutinho, 2011; Strubel and Petrie, 2016) and have confirmed that a high level of product involvement is associated with a higher likelihood of intention to purchase (Strubel and Petrie, 2016).

Finally, regarding consumption involvement, unlike the purchase decision and product involvement, consumption involvement has received only limited attention. According to Cass and O’Cass (2000), the concept of consumption involvement relates to the extent to which a consumer is involved in the consumption of a product or service. Khare et al. (2020) found that consumption involvement significantly influenced purchase behaviour. Arguably, consumption involvement can only fully develop in a post-purchase stage, and therefore, the tridimensional conceptualization of consumer involvement developed and validated by Cass and O’Cass (2000) may depend on effective experience with the products or services being analyzed.

According to Lou and Yuan (2019), a social media influencer is “first and foremost a content generator: one who has a status of expertise in a specific area, who has cultivated a sizable number of captive followers-who are of marketing value to brands-by regularly producing valuable content via social media (p.2)”. In addition, Moreno et al. (2015, p. 246) describe influencers as “opinion leaders who can use their online platforms to diffuse information and affect the attitudes and behaviours of their audiences.” Likewise, (Engel et al., 2024, p. 3) define influencers as “online users who have a significant number of followers, regularly share content, and possess the ability to influence others.” In line with these definitions, and in the case of Instagram, this study defines an influencer as an Instagram user with a large number of followers who consistently shares content related to products and brands (e.g. cosmetic products), and has the ability to influence the opinions, attitudes, and purchase decisions of his/her followers.

Influencers often can affect the opinion of followers and shape audience attitudes (Freberg et al., 2011). Due to influencers' expertise in particular fields, customers typically appreciate the advice given by them (De Veirman et al., 2017). Previous studies indicated that influencer-product congruence (Belanche et al., 2021) and physical attractiveness (Silvera and Austad, 2004) positively impact attitudes toward influencers' advertising. In contrast, researchers found an insignificant influence of attractiveness (Ohanian, 1991), trustworthiness (Al Mamun et al., 2023; Ohanian, 1991), and expertise (Cheung et al., 2008) on the customer attitude and intention to purchase.

The present study focuses on two types of attitudes as follows: attitude towards the Ads on Instagram and attitude towards influencer-endorsed products. Attitude towards the Ads is one of the significant indicators of advertising effectiveness (Wang and Sun, 2010). Attitude toward advertising is defined as a “predisposition to respond favourably or unfavourably to a particular advertising stimulus during a particular exposure occasion” (Ho Nguyen et al., 2022, p. 4). In line with the study of Belanche et al. (2021) and Magano et al. (2022), we defined attitude towards influencer-endorsed products as a follower's positive (negative) feelings towards advertising and/or product endorsement by an influencer on Instagram.

The existing literature showed a strong impact of advertiser credibility (MacKenzie and Lutz, 1989) and endorser credibility (Lafferty and Goldsmith, 1999) on attitude towards Ads. Furthermore, studies revealed that credibility has a significant influence on attitudes toward Ads (Al Mamun et al., 2023; Ho Nguyen et al., 2022; Thomas and Johnson, 2017). In addition, previous studies found that the credibility of an influencer (Belanche et al., 2021; Magano et al., 2022; Taillon et al., 2020) plays a significant role in attitude towards influencer-endorsed products. Based on these contributions, it is expected that:

Most influencers share their lifestyle with their audiences. The success of influencers depends on their strong relationship with followers (Dhanesh and Duthler, 2019), which impacts the attitudes and intentions of their followers (Durau et al., 2022). Overall, purchase intention reflects consumers' preference for a brand or product (Al Mamun et al., 2023) and refers to the “possibility that consumers will plan or be willing to purchase a certain product or service in the future” (Wu et al., 2011, p. 32). Empirical studies generally emphasize that an increase in purchase intention enhances the likelihood of behaviour (Dodds et al., 1991; Schiffman and Kanuk, 2007). In line with this, previous research demonstrates the influence of the credibility of the endorser on intentions (Amos et al., 2008; Munnukka et al., 2016). Likewise, several empirical studies support the positive impact of source credibility dimensions on consumers' intention to purchase products or services (Osei-Frimpong et al., 2019; Thomas and Johnson, 2017). Hence, it is hypothesized that:

Involvement is one of the significant concepts in the study of product-consumer relationships and purchase behaviour (Mao and Zhang, 2013). According to Schaefer and Keillor (1997), advertisers dealing with high-involvement products should consider choosing endorsers that are well-matched with their products because source attractiveness creates persuasive effects in situations where individuals have low (high) involvement through a mechanism of liking (El Hedhli et al., 2021), as well as other reasons such as qualification, success, values, and conduct (McGuire, 1985). In addition, influencer attractiveness appeals to customers, thereby increasing consumer involvement (Liu, 2022). Studies have demonstrated that when consumers have positive feelings for an influencer, they are more likely to develop positive and favourable feelings about the brand endorsed by that influencer (Min et al., 2019), which increases consumer involvement. Furthermore, studies indicated that there is a relationship between influencers and customer engagement (Qiu et al., 2021) and product involvement (Jiang et al., 2022). However, some studies (Eisend and Langner, 2010) found that neither attractiveness nor expertise had a significant effect on involvement (control variable). Based on the literature, it is hypothesized that:

ELM can clarify how consumer involvement influences attitude in social media. According to ELM research, Abbasi et al. (2023) confirmed that involvement has a positive effect on attitudes and purchase intention. Considering ELM, consumers with high involvement tend to search for more information about a product, which might impact their attitude (Das and Ramalingam, 2022). The literature includes numerous studies that have examined the moderating effect of different types of consumer involvement (i.e. product involvement) (Belanche et al., 2017; Chavadi et al., 2021; Park and Keil, 2019). However, few studies have examined the direct effects of consumer involvement on consumer behaviour. Researchers argued that consumer involvement has a direct impact on attitudes (brand and advertisement) (Xue and Phelps, 2013). Thus, the following hypotheses are formulated:

Involvement plays a significant role in consumer behaviour (Lin and Chen, 2006). Bateman and Valentine (2021) argued that involvement encompasses various levels of cognitive and affective aspects at different purchase stages. A higher level of consumer involvement may lead to higher purchase intention (Xue and Phelps, 2013). However, high involvement may lead to a sense of risk in the consumer decision-making process (Atkinson and Rosenthal, 2014). Overall, studies indicated that involvement positively affects purchase intention (Bian and Moutinho, 2011; Chang and Chen, 2022). Therefore, it is expected that:

For several decades, researchers have examined attitudes toward advertising based on the studies conducted by Mitchell and Olson (1981) and Shimp (1981). These studies indicate that attitude toward ads impacts purchase intentions (Ho Nguyen et al., 2022; Thomas and Johnson, 2017) and consumer behaviour (Ho Nguyen et al., 2022). Previous studies have suggested that a positive attitude toward influencers (Durau et al., 2022; Min et al., 2019; Taillon et al., 2020) and toward ads (Al Mamun et al., 2023; Durau et al., 2022; Ho Nguyen et al., 2022; Thomas and Johnson, 2017) positively affect purchase intention. Thus, it is expected that:

A review of existing literature indicated that most of the studies have investigated the direct effect of credibility on purchase intention. In this study, we examine the indirect effect of influencers' credibility on purchase intention through consumer involvement. Following the suggestion of Ohanian (1990) and the studies of Arora et al. (2020) and McClure and Seock (2020), the following hypothesis is proposed:

Figure 1 summarizes the hypotheses defined for this study.

The current study employed a repeated cross-sectional design, consisting of two iterations of an online survey. It relied on analyses of two consumer samples. The first sample, collected in November 2019 (pre-pandemic), consisted of 297 participants, while the second sample, collected in 2023 (post-pandemic), comprised 307 participants. Data were collected through an online survey targeting Portuguese female consumers of beauty products, aged 18 and above, who are active on Instagram. While both women and men use beautification strategies, women tend to spend more on beauty products. Studies have shown that in 2021, the female segment accounted for the highest revenue share at over 62.2% due to grooming demand for personal care products (Deb, 2025), and it is anticipated that women will capture 53.73% of the market share in 2025 (Fortune Business Insights, 2025). In addition, previous studies, such as the one conducted by Hermans et al. (2024), demonstrate that women are the most interested among cosmetics consumers and consequently the most willing to participate (in their case, 96.6% of the sample were female). Therefore, the current study focuses on female respondents to better understand the consumption behaviours of this relevant consumer segment. Due to the design of the online questionnaire, the collected responses contained no missing data. A total of 604 questionnaires were analyzed using SPSS 25 and IBM AMOS software. The respondents' demographics are reported in  Appendix 1.

The decision to focus on Instagram and beauty products was based on several factors. Firstly, COVID-19 has significantly increased the use of social media, particularly Instagram (Sheth, 2020). Secondly, while Instagram is considered the most popular influencer marketing platform and 68% of companies have adopted Instagram for influencer marketing (Statista, 2023b), previous studies have focused on Facebook users (McClure and Seock, 2020), TikTok (Barta et al., 2023), and Twitter (Lahuerta-Otero and Cordero-Gutiérrez, 2016). Few studies have been conducted among Instagram users. Thirdly, due to its visual nature and filtered images, this platform has become a popular social media for beauty brands (Jin et al., 2019). Finally, while many customers reduced their beauty budget during COVID-19 (Becker, 2021; Gerstell et al., 2020; Verma et al., 2022), the market size of cosmetics retailers in Portugal has grown 4.5% per year on average between 2019 and 2024 (IBISworld, 2022).

The literature review reveals that influencers are commonly categorized based on their number of followers (e.g. micro, macro, and mega-influencers) (Conde and Casais, 2023), their focus on specific topics (e.g. food influencers) (Hendriks et al., 2020), and the presence of influencers on particular social media platforms (Engel et al., 2024). This study specifically focused on cosmetics Influencers on Instagram and without considering the number of followers. To ensure clarity for respondents, we initially explained that the questionnaire was aimed at consumers of beauty products (cosmetics and personal care products) who followed “beauty influencers” on Instagram, i.e. influencers who discussed beauty care and recommended certain beauty products or brands. To participate in the study, the respondent had to provide a positive answer to the question “Do you follow any influencers on Instagram who regularly talk about beauty, including cosmetics brands?”. Afterwards, the participant was requested to identify one influencer of their choice that matched the criteria and then answer the following questions regarding the influencer they chose.

Regarding measurement, the influencer credibility scale was adapted from Ohanian (1990) and comprises three dimensions. Some items were reverse-coded so that 1 corresponded to the minimum value of credibility dimensions (i.e. attractiveness, trustworthiness, and expertise) and 5 to the maximum. The three-dimensional consumer involvement scale was adapted from the one previously developed and validated by Cass and O’Cass (2000). Attitude towards influencer-endorsed products and attitude towards ads were adapted from Speck and Elliott (1997), and purchase intention was measured with a scale originally developed by Dodds et al. (1991), which has been recurrently adapted to new contexts over the years. Responses for the items of consumer involvement, attitudes, and purchase intention ranged from 1 “completely disagree” to 5 “completely agree”. The list of all questionnaire items is presented in Table 1. In this study, beauty products refer to cosmetics and personal care products, as it was carefully explained to participants.

The Skewness and Kurtosis values of all the items ranged between −2 to +2, which confirmed the assumption of normality (George and Mallery, 2010). In addition, the Common Method Bias (CMB) was evaluated in both datasets (pre-pandemic and post-pandemic) using Harman's single-factor test (Podsakoff et al., 2003). The findings revealed that the percent of the accumulated variance by the first factor is 33.8% for the pre-pandemic dataset, 30.8% in the post-pandemic dataset, and 31.5% in the combined sample. These values are all less than the threshold of 50%, suggesting that the likelihood of CMB influencing this research is low.

The confirmatory factor analysis (CFA) was conducted in AMOS ( Appendix 2). In this study, influencers' credibility (including attractiveness, trustworthiness, and expertise) and consumer involvement (including purchase decision involvement, product involvement, and consumption involvement) are treated as second-order constructs. The results of both models (first-order and higher-order factors) are presented in Table 2. In this step, due to the low reliability, six items (in your opinion, advertising on Instagram is … excessive (Atad3), annoying (Atad4), Wastes my time (Atad6); in your opinion, the advertising that influencers' do for beauty products is … excessive (Atb3), annoying (Atb 4), and waste my time (Atb6)) were excluded from the measurement model.

As indicated in Table 2, the overall model fit statistics for both measurement models were adequate. The values of standardized loadings for all remaining items exceed the cut-off point of 0.5 (Chin, 1998). The results of Cronbach's Alpha and composite reliability (CR) for all constructs were above 0.70 (Bagozzi and Youjae Yi, 1988; Sekaran, 1993), supporting acceptable reliability for both models. The results show that the values for average variance extracted (AVE) for both measurement models exceed the 0.50 threshold value (Fornell and Larcker, 1981), indicating convergent validity of the constructs. In addition, the Fornell-Larcker criterion (Fornell and Larcker, 1981) was employed to assess discriminant validity. The square root of the AVE values (see the diagonal in Table 3) is higher than the inter-factor correlations. Thus, the discriminant validity of both models was confirmed.

To compare the two sets of data (before and after the pandemic) and to ensure the validity of the study's results and conclusions, establishing measurement invariance is essential (Millsap, 2012). Thus three-step measurement invariance (configural, metric, and scalar invariance) test was assessed through chi-square tests and goodness-of-fit indexes (e.g. CFI, RMSEA).

As shown in Table 4, the initial model assessed configural invariance for both groups (pre-pandemic and post-pandemic), which is tested by evaluating the overall fit of the model (Putnick and Bornstein, 2016). The configural invariance had an acceptable fit (χ² (df) = 1577 (738), CFI = 0.95 and RMSEA = 0.04), indicating that the model is configurally invariant. In addition, metric and scalar invariance were evaluated through chi-squared difference tests (Δ Chi-Square) and using a threshold of −0.01 for CFI and RMSEA changes (Putnick and Bornstein, 2016). As indicated in Table 4, results show that the chi-square difference between configural and metric invariance is not statistically significant (Δ χ² (df) = 29.3 (21), p > 0.01). In addition, the changes in CFI and RMSEA are below the threshold of 0.01. Thus, full metric invariance was established. Following the metric invariance, the scalar invariance was evaluated. Due to the significant chi-square difference (Δ χ² (df) = 54.4 (30), p < 0.01), the full Scalar invariance (M3) was not supported. Therefore, for achieving partial-scalar invariance, two items (AT2: beautiful–ugly and AT2: Sexy –Not sexy) were allowed to be freely estimated in both groups. The partial-scalar invariant model (M3a) is accepted as the chi-square is not significant (Δχ2 (df) = 40.1 (28), p > 0.05). Thus, the results indicated sufficient quality and invariance for comparing samples before and after the pandemic.

After the measurement model met the validity requirements, the set of hypotheses was tested. As indicated in Figure 2, the SEM model provided a good fit to the data (χ2 = 1184, df = 390, χ²/df = 3.04, CFI = 0.95, TLI = 0.95, IFI = 0.95, and RMSEA = 0.06).

As illustrated in Figure 2 and considering the guideline of Cohen (1998) for R² (0.26 “substantial”, 0.13 “moderate”, 0.02 “weak”), although the R² for consumer involvement is weak (R² = 0.05), the R² values for attitude towards influencer-endorsed products (0.37), attitude towards ads (0.18), and purchase intention (0.64) are substantial.

As indicated in Table 5, the results of H1, H2, and H4, which predicted an effect of influencers' credibility on attitude towards ads (H1: β = 0.09, t = 1.97, p < 0.05), attitude towards influencer-endorsed products (H2: β = 0.35, t = 7.69, p < 0.01), and consumer involvement (H4: β = 0.22, t = 4.68, p < 0.01) were confirmed. H3 was not supported since there is a non-significant relationship between influencers' credibility and purchase intention (H3: β = 0.06, t = 1.79, p > 0.05). Similarly, the anticipated positive effect of Consumer Involvement on attitude towards ads (H5: β = 0.39, t = 8.21, p < 0.01), attitude towards influencer-endorsed products (H6: β = 0.43, t = 10.46, p < 0.01), and purchase intention (H7: β = 0.26, t = 6.36, p < 0.01) were supported. The impact of attitude towards ads (β8 = 0.08, t = 2.29, p < 0.05) and attitude towards influencer-endorsed products (H9: β = 0.58, t = 14, p < 0.01 on purchase intention was also supported.

Finally, the mediation relationships posited were evaluated using a bootstrapping procedure involving 5000 bootstrap resamples (Hair et al., 2011). As shown in Table 6, the indirect impact of influencers' credibility on purchase intention (H10a: B = 0.11, p < 0.01), attitude towards ads (H10b: B = 0.13, p < 0.05), and attitude towards influencer-endorsed products (H10c: B = 0.16, p < 0.01) through consumer involvement is significant.

The comparison between the two samples of the study was evaluated using the chi-square difference (Δχ2) test between the unconstrained models (without imposing any equality constraints) with a series of constrained models for each of the hypothesized paths (See Table 7).

As shown in Table 7, out of four hypothesized paths for the impact of influencers' credibility, the path to attitude towards influencer-endorsed products is statistically significantly different across the two samples (Δ χ² (df) = 6.5 (1), p ≤ 0.05). The regression coefficients before the pandemic (β = 0.50) indicate a stronger impact compared to after the pandemic (β = 0.25) in terms of the effect of influencers' credibility on attitude towards influencer-endorsed products. In addition, the remaining paths (influencers' credibility on Attitude towards ads, consumer involvement, and purchase intention) did not significantly differ across the two samples.

The next paths that were examined were the ones connecting consumer involvement with attitude towards ads, attitude towards influencer-endorsed products, and purchase intention. As indicated in Table 7, while the p-value of the chi-square difference test for consumer involvement and attitude towards ads and purchase intention is not significant, the consumer involvement on attitude towards influencer-endorsed products (Δ χ² (df) = 10.8 (1), p ≤ 0.01) is statistically significant. The regression coefficients before the pandemic (β = 0.29) indicate a weaker effect compared to after the pandemic (β = 0.54) regarding the influence of consumer involvement on attitude towards influencer-endorsed products. Finally, the last two paths analyzed were the link between attitudes (toward ads and influencer-endorsed products) and purchase intention. The chi-square differences for both paths are not statistically significant.

The current study evaluated the impact of influencers' credibility on consumer involvement, attitudes, and purchase intention, both directly and indirectly through consumer involvement among users of beauty products. The study provided valuable insights into the impact of influencers' credibility and consumer involvement. The findings confirmed the positive effect of influencers' credibility on attitudes toward ads. Studies revealed that credibility has a significant influence on attitude toward the ads (Al Mamun et al., 2023; Ho Nguyen et al., 2022; Thomas and Johnson, 2017), attitude towards influencer-endorsed products (Belanche et al., 2021; Magano et al., 2022; Min et al., 2019; Taillon et al., 2020), and consumer involvement (Caballero and Solomon, 1984). Despite the positive link between influencer credibility and purchase intention, as confirmed by previous research (Osei-Frimpong et al., 2019; Thomas and Johnson, 2017), the results demonstrate that influencers' credibility has not affected purchase intention in both samples (before and after the pandemic). The insignificant result could be attributed to the fact that influencers may not always successfully endorse products and campaigns (Jun et al., 2023). Moreover, considering the “match-up hypothesis/theory” (Kahle and Homer, 1985; Till and Busler, 1998), influencer–object match matters when it comes to the purchasing decision process (Knoll and Matthes, 2017). Thus, this unexpected finding may be due to the fact that the authors of this study have not considered the match-up between the influencers and the beauty products. In addition, some studies show that the use of influencers does not add considerable value to an advertisement (Sliburyte, 2009). In the context of source credibility, previous researchers found an insignificant influence of credibility (Ohanian, 1991), brand credibility (Jeng, 2016), and the credibility of authenticity claims (Kim and Song, 2020) on purchase intention.

While most of the prior researchers are focused on the moderating effect of different types of consumer involvement (Chavadi et al., 2021; Park and Keil, 2019), this study examined the direct and indirect effects of consumer involvement on attitudes and purchase intention. Considering the findings of (Belanche et al., 2017; Chavadi et al., 2021; Park and Keil, 2019; Wu and Wang, 2011), the findings highlighted the direct impact of consumer involvement on consumer attitudes and purchase intention. In addition, the mediation role of consumer involvement between influencers' credibility and attitude towards Ads, attitude towards influencer-endorsed products, and purchase Intention is supported (Arora et al., 2020; McClure and Seock, 2020). This study confirmed the positive impact of attitude towards influencer-endorsed products (Durau et al., 2022; Min et al., 2019; Taillon et al., 2020), and towards ads (Al Mamun et al., 2023; Durau et al., 2022; Ho Nguyen et al., 2022; Thomas and Johnson, 2017) on purchase intention.

Furthermore, the study compared structural paths across two samples from before and after the pandemic. The study found significant differences in the impact of influencers' credibility on attitude towards influencer-endorsed products. The findings suggest that influencers' credibility had a weaker effect on attitude towards influencer-endorsed products after the pandemic. The reduction may have occurred due to beauty and cosmetic products consumers returning to their pre-pandemic shopping habits from physical stores. Also, consumers of beauty products may be aware that influencers are paid for endorsements (Rossiter and Smidts, 2012), which has decreased the effectiveness of influencers on their attitude. In addition, negative actions by some influencers during the pandemic (Yoosefi Lebni et al., 2022) may have contributed to a decline in their credibility, leading to a reduced influence on attitudes towards influencer-endorsed products.

Furthermore, the impact of consumer involvement on attitudes towards influencers' advertisements was significantly different. Considering the elaboration likelihood model (ELM), involvement enables customers to be more proficient in processing information, and a high degree of consumer involvement with endorsed products results in an increased focus on the advertisement and enhanced processing of commercial information. Consequently, the findings of this study indicate that consumer involvement had a strong influence on attitudes toward influencer-endorsed products after the pandemic.

The findings of this study contribute significantly to the literature in several ways. Literature reviews indicated that the Elaboration Likelihood Model (ELM) is useful for explaining information processing, especially in the social media environment. In this context, social media influencers provide comprehensive information about various products to consumers (Calvo-Porral et al., 2021), which is important for enhancing consumer involvement. According to ELM, involvement makes consumers more motivated and able to process information (Belanche et al., 2017). In other words, when consumers are actively involved with influencer content (as seen in this study after the pandemic), they are more likely to pay attention to the information provided by the influencer, leading to greater influence on their attitudes and purchase intentions. Thus, this study extends the theoretical understanding of ELM by examining the direct effect of influencers' credibility on consumer involvement.

Another theoretical implication of this study is the focus on the role of consumer involvement. While most previous research has explored the interaction effect of consumer involvement (low vs. high involvement), this study developed the model and hypotheses based on the mediation effect, which has been addressed by only a few researchers (Arora et al., 2020; McClure and Seock, 2020). The findings of an insignificant direct effect of influencer credibility on purchase intention, coupled with a significant indirect effect through consumer involvement, highlight the critical importance of considering consumer involvement for a more comprehensive understanding of influencer marketing effectiveness.

One strength of this study is the availability of data from both before and after the COVID-19 pandemic, enabling a repeated cross-sectional analysis to identify changes in consumer behaviours and habits across these distinct periods. Hence, this research contributes to the growing body of literature that examines post-pandemic changes in consumer behaviour. Considering the question of whether COVID-19 changed habits permanently, despite some studies suggesting that COVID-19 changed consumer behaviour in terms of attitudes and purchases, the findings of this study suggest that the influence of attitudes on purchase intention has remained consistent with pre-pandemic patterns. This observation could imply that consumers have reverted to their previous habits, marking a significant advancement in post-pandemic consumer behaviour research.

In addition, among the hypothesized paths, only two paths (Influencers' credibility and Consumer involvement to Attitude toward influencer-endorsed products) demonstrated statistically significant differences across the pre- and post-pandemic contexts. Findings showed that the impact of influencers' credibility on consumers' attitudes toward influencer-endorsed products has declined. This study contributes to the literature on influencers' credibility in marketing by revealing that the effectiveness of influencers' credibility may decline based on social and environmental phenomena, and it might not be an effective predictor of all types of consumer attitudes. The declining influence of influencer credibility on consumer attitudes suggests a possible shift in how consumers evaluate trustworthiness, expertise, or authenticity in post-pandemic digital environments. In contrast, the impact of consumer involvement on these attitudes has increased, highlighting a growing role of personal relevance and engagement in shaping consumer perceptions and favourable attitudes toward endorsed products.

Lastly, this study makes a methodological contribution by treating consumer involvement as a higher-order factor, including three key components of involvement (Cass and O’Cass, 2000), despite the fact that previous studies have primarily focused on specific dimensions of involvement (e.g. product or purchase involvement). It demonstrates the viability of adopting a holistic view provided by the second-order factor, which, arguably, provides a more robust view of the consumers' perspectives.

This study provides several managerial implications to improve marketing strategies for beauty products, especially concerning social media influencer endorsements and consumer involvement. The findings showed a positive direct and indirect impact of influencers' credibility on consumer involvement and attitudes. This result confirms that, although influencer endorsement is expensive and risky, companies may consider this strategy as a relevant marketing strategy. Brands should focus on building and maintaining influencer credibility as it significantly impacts consumer attitude and involvement, which are key drivers of purchase decisions.

In addition, the results showed that consumer involvement has a strong positive effect on attitudes toward ads, attitudes toward influencer-endorsed products, and purchase intention. This suggests that brands should design campaigns that actively engage consumers and increase their involvement with both the advertising content and the endorsed products. Enhancing consumer involvement may lead to more favourable attitudes and ultimately higher purchase intentions, making it a critical focus for effective influencer marketing strategies.

Moreover, the significant indirect effect of influencers' credibility on purchase intention through consumer involvement suggests that, by developing marketing strategies focussing on both influencers' credibility and consumer involvement, managers can enhance consumers' purchase intentions.

Furthermore, the current study has indicated that the impact of influencers' credibility on attitudes toward influencer-endorsed products has decreased after the pandemic. Therefore, marketers should consider relying solely on credibility cues such as expertise or attractiveness. Instead, it is essential to prioritize strategies that enhance consumer involvement to build stronger positive attitudes and drive purchase intentions.

Finally, the findings from post-pandemic data showed that influencer credibility does not influence attitude towards ads on Instagram. This has significant managerial implications, as it demonstrates that relying solely on influencers' credibility may not be an effective strategy for all types of consumer attitudes. Managers may consider alternative strategies to influence the specific types of attitudes.

This study has some limitations and suggestions for future research. In this study, we focused on beauty product users; future studies can examine the current findings for other product categories. It is recommended for future research to consider cosmetic product categories (such as skincare and makeup products) to enhance the overall understanding of consumer behaviour in the beauty industry. As demand for organic and natural beauty products is rapidly increasing, it is recommended for future research to investigate consumer purchasing behaviours, particularly within the eco-friendly beauty product segment. The findings showed that influencer credibility declined after the pandemic, prompting further exploration into the underlying reasons for these shifts. Furthermore, it will be interesting to explore the potential impact of attitude towards ads on attitude towards influencer-endorsed products. In this study, we focused on the Portuguese respondents; future studies can examine and compare the findings of this study with samples from different countries.

One limitation of this study is that the data were collected exclusively from female participants, which may limit the generalizability of the results. Future research may incorporate a more diverse sample to enhance the generalizability of the findings. A comparison between female and male consumers can also provide relevant new insights. Future research could explore the impact of consumption involvement on consumer behaviour since it hasn't been extensively examined by researchers. Furthermore, future studies can consider complementary research methods, such as a causal study design, and study particular influencer types. This could provide additional insights on the topic and further assist business managers in the selection of digital influencers to represent their brands.