The influence of Electronic Word-of-Mouth (eWOM) on museum visit intention: an application of the IAM and IACM models

In today’s digital era, access to information has undergone an unprecedented transformation driven by rapid technological innovation. Consolidation of digital platforms have fundamentally reshaped how individuals create, access and share information (Lamberton and Stephen, 2016; Kannan, 2017; Jun et al., 2022; Dwivedi et al., 2023; Lee et al., 2026). In addition, these technological developments have profoundly reshaped consumer behavior. As a result, individuals increasingly rely on digital environments and social media platforms to seek information about products, services and experiences before making consumption decisions (Appel et al., 2020; Voorveld, 2019; Dwivedi et al., 2023; Cruz-Cárdenas et al., 2025). Within this context, reviews, comments and ratings shared by other users on online platforms and social networks play a decisive role in shaping perceptions and guiding decision-making processes. This phenomenon, known as Electronic Word-of-Mouth (eWOM), has been consistently identified as one of the most influential forms of communication in contemporary marketing (Kusawat and Teerakapibal, 2024), as online recommendations significantly affect users’ attitudes, intentions and behavioral responses toward products and services (Cheung and Thadani, 2012; Wu, 2017; Hong and Pittman, 2020; Akdim, 2021; Verma et al., 2023; Nguyen, 2025).

In service sectors characterized by a high degree of intangibility, such as tourism and leisure, eWOM has become a particularly powerful driver of consumer trust and engagement. Numerous studies have demonstrated that eWOM not only shapes consumer decision-making but also influences perceptions of quality, reputation and authenticity (Ayeh et al., 2013; Fang et al., 2016; Xu and Li, 2016; Esparza-Huamanchumo et al., 2024). However, despite extensive evidence in hospitality, travel and general tourism contexts, the museum sector remains underexplored from a quantitative and theoretical perspective (Hausmann, 2012). Museums represent a distinctive form of cultural consumption where experiences are symbolic, educational and emotionally charged rather than purely transactional. In this setting, digital communication, through social networks, user reviews and virtual communities, plays a key role in shaping visitors’ perceptions of value and authenticity (Orea-Giner and Vacas-Guerrero, 2020; Fernandez-Lores et al., 2022).

This study makes relevant theoretical and empirical contributions to the literature on Electronic Word-of-Mouth (eWOM) and information processing in experiential consumption contexts. First, it extends existing eWOM research by examining its effects beyond attitudinal outcomes, simultaneously analyzing information adoption, visit intention and actual visiting behavior within the museum context. Furthermore, the study provides novel empirical evidence from a noncommercial and culturally embedded setting (Orea-Giner and Vacas-Guerrero, 2020)., thereby challenging the predominance of utilitarian and market-oriented assumptions in traditional eWOM models.

Second, the research advances the application of the Information Adoption Model (IAM) and the Information Acceptance Model (IACM) by testing their explanatory power in an experiential and symbolic consumption environment. In doing so, it deepens theoretical understanding of how digital information is evaluated, accepted and translated into behavior, particularly when consumption decisions are also shaped by emotional and symbolic meanings.

Finally, the study identifies and empirically examines key dimensions of eWOM that influence the adoption of online information and subsequent museum visitation behaviors. This integrated approach contributes to marketing, leisure and tourism literature by offering a more comprehensive framework for understanding the role of eWOM in shaping real-world behavioral outcomes in cultural institutions.

The focus on Generation Z is particularly relevant given that this cohort represents the first generation of true digital natives, whose information search, evaluation and decision-making processes are deeply embedded in social media and user-generated content (UGC) (Perez-Aranda et al., 2024). Compared to previous generations, Generation Z shows higher reliance on online reviews, peer opinions and digital credibility cues when forming attitudes and behavioral intentions, especially in experiential consumption contexts (Gómez-Hurtado et al., 2025). Their strong preference for authentic, socially validated and visually mediated information makes them especially sensitive to eWOM dynamics. Examining the IAM and IACM frameworks within this generational context therefore provides a more accurate understanding of contemporary information acceptance mechanisms (Ngo et al., 2024b). Consequently, Generation Z offers an appropriate and theoretically meaningful context for testing how eWOM translates into visit intention and actual museum attendance.

The article begins with an introduction to Electronic Word-of-Mouth in Section 1, where the research questions and objectives are presented. Next, in Section 2, a literature review is conducted, presenting the various theoretical dimensions of the phenomenon in Section 3. Section 4 describes the methodology used, based on Structural Equation Modeling (SEM). Afterwards, in Section 5, the results derived from the analysis are presented. Finally, the research questions are answered in Section 6, and the initial objectives are discussed in Section 7.

Information acquires strategic value when it influences decision-making. In this sense, Electronic Word-of-Mouth (eWOM) has become one of the most powerful and pervasive mechanisms through which consumers exchange opinions and shape behavioral outcomes in digital environments (Akdim, 2021; Cheung and Thadani, 2012). eWOM is defined as the dissemination of positive, negative or neutral statements made by consumers about products, services or experiences through online media. This phenomenon has been widely studied across disciplines such as marketing, communication and tourism (Akbari et al., 2022; Verma and Yadav, 2020; Liu et al., 2024), with consistent evidence of its impact on trust, attitude formation and behavioral intention (Teixeira et al., 2018; García-De-Blanes-Sebastián et al., 2024).

Compared to traditional Word-of-Mouth, eWOM exhibits distinctive characteristics that enhance its influence. Its interactive and bidirectional nature enables continuous dialogue between users (Abuhjeeleh et al., 2023), while the absence of geographic and temporal limitations allows messages to spread rapidly across global audiences (Donthu et al., 2021). Furthermore, the asynchronous nature of digital communication, where content remains accessible indefinitely, enhances its persistence and long-term persuasive potential (Hennig-Thurau et al., 2004). The anonymity that characterizes many online platforms also encourages more candid and diverse opinions, generating a form of social proof that strongly shapes perception and collective judgment (AlRabiah et al., 2022; Bae, 2015; Kapoor and Gunta, 2016).eWOM occurs in a variety of digital spaces, social media platforms, blogs, forums, review websites and institutional web pages (Yan et al., 2017), and operates as a trust-based communication mechanism that bridges the gap between marketing messages and consumer experience. For marketers, understanding eWOM is essential not only as a communication tool but also as a process of information evaluation and adoption. The ability of eWOM to influence decisions depends largely on how users assess the credibility, quality and usefulness of the information they encounter, dimensions that connect eWOM with cognitive models of information processing and acceptance, such as IAM and IACM.

In cultural and experiential contexts, such as museum visits, eWOM assumes additional significance. Here, UGC does not simply convey functional attributes (e.g. opening hours or ticket prices) but expresses emotional, educational and symbolic values. These values relate to the concept of experiential consumption, where decisions are guided not only by rational assessment but also by the search for meaning, identity and authenticity (Holbrook, 1999; Schmitt, 1999). Hence, exploring eWOM in the museum sector offers a valuable opportunity to expand marketing theory by examining how informational, emotional and symbolic cues interact in noncommercial digital exchanges.

The IAM developed by Sussman and Siegal (2003) provides a theoretical lens to understand how individuals evaluate and integrate online information into their decision-making processes. Drawing from the Technology Acceptance Model (Davis, 1989) and the Elaboration Likelihood Model (Petty and Cacioppo, 1986), IAM combines technological and communicative perspectives, explaining how the characteristics of both the message (e.g. argument quality) and the source (e.g. credibility) determine information adoption. It has been widely applied in digital environments such as e-commerce and social media, where peer-generated information influences purchase and engagement behavior (Ngo et al., 2024a).

In parallel, the Information Acceptance Model (IACM) proposed by Erkan and Evans (2016) builds upon IAM and the Theory of Reasoned Action (Ajzen and Fishbein, 1980) to explain how users accept and act upon information obtained through eWOM. IACM extends the IAM by incorporating attitudinal and motivational factors such as information credibility, perceived usefulness, needs for information and attitude toward information, thus providing a more behaviorally oriented framework. Its application has proven especially relevant in contexts where UGC serves as a substitute for direct product experience, as in tourism and hospitality.

However, most studies applying IAM and IACM have been conducted in commercial settings, where decisions are typically transactional and utilitarian. The museum context, by contrast, presents a noncommercial, symbolic and experiential environment, where the value of information extends beyond utility to include cultural enrichment, emotional resonance and identity expression. This distinction provides an opportunity to expand these models theoretically, testing whether the mechanisms that explain information acceptance in consumption contexts also hold, or require adaptation, when the decision involves cultural or leisure experiences. By integrating IAM and IACM into the study of museum visits, this research aims to offer new theoretical insights into how eWOM operates in experiential consumption and contribute to the broader marketing literature on digital influence and cultural engagement.

In today’s digital environments, the exchange of opinions through digital environments has become a common practice, understanding the factors that determine the impact of eWOM becomes crucial for predicting consumer behavior (Chu and Kim, 2011; Dwivedi et al., 2020; Verhoef et al., 2019). Several variables influence users’ perceptions of the information they find online, influencing their behavioral decisions. In this context, eight key variables that shape the dynamics of eWOM are analyzed: source credibility, information quality, information credibility, needs for information, information usefulness, information acceptance, visit intention and visit. Each of these dimensions will be addressed below with support from current research and a proposal of hypotheses for future studies.

In the field of cultural and tourism promotion, source credibility (SC) in recommending or informing about an experience, such as visiting a museum, plays a crucial role in the perceived usefulness of the content. This credibility is typically based on two key elements: the degree of knowledge or expertise attributed to the sender and the perception of their honesty or integrity. In digital environments like travel blogs, social media or review platforms, these qualities are assessed through signals such as the author’s track record, the quality of their previous posts or other users’ ratings.

When a source is perceived as knowledgeable or authentic, its content not only gains credibility but also increases its perceived usefulness, facilitating decision-making, such as the choice to visit a specific museum. Studies like Filieri and McLeay (2014) highlight that online recommendations perceived as credible significantly influence information usefulness and, consequently, behavioral intentions. Likewise, Zeng and Gerritsen (2014) point out that in the context of cultural tourism, trust in the source enhances the effectiveness of promotional messages. More recent research by Putri and Tjokrosaputro (2024) indicates that SC, specifically the credibility of influencers, positively affects both the usefulness and acceptance of information:

In the context of museums and digital cultural promotion, Information Quality (IQ) refers to the intrinsic characteristics of the content, including clarity, organization, relevance, coherence and completeness. High-quality information facilitates understanding, reduces uncertainty and meets the informational expectations of potential visitors. IQ reflects the structural and technical aspects of the content rather than its perceived trustworthiness.

Prior research in digital tourism and cultural contexts has shown that well-structured, accurate and relevant information can enhance perceived usefulness and increase intention to visit (An et al., 2021; Wang and Yan, 2022). In museum settings, Garcia-Madariaga et al. (2019) demonstrate that clear storytelling, coherent presentation and organized content can improve user experience and perceived usefulness, ultimately foster information acceptance and visit intention:

Information Credibility (IC) refers to the perceived trustworthiness, reliability and accuracy of information as judged by the user. It is distinct from IQ in that it emphasizes subjective evaluation of truthfulness and consistency with prior knowledge or experience, rather than technical or structural quality. IC is particularly relevant in the early stages of decision-making, when users determine whether the content is trustworthy enough to guide their visit-related decisions.

Research indicates that perceived credibility enhances the perceived usefulness of information and fosters positive behavioral intentions. Erkan and Evans (2016) found that credible content on social media strongly increases consumer engagement and intention to act. In the museum context, García-Madariaga et al. (2017) and Blasco López et al. (2021) confirm that credible digital channels, including social media platforms and museum websites, improve user confidence, perceived usefulness and intention to visit:

The motivation to seek information is closely related to the consumer’s level of uncertainty or lack of knowledge. This need can be informational, emotional or comparative and translates into a greater willingness to resort to eWOM to reduce risk.

In the museum context, Booth (1998) identified that visitors have different needs based on their profile (residents, tourists, academics) and that different types and depths of information are required, from technical aspects to emotional narratives, to satisfy their motivations and ensure effective perceived usefulness. Riva and Agostino (2022), analyzing online reviews, found that specific museum attributes (e.g. amenities or ambiance) reflect visitor information needs and are closely tied to perceived information usefulness.

These studies show that the greater the users’ uncertainty or perceived informational gap, whether practical, emotional or experiential, the more likely they are to turn to UGC. This dynamic supports the hypothesis that:

The perceived usefulness of information directly influences its acceptance by the receiver, especially in digital contexts with high content volumes. Usefulness is defined as the extent to which a person believes that information helps them make a relevant decision (Davis, 1989). Various studies have shown that when individuals perceive information obtained through digital channels, such as eWOM, as useful, they are more likely to accept, trust and act on it.

For example, Sánchez-Torres et al. (2018) note that consumers show greater willingness to accept information they deem useful, leading to more informed purchasing decisions. Similarly, Leong et al. (2021) found that perceived content usefulness affects attitudes toward that content, which then mediates the relationship between eWOM exposure and purchase intention. Furthermore, Cheung and Thadani (2012) emphasize that perceived message usefulness is a key determinant of eWOM credibility, which, in turn, affects its acceptance.

It has also been found that information acceptance is closely related to factors such as channel trust, topic familiarity and content quality (Filieri, 2015). In other words, usefulness not only increases the likelihood of acceptance but also fosters a positive attitude toward the information, raising the chance of converting intention into action:

Visit intention refers to an individual’s conscious desire to go to a specific place, because of previously received information, especially through Electronic Word-of-Mouth (eWOM). This intention is considered a fundamental precursor to actual behavior and is influenced by multiple factors such as content quality, source credibility and information usefulness shared by other users (Ajzen, 1991; Jalilvand and Samiei, 2012). In tourism and hospitality contexts, this variable has gained importance as an indicator of digital strategy success, representing a favorable predisposition which, although not guaranteeing a real visit, reflects a high level of consumer consideration. Recent studies such as (Aboalganam et al., 2025) show that positive reviews and interaction with traveler-generated content directly influence the formation of this intention:

The variable Visit refers to the actual behavior of the user who, after being influenced by eWOM content, decides to physically attend or engage in a specific consumption experience at a destination, business or event. This variable represents the final conversion in the eWOM influence process and serves as empirical validation of the real impact of consumer-to-consumer communication in digital environments (Litvin et al., 2007; Sparks and Browning, 2011).

The decision to visit is strongly tied to prior variables such as intention, trust in sources and consistency between the information and user expectations. Recent studies such as (Wang et al., 2025) confirm that eWOM can predict actual consumer behavior, especially when the content aligns with consumer motivations and demonstrates a high degree of authenticity:

This paper first conducts a comprehensive literature review of the key concepts underlying this research: eWOM, Information Acceptance Models (IAM) (Sussman and Siegal, 2003) and IACM (Erkan and Evans, 2016), factors that influence visitor intention and visits to museums and cultural spaces, and the use of information about museums and cultural spaces on the Internet and in social media. Based on this literature review, a theoretical model is constructed, which is presented in the following Figure 1.

Likewise, to verify the empirical validity of the proposed theoretical model, a survey was developed, obtaining various analysis. First, descriptive analysis was conducted to diagnose the characteristics of the individuals surveyed. Subsequently, a quantitative analysis was performed using SEM of the sample obtained, allowing for the development of three types of research using this technique (Henseler, 2018, 2021; Henseler et al., 2018).

First, confirmatory research was conducted to detect how the constructs proposed in the theoretical model are related, thereby testing the hypotheses posed based on these relationships.

Second, SEM is used to demonstrate whether respondents intend to adopt information provided by eWOM about museums, and whether this ultimately influences visits to museum spaces (explanatory research).

Finally, SEM is used to determine the predictive capacity of the proposed theoretical model in other social, economic, geographical and/or temporal situations other than those covered in this research. This would allow for the replicability of the model presented here in the future (predictive research) (Sarstedt and Danks, 2021; Shmueli and Koppius, 2011).

As previously mentioned, the statistical analysis used in this study was conducted through SEM, a multivariate statistical technique that enables the analysis of complex relationships between observed (manifest) and unobserved (latent) variables. SEM makes it possible to test hypotheses about theoretical causal relationships by estimating a system of linear equations represented graphically in a structural model (Hair et al., 2022).

In this regard, the use of SEM is common in multivariate research fields such as marketing and digital social media communication (García-De-Blanes-Sebastián et al., 2024; Herrero et al., 2017; Sarstedt and Cheah, 2019; Sarstedt et al., 2021; Verma et al., 2023), as it allows for the simultaneous analysis of the relationships between unobserved variables and their manifest indicators, as well as the relationships between the latent variables themselves (Gefen et al., 2000; Hair et al., 2022; Williams et al., 2009).

Specifically, Smart-PLS 4 (Ringle et al., 2024) was chosen for this study due to its ease of use and interpretation of data, and because it enables the application of both widely known SEM techniques: covariance-based structural equation modeling (CB-SEM) and partial least squares structural equation modeling (PLS-SEM) (Barroso et al., 2010). In this case, PLS-SEM was selected, as the theoretical model presented is highly complex, involving numerous items and latent variables (Chin, 1998; Hair et al., 2022; Hair et al., 2017; Sarstedt et al., 2021). In addition, PLS-SEM is recommended when working with small samples drawn from relatively small populations (Richter et al., 2016).

Based on the proposed theoretical model, a questionnaire was designed based on the proposed theoretical model to collect necessary data. It included sociodemographic questions and information on Internet and social media use related to museums. Items were developed around the model’s eight constructs, adapted to the context of eWOM about museums, using the IACM framework as a reference (Erkan and Evans, 2016). Responses were measured on a five-point Likert scale (strongly disagree–strongly agree), a format commonly used in SEM studies (Hair et al., 2022).

Table 1 presents the constructs, items and references used in the questionnaire design.

Once the measurement instrument was designed, data were collected using a self-administered online survey via Google Forms, valued for its ease of design, low cost, response organization and respondent anonymity (Dillman et al., 2014). Collection occurred face-to-face through quick response (QR) code access between December 2024 and February 2025, targeting Generation Z in Spain, yielding 240 responses. A homogeneous convenience sampling method was applied to control the specific context, enhance accessibility and reduce time and costs (Jager et al., 2017).

To evaluate the sociodemographic data of the collected sample, a frequency analysis was conducted, as shown in Table 2. In this regard, it is evident that there is a higher number of female respondents, which is linked to the traditionally greater number of women pursuing secondary and higher education in Spain. As most of the respondents reported an educational level above high school (97.5%), this may explain the lower number of male participants in the survey.

Regarding the use of social media by the sample, most participants reported spending between 4 and 5 hours on these platforms, with TikTok and Instagram being the most used. As for the use of podcasts, which are commonly used to share museum-related content, only about 15% reported using them. In addition, around 18% stated they had posted some form of comment on digital media about museums.

Finally, over 80% of the sample indicated they had visited at least one museum in the past year. The most frequent response was visiting two to three museums per year, with over 60% of participants stating they booked their tickets online.

To determine the necessary sample size for analysis using SEM, the rules established by Chin and Newsted (1999) were followed. Specifically, assuming a medium effect size, a statistical power of 0.80, a significance level of 0.05 and five independent variables in the model, the minimum required sample size is 91 cases (Cohen, 1988; Green, 1991). Likewise, following the recommendations of Hair et al. (2022) for sample size in PLS-SEM, with a power of 0.80, aiming for a minimum R² of 0.50, at a significance level of 0.05 and considering the five independent variables in the theoretical model, the minimum required sample would be just 20 cases.

Therefore, with 240 valid responses, the obtained sample is more than sufficient to conduct the analysis.

Once the descriptive data of the sample were reviewed, a series of preliminary steps were taken before initiating the PLS-SEM analysis. First, the data set was checked for any missing or outlier values, none of which were found. In addition, a review of skewness and kurtosis values of the analyzed items was conducted. Although PLS-SEM is a nonparametric method, some authors argue that it is important to assess whether the data approximate a normal distribution, as extreme values could potentially distort the subsequent analysis (Hair et al., 2022). After analyzing the sample data, it was confirmed that all items had skewness values below 3 and kurtosis values below 10 (Matas-Terron, 2023), indicating that the data approximately follows a normal distribution.

The PLS-SEM analysis begins by assessing the relationships between the items and the latent variables, to determine whether the measurement model demonstrates adequate consistency and reliability (Hair et al., 2017).

The first step is to analyze the outer loadings of each item with its corresponding construct. A reference value of ≥ 0.707 is used (Carmines and Zeller, 1979) to determine whether the proposed indicators are reliable. Some items with loadings below this threshold were identified. Consequently, those indicators were removed from the model (Hair et al., 2011), resulting in a better model fit and, ultimately, individual reliability for the remaining items above the reference value (Table 3).

Regarding the reliability of each latent variable, the results of Cronbach’s alpha, Composite Reliability (ρc) and Dijkstra–Henseler’s (ρA) coefficients were analyzed, using 0.7 as the reference value indicating adequate reliability (Nunnally and Bernstein, 1994). Table 3 shows that all these coefficients for every construct exceed this reference value, meaning that the indicators for each construct are adequately measuring their respective latent variables.

Removal of indicators in reflective models does not compromise content validity if the retained items continue to represent the conceptual domain and facet coverage is adequately documented (Haynes et al., 1995; Sireci, 1998; Hair et al., 2022). For example, in Information Acceptance, the core facet was fully covered (AI3–AI4), while the cognitive consequences facet (AI1–AI2) showed lower psychometric performance in this sample. Convergent evidence supports the conclusion that the retained measure continues to capture the construct of interest, although it is suggested that future work reexamine artificial intelligence cognitive indicators with alternative samples and wordings.

On the other hand, Hair et al. (2019) recommend using 95% confidence intervals via bootstrapping to verify whether the internal consistency coefficients are statistically significant. The results in Table 4 indicate that all coefficients are above the lower limit (0.7) and below the upper limit (0.95).

This section analyzes the extent to which each construct shows agreement in explaining the variance of its items (Hair et al., 2019). The measure used for this is the average variance extracted (AVE), with a value of 0.50 considered acceptable to establish convergent validity. That means each construct should explain at least 50% of the variance of its indicators (Fornell and Larcker, 1981; Hair et al., 2019). As shown in Table 2, all AVE values exceed 0.50, indicating that the model demonstrates convergent validity for all constructs.

Regarding discriminant validity, that is, whether the constructs in the developed model are empirically distinct from each other, the results from the Heterotrait–Monotrait (HTMT) matrix ratio developed by Henseler et al. (2015) are analyzed. Reference values of 0.85 (Kline, 2011) and 0.90 (Gold et al., 2001) are used, and as shown in Table 5, all values fall below 0.90 and 0.85, indicating that, preliminarily, discriminant validity exists in the model.

In addition, it is possible to use 95% confidence intervals via bootstrapping to verify whether all values in the HTMT matrix are statistically valid. Table 6 shows that all values fall within the proposed confidence intervals, thereby confirming the existence of discriminant validity.

Once the revision of the measurement model has been completed, the next step is to analyze the structural model and verify how the constructs relate to each other.

First, it is necessary to verify possible multicollinearity problems between the antecedent variables of each of the endogenous constructs. To do this it is necessary to consider the Variance Inflation Factor (VIF) statistic and using a threshold of 3 for detecting possible collinearity issues (Hair et al., 2019). Table 7 shows that all VIF values are below 3, indicating no evidence of multicollinearity in the structural model.

Finally, the results in Table 8 are analyzed, showing the path coefficients obtained between the different constructs, as well as the statistical significance of each. For the latter, one-tailed bootstrapping technique is used based on the original samples (Hair et al., 2011), using 10,000 bootstrap samples (Streukens and Leroi-Werelds, 2016), from which the standard errors, t-statistics and 90% confidence intervals are obtained.

With this information, it is possible to verify whether the hypotheses proposed in the research are accepted or rejected. In Table 7, the path coefficients fall within the proposed confidence intervals. However, analyzing their statistical significance reveals that the supported hypotheses are H1, H5, H7, H8, H9, H10 and H11, while the unsupported hypotheses are H2, H3, H4 and H6.

Once the supported and unsupported hypotheses within the structural model have been determined, Figure 2 presents the results of the obtained PLS-SEM model, showing the values of the outer loadings for each item and the path coefficients for each proposed hypothesis. In addition, the R² coefficients are included, which indicate the explanatory power of the model.

In this regard, for the dependent variable Visit, an R² of 0.722 and an adjusted R² of 0.721 were obtained, indicating that this variable is explained 72% by the model’s independent variables. Likewise, it is observed that the independent variables Information Acceptance and Visit Intention also have high R² values, both exceeding 0.67, the minimum threshold established by Chin (1998).

Next, Table 9 shows the effects produced by the exogenous constructs on the endogenous constructs of the model. It can be observed that the largest effects occur between the constructs Visit Intention and Visit (f² = 2.601), Information Acceptance and Visit Intention (f² = 2.457), Needs of Information and Information Usefulness (f² = 0.937) and between Needs of Information and Information Acceptance (f² = 0.455). These effects exceed the threshold to be considered large (f² = 0.35) (Cohen, 1988).

The fit of both the measurement (saturated) and structural (estimated) models was evaluated using goodness-of-fit indicators, as recommended by some authors (Henseler, 2017; Henseler, Hubona and Ray, 2016), although others question this approach (Hair et al., 2019). The approximate fit was analyzed with the SRMR (Standardized Root Mean Square Residual) coefficient, and Table 10 shows all SRMR values below 0.08, meeting the recommended criterion (Hu and Bentler, 1998).

Bootstrap tests of exact fit, SRMR (Standardized Root Mean Square Residual), d_ULS (Unweighted Least Squares Discrepancy) and d_G (Geodesic Discrepancy) were conducted for 95% and 99% confidence intervals (Henseler, 2017; Henseler et al., 2016). For the saturated model, all tests were below the confidence limits. For the estimated model, some values exceeded the 95th percentile but remained below the 99th percentile, indicating that the model is acceptable at a 99% confidence level. Adjustments by removing three low-weight structural relationships worsened overall fit.

Consequently, the original model was retained due to its theoretically justified structure, high explanatory power (R² between 0.65 and 0.72) and statistically significant key relationships. Following Hair et al. (2022), priority was given to explanatory capacity and structural validity over exact model fit.

Finally, the model is assessed to determine whether it has sufficient predictive capability and whether it can be applied to future samples different from the one used in this study. To do so, the PLSpredict algorithm is used to verify the predictive power of the model (Shmueli et al., 2019).

Table 11 presents the Q² results for each item of the endogenous constructs, all of which are positive, indicating that the designed PLS model has better predictive ability than the predictions generated by the algorithm itself (Shmueli et al., 2019).

In addition, the prediction errors of the PLS model and those of the Linear model (LM) generated by the algorithm are analyzed (Shmueli et al., 2019). As shown in Table 11, all Root Mean Squared Errors of the PLS model are smaller than those of the LM, except for three indicators of the Visit construct. This demonstrates that the constructs Information Acceptance, Information Usefulness and Visit Intention have strong predictive ability for new observations.

In addition, the predictive power of the model is further assessed using the Cross-validated Predictive Ability Test. It is essential that the model demonstrates a lower average loss difference than both the average values of the indicators average (IA) (Sharma et al., 2022) and the Linear Model (LM) prediction benchmark (Shmueli et al., 2019).

Table 12 shows that the average loss of the model is indeed significantly lower than the average values of the indicators, which supports the predictive validity of the constructs and of the model. However, the model’s average losses compared to the Linear Model (LM) prediction benchmark are only significantly lower for the constructs Information Acceptance and Visit Intention, indicating that these two constructs are the ones for which a high predictive power can be confirmed.

Results provides empirical evidence on the role of electronic Word-of-Mouth (eWOM) in information adoption, visiting intention and actual visits to museums by Generation Z in Spain. Using a theoretical model based on IAM and IACM, eleven hypotheses were tested.

Results show that Source Credibility (H1) and Information Credibility (H5) significantly influence Information Usefulness, confirming previous findings (Blasco López et al., 2021; Erkan and Evans, 2016; Filieri and McLeay, 2014; Zeng and Gerritsen, 2014). This highlights the importance of selecting trustworthy endorsers and influencers capable of sharing authentic, first-hand experiences to enhance perceptions of usefulness. In contrast, Information Quality (H3) did not show a significant effect on usefulness, representing a novel contribution to literature. This suggests that young audiences prioritize authenticity and trust in the source over message quality. For example, viral content shared by public figures, such as Dua Lipa’s visit to the Prado Museum (Museo del Prado, 2025), may generate sufficient perceived usefulness to influence adoption, intention and visits. All of this reflects the characteristics of digital native audiences who value experiential and peer-generated information (Erkan and Evans, 2016; Virto et al., 2024). This unexpected finding calls for further research in other contexts.

Needs of Information (H7) emerged as the strongest predictor of Information Usefulness (β = 0.601). Users with greater informational motivation are more likely to positively evaluate UGC (Booth, 1998; Riva and Agostino, 2022). Moreover, Needs of Information directly influenced Information Acceptance (H8) (β = 0.470), showing its central role in the adoption process. Users may validate eWOM content primarily to satisfy an informational need, which subsequently drives their visit intention and actual visitation behavior. These relationships also exhibited large effect sizes (f²), strengthening the robustness of the results.

Information Usefulness strongly determined Information Acceptance (H9), in line with the assumptions of IACM and prior studies (Cheung and Thadani, 2012; Chu and Kim, 2011; Erkan and Evans, 2016; Leong et al., 2021). Once museum-related eWOM is perceived as useful, it is adopted by users. This adoption significantly influenced Visit Intention (H10), which, in turn, predicted actual visits (H11). The large effect sizes found for these relationships further support the proposed causal chain.

Conversely, no significant relationships were found between Source Credibility, Information Quality and Information Credibility with Information Acceptance (H2, H4, H6). This suggests that, although these factors can affect perceived usefulness (except for Information Quality), they do not directly lead to information acceptance. As Chu and Kim (2011) and Erkan and Evans (2016) argue, the abundance of online content forces users to filter and prioritize information perceived as useful for their specific goals – in this case, deciding whether to visit a museum, an inherently intangible experience (Li and Lv, 2024). In other words: a credible source, or information, can increase perceived usefulness, but it does not automatically result in behavioral adoption. This aspect highlights the mediating role of perceived usefulness in the eWOM adoption process. Practically, young visitors seem to accept museum-related eWOM only when it proves instrumentally useful for an imminent visit (e.g. planning, expectations), even if they recognize credible sources and high-quality messages.

From a methodological perspective, the model demonstrated high explanatory power, with R² values exceeding recommended thresholds (Chin, 1998), particularly for Visit Intention (R² = 0.711) and Visit (R² = 0.722). This suggests that the constructs included in the model account for a substantial proportion of the variance in behavioral outcomes, thereby strengthening the validity of the IACM framework. Predictive validity analyses with PLS-Predict further confirmed the robustness of the model and its potential applicability to other cultural or geographic contexts. However, adaptations may be necessary, as some items in the Visit construct showed limited predictive capacity (Table 11). Moreover, while overall predictive validity was confirmed (Table 12), strong predictive power was found mainly for Information Acceptance and Visit Intention, suggesting that refinement of other constructs is needed in future research.

This study has provided valuable insights into the role played by Electronic Word-of-Mouth (eWOM) communication in information adoption, visit intention and actual visits to museums by Generation Z in Spain. By applying the IAM and IACM models, it has been confirmed that these theoretical frameworks are suitable for analyzing information acceptance in the cultural context, as evidenced by the high R² values obtained in the variables Visit Intention and Visit.

In relation to the theoretical contributions, the study has provided an updated definition of eWOM, emphasizing its distinctive characteristics in digital environments, such as its global reach, permanence over time and capacity for virality. Furthermore, the key dimensions shaping this phenomenon, have been identified and analyzed in terms of their influence on information adoption and museum visit behavior.

Among the most significant findings, the variable Needs of Information emerged as the strongest predictor of Information Usefulness, reinforcing the idea that users with greater informational motivation are more likely to positively evaluate UGC. Likewise, both Source Credibility and Information Credibility were found to significantly impact Information Usefulness, confirming the importance of these factors in fostering favorable attitudes toward information. By contrast, an unexpected result was observed: Information Quality did not have a significant relationship with Information Usefulness or with Information Acceptance, posing a challenge to previous literature and opening new research horizons.

Interestingly, the nonsignificant effect of Information Quality on perceived usefulness and acceptance highlights a distinctive aspect of Generation Z’s engagement with museum eWOM. Rather than valuing formal attributes such as completeness or structural clarity, this generation appears to prioritize authenticity, emotional resonance and the perceived sincerity of the content. This aligns with theories on digital authenticity and influencer culture, suggesting that trust signals and heuristic cues may guide their information processing more than technical quality. Consequently, this finding offers a relevant theoretical insight into how experiential and symbolic factors shape digital cultural consumption, underscoring the importance of credibility and authenticity in influencing visit intention and audience behavior.

This study clarifies that Information Quality (clarity, organization and structural relevance) and Information Credibility (perceived reliability and truthfulness) are conceptually distinct. Although both influence information processing, our findings highlight that, in the context of museum eWOM, credibility plays a more decisive role than technical quality, reflecting the symbolic and experiential nature of cultural visits. This distinction reinforces the validity of the constructs and helps explain the nonsignificant effect of Information Quality.

From a theoretical perspective, this study extends the IAM and IACM models by highlighting how the experiential, symbolic and noncommercial nature of museums nuances traditional assumptions about information adoption. Unlike typical tourism or hospitality contexts, cultural consumption involves motivations centered on learning, personal enrichment and meaning making, which shape how eWOM is perceived and acted upon. The findings demonstrate that factors such as Needs of Information and Credibility take on heightened importance in this setting, while Information Quality may play a less central role. This suggests that theoretical models of information adoption must account for the unique characteristics of cultural experiences, including intrinsic motivations and symbolic value. Consequently, the study offers a refined understanding of eWOM dynamics in noncommercial, culturally rich contexts, contributing to conceptual advancement beyond mere contextual application.

Overall, the study confirms the causal chain proposed in the theoretical model: information perceived as useful is adopted by users, which leads to visit intention and ultimately to the actual visit to the museum. This process highlights the essential role of usefulness and information acceptance as critical links between eWOM and visitor behavior.

Based on the results obtained, several managerial contributions can be made to museum and cultural spaces managers.

First, it is essential that museums carefully select their endorsers, sponsors or influencers, prioritizing those who convey authenticity and credibility, as these attributes are key to increasing the perceived usefulness of eWOM. Moreover, it is advisable to encourage the creation and dissemination of authentic and personal content on social media, as this appears to have a greater impact on young audiences than the objective quality of the message itself. Real experiences and first-person narratives from visitors or relevant figures can serve as more effective communication tools than formal descriptions.

Second, museums should actively engage with UGC by encouraging reviews, testimonials and posts that reflect personal experiences. Interactive campaigns, such as contests, hashtags or “visitor of the month” features, can increase participation and the visibility of authentic content. Platforms such as Instagram, TikTok and YouTube should be leveraged strategically, considering the content format most appealing to Generation Z (short videos, reels, stories).

Third, given that informational needs strongly predict information usefulness and acceptance, museums should provide targeted and accessible content that meets these needs. This could include personalized digital guides, frequent asked questions (FAQ)s, behind-the-scenes insights and curated thematic content that helps potential visitors plan and enrich their experience. Tailoring content to specific visitor interests can enhance perceived usefulness and motivate actual visits.

Fourth, although Information Quality did not prove significant in this study, museums should maintain clarity, accessibility and user-friendliness on their digital platforms. This includes mobile optimization, easy navigation and visually appealing content, which contribute to user satisfaction and reinforce institutional credibility.

Finally, museums should monitor the effectiveness of their eWOM campaigns through analytics and feedback mechanisms. Tracking engagement metrics, user comments and shares can help managers identify which types of content are most effective in driving perceived usefulness, information acceptance and visit intention. Periodic evaluation and adaptation of content strategies will ensure that digital communications remain relevant, engaging and aligned with the evolving preferences of Generation Z audiences.

This study opens several opportunities for future research. It would be particularly interesting to replicate the study in other geographic and socioeconomic contexts to determine whether the findings of this research, especially regarding the lack of influence of Information Quality on Information Usefulness and Acceptance, are consistent in different settings. In addition, future studies should explore the use of mediating variables in the relationship between Information Usefulness and Information Acceptance, such as attitude toward the information or the user’s previous experience. There is also potential to extend the sample to other population segments, such as other generations or profiles with different cultural motivations, to assess whether the proposed model remains valid. Finally, longitudinal studies could provide a more dynamic view of how Visit Intention and actual Visit evolve over time.

Several limitations must be acknowledged. The study focused exclusively on Generation Z in Spain, which may limit the generalizability of findings to other populations or cultural contexts. The cross-sectional design prevents establishing causality with certainty and does not allow for the analysis of temporal changes.

It should also be noted that, during the analysis process, some items were removed from the original theoretical model due to their low outer loadings. After examining the reasons behind the issues associated with these items, we concluded that they are related to their adaptation to the sample used (Gen Z), the context in which the model was applied (museums) and to potential ambiguity in the wording of the items. For future applications of this model in other contexts and samples, the selected items should be carefully examined to support the study’s replicability.

Although the model showed strong explanatory power, predictive capacity was more robust for certain constructs (Information Acceptance and Visit Intention) than for others, suggesting the need for refinement before applying the model in different contexts. Finally, the questionnaire could be improved, particularly items measuring actual visits, which exhibited weaker predictive validity.

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.