Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2, 1228-1244 2025 Publisher: Learning Gate DOI: 10.55214/25768484.v9i2.4745 © 2024 by the authors; licensee Learning Gate © 2025 by the authors; licensee Learning Gate History: Received: 11 December 2024; Revised: 16 January 2025; Accepted: 21 January 2025; Published: 12 February 2025 * Correspondence: dangph@hub.edu.vn Factors influencing purchase decisions on social media platforms: The role of explainable artificial intelligence Nguyen Van Dat1, Pho Hai Dang2*, Nguyen Van Thich3 1,3Ho Chi Minh University of Banking, Faculty of Business Administration, Vietnam; datnv@hub.edu.vn (N.V.D.). thichnv@hub.edu.vn (N.V.T.). 2Ho Chi Minh University of Banking, Faculty of Management Information Systems, Vietnam; dangph@hub.edu.vn (P.H.D.). Abstract: This study aims to measure the impact of privacy concerns and perceptions of personalization on purchase decisions on social media platforms. It focuses on the mediating role of attitudes toward advertising and the moderating role of Explainable Artificial Intelligence (XAI). The study investigates consumers who frequently shop on social media. The research model was implemented using an online questionnaire and direct interviews, yielding 515 valid responses. To assess the reliability of the measurement scales, SPSS 26 software was employed. The research hypotheses were tested, and the measurement and structural models were evaluated using AMOS 28. The proposed model is grounded in the Elaboration Likelihood Model (ELM), causal models, interpretability in human-AI interaction, computational privacy theory, and the Theory of Reasoned Action (TRA). The findings indicate that consumers' perceptions of personalized advertising content positively influence their attitudes toward advertisements. Privacy concerns negatively affect users' attitudes toward advertisements. Positive attitudes toward advertising, in turn, influence purchase decisions on social media. This study enriches the theoretical understanding of consumer behavior toward AI-enabled technological products and offers managerial implications for producers to enhance advertising quality and meet consumer demands in the context of social media shopping. Keywords: Explainable Artificial Intelligence, Perceived Personalization, Social Media Purchasing. 1. Introduction The rapid growth of social media platforms has significantly transformed how businesses approach advertising. As users are increasingly bombarded with content, advertisers face the challenge of cutting through the noise to effectively engage and influence potential customers. Personalization has emerged as a key strategy in digital marketing, allowing brands to tailor content to individual preferences and behaviors, thereby improving engagement and conversion rates. In the current digital era, advertising on social media platforms has become a vital component of marketing strategies for businesses in the e- commerce sectors. With the rapid advancement of artificial intelligence (AI), the personalization of advertising content has achieved unprecedented effectiveness. However, despite these clear benefits, the lack of transparency in AI operations has raised significant concerns regarding fairness and ethics in advertising. AI can generate highly targeted advertising that resonate with specific audiences by analyzing vast amounts of data from users' social media activities. This personalization is achieved through sophisticated algorithms that assess user behavior, preferences, and interactions on social media platforms. For instance, AI can optimize ad placements by predicting which content will elicit positive emotional responses from users, thus driving engagement and conversion rates [1, 2]. Moreover, the use of generative AI in conversational marketing enables brands to engage in real-time, personalized https://orcid.org/0009-0007-5012-289X 1229 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate dialogues with consumers, further enhancing the relevance of advertisements [2, 3]. This dynamic interaction not only improves customer satisfaction but also increases the likelihood of purchase decisions being influenced by the personalized content presented to them. However, traditional AI algorithms often operate as "black boxes," providing limited transparency regarding how decisions are made. This lack of explainability can lead to consumer trust issues, potentially affecting their response to personalized advertising. The ethical implications of AI in advertising cannot be overlooked. As organizations deploy AI to automate and optimize marketing strategies, they must also consider the ethical dimensions of these technologies. The use of AI in advertising raises questions about consumer privacy, data security, and the potential for bias in algorithmic decision-making [4]. XAI addresses these concerns by providing insights into how AI models operate, which can help mitigate data misuse fears and enhance consumer trust [5]. By making AI's decision pathways more transparent, advertisers can align their strategies with ethical standards, ultimately leading to more responsible marketing practices. Furthermore, the effectiveness of personalized advertising is significantly enhanced when consumers understand the rationale behind the recommendations they receive. Research indicates that when consumers perceive AI-driven advertisements as relevant and trustworthy, they are more likely to engage with the content and make purchasing decisions [6, 7]. XAI not only improves the interpretability of AI systems but also empowers consumers by allowing them to comprehend the factors influencing their ad experiences. This understanding can lead to a more positive perception of the brand and its offerings, thereby influencing purchasing behavior favorably. The existing literature on XAI in marketing is limited and lacks comprehensiveness, with the aforementioned gaps either insufficiently addressed or still under debate. To bridge this research gap, this paper aims to develop a model to investigate users' attitudes toward social media advertising through their perceptions of XAI quality. The model includes perceptions of personalization, privacy concerns, attitudes toward advertising, and purchasing decisions. By exploring the relationship between XAI, advertising personalization, and purchase decisions, this study aims to provide valuable insights for marketers, enabling them to optimize their advertising strategies while fostering consumer trust and loyalty in the commerce sector as well as in the logistics industry. The following sections of the study will present the foundational theories related to the research topic, hypotheses, proposed research model, research methodology, results, and discussion. The final section will offer conclusions and managerial implications. 2. Literature Review and Research Model 2.1. The Elaboration Likelihood Model (ELM) The Elaboration Likelihood Model is a critical framework for understanding how advertising content influences consumer purchasing decisions on social media platforms. ELM posits that individuals process persuasive information through two distinct routes: the central route and the peripheral route. The central route involves careful and thoughtful consideration of the information presented, typically employed in high-involvement decisions, while the peripheral route relies on superficial cues, such as visual elements or the attractiveness of endorsers, in low-involvement situations [8, 9]. This dual processing framework is particularly relevant in the context of social media, where users are often bombarded with a plethora of advertisements, necessitating a nuanced understanding of how different types of content can effectively engage consumers. Research indicates that the effectiveness of advertising on social media can vary significantly based on the level of consumer involvement with the product being advertised. For high-involvement products, where consumers are likely to engage in extensive information processing, advertisements that provide detailed, relevant content are more effective in persuading consumers [10, 11]. Conversely, for low-involvement products, peripheral cues such as celebrity endorsements or visually appealing graphics can significantly enhance the persuasiveness of the advertisement [12]. This distinction underscores the importance of tailoring advertising strategies to align with the target audience's level of involvement, thereby 1230 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate optimizing the likelihood of positive purchasing decisions. Moreover, the role of social media influencers and their perceived attractiveness can significantly impact consumer attitudes and behaviors. Studies have shown that when influencers are congruent with the product being advertised, they can serve as effective peripheral cues that enhance the persuasive power of advertisements, even in low-involvement contexts [10, 12]. This phenomenon is particularly relevant on platforms like Instagram and TikTok, where influencer marketing has become a dominant strategy for brands seeking to engage younger audiences. The ELM framework helps elucidate why consumers may respond favorably to advertisements featuring influencers, as these endorsements can evoke emotional responses that facilitate purchasing decisions [10, 12]. In addition to the content and presentation of advertisements, the context in which they are encountered also plays a crucial role in shaping consumer responses. For instance, the popularity of a post can serve as a peripheral cue that influences consumer perceptions and intentions. Research has demonstrated that advertisements with higher engagement metrics (likes, shares, comments) are perceived as more credible and appealing, thereby enhancing their effectiveness in driving purchasing decisions [13, 14]. This interplay between content, context, and consumer involvement highlights the complexity of consumer behavior on social media and the necessity for marketers to adopt a multifaceted approach when designing advertising campaigns. Figure 1. The elaboration likelihood model. Source: Cacioppo, et al. [15] 2.2. The Causal Models and Explainability in Human-AI (CMEH-AI) Causal models provide a framework for identifying and analyzing the relationships between various consumer decision factors. For instance, AI systems can analyze consumer data to predict purchasing behavior, but without a clear understanding of the causal relationships at play, marketers may struggle to optimize their strategies effectively. By employing causal inference techniques, marketers can discern which aspects of their advertising campaigns are genuinely influencing consumer behavior, allowing for more targeted and effective marketing strategies. This is particularly relevant in social media contexts, where the rapid dissemination of information can lead to complex interactions between consumer perceptions and advertising content. Explainability in AI is crucial for fostering trust and enhancing user experience in human-AI interactions. As consumers become more aware of AI's role in shaping their purchasing decisions, the demand for transparency in how these systems operate increases. Explainable AI (XAI) initiatives aim to demystify AI decision-making processes, allowing consumers to understand the rationale behind personalized advertisements. This understanding can mitigate skepticism and enhance the perceived credibility of AI-driven recommendations, ultimately influencing purchasing decisions positively. For example, when consumers are informed about how their data is 1231 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate used to tailor advertisements, they are more likely to engage with the content and make informed purchasing choices. The findings of Shin [16] highlight the dual role of causality and explainability in establishing trust, which ultimately influences user behavior. The study demonstrates that providing clear explanations of personalized suggestions or recommendations enhances user trust. According to Shin [16] an explainable system must ensure three key factors: fairness, clarity, transparency, and accountability. These factors are also discussed in studies by Meske, et al. [17]; Gerlings, et al. [18] and Shin and Park [19]. The three elements of fairness, transparency, and accountability in XAI foster corresponding user perceptions: perceived fairness, transparency, and accountability. These perceptions, in turn, influence users' trust in artificial intelligence [16]. Moreover, the integration of explainability into AI systems can help address ethical concerns related to data privacy and algorithmic bias. As AI technologies become more pervasive in marketing, ensuring that these systems operate transparently is essential for maintaining consumer trust. Ethical AI practices involve not only providing clear explanations of how AI systems function but also ensuring that these systems are designed to avoid reinforcing existing biases in advertising. This ethical consideration is particularly important in social media environments, where diverse consumer demographics interact with targeted advertising. Figure 2. The causal models and explainability in human-AI. Source: Shin [16] 2.3. The Privacy Calculus Theory Privacy Calculus theory plays a significant role in shaping purchasing decisions on social media platforms, particularly as consumers navigate the complexities of personalized advertising and data privacy. The intersection of privacy concerns and consumer behavior is critical, as individuals increasingly weigh the benefits of personalized recommendations against their apprehensions about data misuse and privacy violations. One of the foundational concepts in computational privacy theory is the "privacy paradox," which describes the discrepancy between consumers' stated privacy concerns and their actual online behaviors. Research indicates that while consumers express significant privacy concerns, they often fail to take protective actions when engaging in online transactions [20]. This paradox is particularly relevant in the context of social media, where personalized advertising relies heavily on user data. For instance, studies have shown that privacy concerns can inhibit consumers' willingness to engage with personalized recommendations, thereby negatively impacting their purchase intentions [21, 22]. This suggests that marketers must navigate a delicate balance between leveraging consumer data for personalization and addressing privacy concerns to foster trust and encourage purchasing behavior. Computational models can provide valuable insights into user behavior regarding privacy and purchasing decisions. By employing formalized computational modelling, researchers can derive precise predictions about how privacy concerns influence consumer behavior in online contexts [23]. Such 1232 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate models can help identify the mechanisms through which privacy concerns affect purchasing decisions, allowing marketers to tailor their strategies accordingly. For example, understanding the factors that contribute to the privacy calculus where consumers weigh the perceived benefits of personalized advertising against the risks of privacy loss can inform the design of more effective advertising campaigns [24]. Moreover, the role of privacy signals in online advertising is crucial for shaping consumer perceptions and behaviors. Transparency regarding data usage and privacy policies can significantly influence purchasing intentions. Research has demonstrated that when consumers are presented with clear and accessible privacy information, they are more likely to incorporate privacy considerations into their purchasing decisions [25]. This highlights the importance of providing consumers with control over their data and clear communication about how their information will be used, which can mitigate privacy concerns and enhance trust in the advertising process [26, 27]. Figure 3. The privacy calculus theory model. Source: Culnan and Armstrong [28]. 2.4. The Relationship between Perceived Personalization and Attitudes toward Personalized Advertising Based on the scrutiny model theory [29], two information processing routes include the central route, which involves careful and thorough consideration of advertising content, and the external route, based on superficial signals at the surface level. The topic considers the ELM model as a tool to explain how consumers process personalized advertising on social networks influenced by AI. In personalized advertising, perceived personalization can lead to customers processing the central route [30]. According to Cacioppo, et al. [15], scrutiny is defined as the degree to which an individual thinks carefully about an argument. The depth and relevance of personalized content can drive more positive and lasting attitudes toward personalized advertising, increasing customer purchase intent. Therefore, the following hypothesis is proposed: H1: Perceived personalization positively impacts attitudes toward personalized advertising. 2.5. The Relationship between the Privacy Concerns and Attitudes Toward Personalized Advertising Privacy concerns may trigger the peripheral processing route because it focuses on the credibility of the information source rather than the content. This aspect is consistent with the ELM view that peripheral cues (such as trust and credibility) can influence attitudes. The greater the privacy concern, the more negatively customers tend to react to personalized advertising [31, 32]. Therefore, the following hypothesis is proposed: H2: Privacy concerns negatively impact attitudes toward personalized advertising. 1233 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate 2.6. The Relationship between the Attitude Advertising and Purchase Decision The relationship between attitudes toward advertising on social media platforms and purchasing decisions is a critical area of research, particularly as digital marketing continues to evolve. Numerous studies have explored how consumer attitudes toward social media advertising influence their purchase intentions, revealing various factors that mediate this relationship. One significant study by Genç and Turna highlights the mediating effect of attitudes toward online advertising in the context of social media addiction and online purchase intention. Their findings suggest that positive attitudes toward online advertisements can enhance consumers' purchase intentions, particularly among those with higher levels of social media addiction [33]. This underscores the importance of crafting advertisements that resonate positively with users to drive purchasing behavior. Celebrity endorsements also play a crucial role in shaping consumer attitudes. Melati et al. found that the congruence between a celebrity endorser and the product being advertised significantly influences consumer attitudes and, consequently, their purchase intentions. The familiarity and credibility of the celebrity can enhance the effectiveness of the advertisement, leading to more favorable consumer responses [34]. This aligns with the notion that consumers are more likely to engage with advertisements that feature relatable and trustworthy figures. Furthermore, Yang et al. explored consumer attitudes toward online video advertisements on platforms like YouTube, demonstrating that factors such as creativity and emotional appeal significantly impact shopping intentions and purchase behavior. Their research indicates that engaging video content can lead to more favorable attitudes toward advertisements, ultimately influencing purchasing decisions [35]. This highlights the necessity for marketers to focus on the quality and creativity of their advertising content to capture consumer interest effectively. Therefore, the following hypothesis is proposed: H3: Attitude advertising positively impact Purchase Decision. 2.7. The Moderating Role of XAI Quality Perception on the Relationship between Perceived Personalization and Attitudes Toward Advertising Investigating how XAI affects consumers’ perceptions of personalization and their subsequent attitudes toward advertising is confirmed by Tsai, et al. [20] in their study. The authors found that when consumers were presented with explanations of how AI systems generated personalized recommendations, they developed higher trust in the AI. As a result, their perception of the personalization was enhanced, leading to a more positive attitude toward the advertising. The authors emphasized the importance of transparency in AI-driven personalization to improve consumer satisfaction and engagement with advertising content. XAI quality significantly improved users' trust in the advertising system, which in turn strengthened the relationship between perceived personalization and consumer attitudes toward advertising. Evans et al., explored how explanations enhanced consumers' perception of relevance, further improving consumers' ad experience [36]. Exploring the role of XAI based advertising on consumer engagement. The authors hypothesized that high quality XAI would enhance the perceived personalization of advertising and increase engagement and ad effectiveness [37]. Their findings supported this hypothesis, showing that consumers engaged more with personalized advertising when they understood how the AI systems arrived at their recommendations. Explainability provided clarity, enhancing the perceived relevance and boosting overall attitudes toward the advertising. Consumers were more likely to accept and appreciate personalized advertising when the AI systems provided clear, understandable explanations for generating customized content. XAI quality positively influenced perceived personalization, resulting in higher levels of ad acceptance and more favorable attitudes toward advertising in general [38]. Therefore, the following hypothesis is proposed: H4: Perceived quality of XAI moderates the relationship between perceived personalization and attitudes toward personalized advertising. 1234 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate 2.8. The Moderating Role of XAI Quality Perception on the Relationship between Privacy Concerns and Attitudes Toward Advertising Privacy concerns significantly influence individuals' attitudes toward personalized advertising [36]. Consumers often express concerns about the impact of AI technology on privacy, including personalized advertising [39]. XAI has attracted attention for its ability to bring transparency and interpretability to AI systems [38]. XAI can improve consumers' understanding of how AI algorithms make decisions, potentially alleviating privacy concerns associated with AI technology. The transparency provided by XAI can promote trust and reduce concerns related to data privacy [38]. Furthermore, the perceived quality of XAI may act as a moderator in the relationship between privacy concerns and attitudes toward personalized advertising. Therefore, the following hypothesis is proposed: H5: Perceived quality of XAI moderates the relationship between privacy concerns and attitudes toward personalized advertising. Through a comprehensive review of the literature and relevant foundational theories, the authors identified a research gap pertaining to the topic under study. Based on this insight, they formulated a hypothesis and proposed a research model. The model integrates elements from the Elaboration Likelihood Model (ELM), the Cognitive-Motivational-Emotional Heuristic-AI (CMEH-AI), and Privacy Calculus Theory (Figure 4). Figure 4. Proposed research model. 3. Research Methods 3.1. Instrument Development Scales are important tools for building survey questionnaires. Based on reviewing studies related to the topic, the authors inherited and synthesized the scales, adjusted the scales to suit the research context, and then carried out the following steps: data analysis and testing of research hypotheses. The measurement structures use a 5-point Likert scale, from 1 is completely disagree, to 5 is completely agree. Appendix 1 presents the measurement scales inherited from previous studies. 3.2. Data Collection Data were collected through online survey methods using convenience sampling. More than 1,500 questionnaires were sent out via email and other communication channels, and after more than 2 months of sending, the number of responses received was 545 (36%). After cleaning and removing 20 invalid responses, the number of remaining questionnaires included in the official study was 525. 1235 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate 3.3. Data Analysis This study uses Structural Equation Modeling (SEM) to test the proposed research model. Linear structural models represent cause-and-effect relationships between independent and dependent constructs [40]. Therefore, using CB-SEM is entirely suitable for this study. The tool used to perform the analysis is AMOS 28 software, which uses a maximum likelihood estimation method. SPSS 26 was used for exploratory factor analysis (EFA). Confirmatory factor analysis (CFA) and average variance extracted (AVE) were used to evaluate the measurement model. 4. Research Results 4.1. Profile of the Sample The statistical results of the research sample show that female demographics account for 64% of the sample, male demographics account for 36%, this ratio shows that women tend to shop on social media more than men. Regarding age distribution, people under 18 years old account for 6.3%, from 18 to 35 years old account for 59.2%, from 35 to 50 years old account for 26.1% and people over 50 years old account for 8.4%. Regarding income, the collected data shows that customers with income under 10 million VND account for 12.2%, from 10 to 20 million account for 63.2%, from 20 to 30 million account for 18.3%, over 30 million account for 6.3%. For the question 'Are you satisfied with shopping on social media?', the number of people answering Yes is 375, accounting for 71% and the number of people answering No is 150, accounting for 29%. Table 1 shows detailed statistics of the research sample. Table 1. Research sample size and structure. Characteristics Frequency Percentage Gender Male Female 189 336 36 64 Age Under 18 years old 18-35 years old 35-50 years old Over 50 years old 33 311 137 44 6.3 59.2 26.1 8.4 Income (million VND) Under 10 From 10 to under 20 From 20 to 30 Over 30 64 332 96 33 12.2 63.2 18.3 6.3 Are you satisfied with shopping on social media? Yes No 375 150 71 829 4.2. Assessment of Measurement Model The analysis was performed in two steps. Step one analyzes each independent factor to more clearly determine the contents that must be considered in testing the scale's reliability. Step two uses the oblique rotation method to test the convergent and discriminant validity of all variables in the model. The criteria for calculating the reliability of variables are as follows: the threshold value of Cronbach's Alpha coefficient and composite reliability is 0.7 [40]. Initially, the analysis encompassed 32 variables. However, after a meticulous examination of Cronbach's Alpha reliability, it was found that the values of the three variables SYS1, PEU1, and SAT5 were all small and unreliable. Consequently, these variables were eliminated from the model, resulting in a refined structural model analysis with 29 variables. Notably, the smallest alpha value is 0.711, corresponding to the smallest value of CR, which is 0.811. All of these elements adhere to the standard, further validating the model. 1236 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate For convergent validity, the AVE threshold is 0.5, and the minimum factor loading is 0.6 [41]. The EFA results of all variables show that the research model's concepts achieve convergent and discriminant validity. The results of the variance analysis extracted using EFA for the scales are presented in a summary in Table 2. Table 2. Reliability and validity measures. Variables Items Loading C.A CR AVE Purchase decision (PD) PD1 0.767 0.86 0.863 0.559 PD2 0.701 PD3 0.695 PD4 0.793 PD5 0.778 Perceived of personalization (PP) PP1 .792 0.87 0.875 0.583 PP2 .651 PP3 .817 PP4 .804 PP5 .706 Attitude advertising (AA) AA1 .792 0.84 0.836 0.560 AA2 .704 AA3 .700 AA4 .641 Privacy concerns (PC) PC1 .849 0.86 0.857 0.601 PC2 .752 PC3 .711 PC5 .760 Perceived quality of XAI (PQ) PQ1 .708 0.82 0.825 0.541 PQ2 .731 PQ3 .764 PQ4 .744 The results of the CFA analysis presented show that the critical model is suitable because the general goodness-of-fit indexes are: Chi-Square/df =1.248 (<3); GFI = 0.959; CFI = 0.991 (>0.9); TLI = 0.989 (>0.9); RMSEA = 0.022 (<0.08) meets the requirement (Figure 2). 1237 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate Figure 5. CFA model. The results of the discriminant validity test are presented in Table 3; the square root of AVE is larger than the correlation value in the rows and columns. Therefore, according to Fornell-Larcker criteria [42], the theoretical model's research concepts meet the discriminant validity requirement. Table 3. Discriminant validity test. Items PD PP AA PC PQ PD 0.748 PP 0.424 0.764 AA 0.736 0.618 0.749 PC -0.441 -0.249 -0.593 0.775 PQ -0.106 -0.147 -0.034 -0.016 0.735 1238 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate 4.3. Structural Model and Hypotheses Test Five hypotheses were developed to validate the conceptual framework of the study. The results of the analysis are presented in Figure 3. The overall fit values of the model all meet technical requirements: GFI = 0.965; TLI = 0.987 (>0.9); CFI = 0.989 (>0.9); RMSEA = 0.053 (<0.08) meets the requirements. Figure 6. Test research hypotheses using SEM. Research results show that the factors of PP, PC have a positive impact on the AA; AA has a positive effect on the PD Table 4. Table 4. Regression weights of theoretical relationships. Hypothesis Relationship Wei S.E. C.R p Conclusion H1 AA <--- PP 0.377 0.036 10.461 *** Accepted H2 AA <--- PC -0.319 0.031 -10.250 *** Accepted H3 PD <--- AA 0.701 0.058 12.050 *** Accepted Figure 7. Results of moderating variables on the relationship between PP and AA. 1239 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate 4.4. Moderator Variable Analysis Results Variable Int_1 has a t-test p_value of 0.020 < 0.05 which is statistically significant, variable PQ plays a moderating role in the impact of PP on AA. The regression coefficient of the moderating impact is 0.130 > 0, so when PQ increases, PP will have a stronger impact on AA. Figure 8. Results of moderating variables on the relationship between PC and AA Variable Int_1 has a t-test p_value of 0.022 < 0.05 which is statistically significant, variable PQ plays a moderating role in the impact of PC on AA. The regression coefficient of the moderating impact is 0.084 > 0, so when PQ increases, PC will have a weaker impact on AA. 5. Discussion Hypothesis 1 (H1): The perceived personalization of advertising positively impacts attitudes toward advertising. The results confirm this hypothesis, indicating that consumers' perceptions of personalization in advertising foster positive attitudes. This aligns with the findings of [30], who demonstrated that well-personalized content enhances engagement and positive reception of advertising. Similarly, Cacioppo, et al. [15] found that personalized content in advertising to deeper processing and stronger attitudes toward advertising, thus supporting this study’s findings. Hypothesis 2 (H2): Privacy concerns negatively impact attitudes toward personalized advertising. This study's findings support this hypothesis, with privacy concerns decreasing positive attitudes toward advertisements. This outcome is consistent with van Ooijen, et al. [32] and Lina and Setiyanto [31] who noted that increased privacy concerns lead to skepticism and reluctance towards advertisements. Further support comes from Van Doorn and Hoekstra [43] who discussed how privacy concerns create barriers in consumer trust, thereby affecting attitudes. Hypothesis 3 (H3): Attitudes toward advertising positively impact purchasing decisions. The study results confirm this, showing that positive attitudes directly enhance purchasing decisions. This is in line with Genç and Turna [33] who found that favorable attitudes toward advertisements on social media platforms significantly boost purchase intentions, particularly among active social media users. Melati, et al. [34] also highlighted that consumer attitudes are crucial in influencing decisions, especially when the ad content is engaging or features credible figures, reinforcing this study’s outcome. Hypothesis 4 (H4): Perceived quality of XAI moderates the relationship between perceived personalization and attitudes toward advertising. Findings confirm that high-quality XAI enhances consumers' perception of personalization, thereby improving their attitudes toward advertisements. These results are consistent with Tsai, et al. [20], who emphasized the importance of transparency in AI systems to build trust and positively influence consumer perceptions. Similarly, Huang and Wang 1240 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate [37] demonstrated that when AI explanations are clear, they enhance the relevance of personalized advertising and foster better engagement, which aligns with the findings of this study. Hypothesis 5 (H5): Perceived quality of XAI moderates the relationship between privacy concerns and attitudes toward advertising. The results support this hypothesis, showing that effective XAI quality can mitigate negative impacts caused by privacy concerns. This finding resonates with Vilone and Longo [38], who argued that XAI improves transparency, which in turn alleviates privacy-related skepticism. Evans, et al. [36] also noted that transparency through XAI has the potential to reduce privacy-related fears, supporting this study’s conclusions. 6. Conclusions Based on the research findings, this study concludes that perceived personalization and privacy concerns significantly influence consumer attitudes toward advertising, which subsequently impacts purchasing decisions on social media platforms. Specifically, the positive impact of perceived personalization highlights the value of tailored content in creating favorable consumer attitudes and driving engagement. Conversely, privacy concerns negatively affect attitudes toward advertising, underscoring the importance of addressing consumer apprehensions about data usage in personalized marketing efforts. The study also confirms that the quality of XAI plays a moderating role in these relationships. High quality XAI enhances perceptions of personalization by increasing transparency, thereby building consumer trust and improving attitudes toward AI-driven advertisements. Furthermore, XAI quality mitigates the adverse effects of privacy concerns, showing that when consumers understand AI decision- making processes, they are more likely to respond positively to personalized advertising. The distribution of this study holds significant importance and value in providing insights into how Explainable Artificial Intelligence (XAI) technology can enhance the effectiveness of personalized advertising on social media platforms, thereby influencing consumer purchase decisions. Investigating the role of XAI in advertising not only helps mitigate concerns about the lack of transparency in AI but also expands understanding of how factors such as trust and transparency can drive consumer engagement and increase conversion rates in digital marketing strategies. The value of this research lies in the development of a model that analyzes the relationship between XAI, personalized advertising, and purchase decisions, offering valuable insights for marketers. The research findings can aid in optimizing advertising strategies, fostering consumer trust, and enhancing customer loyalty. Furthermore, the study will clarify factors such as privacy concerns and their relationship with consumer attitudes toward advertising, thereby providing practical conclusions and recommendations for managers and marketing professionals in developing more transparent and effective advertising strategies. While this study offers important insights, several limitations should be noted. First, the research sample was geographically limited to specific urban areas in Vietnam, which may limit the generalizability of the findings to broader populations. Expanding the survey scope to include diverse regions and cultural contexts could enhance the robustness and applicability of the results. Second, the study focused on consumers between the ages of 18 and 60, potentially overlooking the attitudes and behaviors of younger or older demographics. Future research could include a wider age range to capture variations in responses across different generational groups. Lastly, this study primarily examined technical and emotional factors influencing consumer behavior in AI-driven advertising but did not account for evolving technological or social factors, such as changes in privacy regulations or AI capabilities. Future research could explore additional variables, such as regulatory impacts or ethical considerations, to provide a more comprehensive understanding of the factors shaping consumer responses to AI-powered personalization. In summary, while these findings contribute to the theoretical and practical understanding of consumer attitudes toward AI-driven advertising, addressing these limitations will provide a more 1241 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate nuanced perspective on how personalization, privacy, and explainability shape consumer behavior across various contexts and demographic groups. Transparency: The authors confirm that the manuscript is an honest, accurate, and transparent account of the study; that no vital features of the study have been omitted; and that any discrepancies from the study as planned have been explained. This study followed all ethical practices during writing. Copyright: © 2025 by the authors. This open-access article is distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). References [1] S. Noranee and A. K. bin Othman, "Understanding consumer sentiments: Exploring the role of artificial intelligence in marketing," JMM17: Jurnal Ilmu Ekonomi dan Manajemen, vol. 10, no. 1, pp. 15-23, 2023. https://doi.org/10.30996/jmm17.v10i1.8690 [2] H. N. D. Senyapar, "Artificial intelligence in marketing communication: A comprehensive exploration of the integration and impact of AI," Technium Social Sciences Journal, vol. 55, pp. 64-81, 2024. https://doi.org/10.47577/tssj.v55i1.10651 [3] V. Kumar, B. Rajan, R. Venkatesan, and J. Lecinski, "Understanding the role of artificial intelligence in personalized engagement marketing," California Management Review, vol. 61, no. 4, pp. 135-155, 2019. https://doi.org/10.1177/0008125619859317 [4] W. Rodgers and T. Nguyen, "Advertising benefits from ethical artificial intelligence algorithmic purchase decision pathways," Journal of Business Ethics, vol. 178, no. 4, pp. 1043-1061, 2022. https://doi.org/10.1007/s10551-022- 05048-7 [5] D. Gunning, M. Stefik, J. Choi, T. Miller, S. Stumpf, and G.-Z. Yang, "XAI—Explainable artificial intelligence," Science Robotics, vol. 4, no. 37, p. eaay7120, 2019. [6] V. Nesterenko and O. Olefirenko, "The impact of AI development on the development of marketing communications," Marketing i Menedžment Innovacij, vol. 14, no. 1, pp. 169-181, 2023. https://doi.org/10.21272/mmi.2023.1-15 [7] H. Sun, M. Z. Zafar, and N. Hasan, "Employing natural language processing as artificial intelligence for analyzing consumer opinion toward advertisement," Frontiers in Psychology, vol. 13, p. 856663, 2022. https://doi.org/10.3389/fpsyg.2022.856663 [8] A. R. McAlister and D. Bargh, "Dissuasion: The elaboration likelihood model and young children," Young Consumers, vol. 17, no. 3, pp. 210-225, 2016. https://doi.org/10.1108/yc-02-2016-00580 [9] Y. Pan, "Research on the influence of advertising content on consumer purchasing behavior based on elaboration likelihood model (ELM)," Journal of Education, Humanities and Social Sciences, vol. 27, pp. 439-444, 2024. https://doi.org/10.54097/mk0f1454 [10] S. Park, X. Wei, and H. Lee, "Revisiting the elaboration likelihood model in the context of a virtual influencer: A comparison between high‐and low‐involvement products," Journal of Consumer Behaviour, vol. 23, no. 4, pp. 1638- 1652, 2024. https://doi.org/10.1002/cb.2290 [11] M. H. Shahab, E. Ghazali, and M. Mohtar, "The role of elaboration likelihood model in consumer behaviour research and its extension to new technologies: A review and future research agenda," International Journal of Consumer Studies, vol. 45, no. 4, pp. 664-689, 2021. https://doi.org/10.1111/ijcs.12658 [12] Y. Lee and J. Koo, "Can a celebrity serve as an issue‐relevant argument in the elaboration likelihood model?," Psychology & Marketing, vol. 33, no. 3, pp. 195-208, 2016. [13] H. Wang and J. Yan, "Effects of social media tourism information quality on destination travel intention: Mediation effect of self-congruity and trust," Frontiers in Psychology, vol. 13, p. 1049149, 2022. [14] D. Lee, K. Hosanagar, and H. S. Nair, "Advertising content and consumer engagement on social media: Evidence from Facebook," Management science, vol. 64, no. 11, pp. 5105-5131, 2018. [15] J. T. Cacioppo, R. E. Petty, C. F. Kao, and R. Rodriguez, "Central and peripheral routes to persuasion: An individual difference perspective," Journal of Personality and Social Psychology, vol. 51, no. 5, p. 1032, 1986. https://doi.org/10.1037//0022-3514.51.5.1032 [16] D. Shin, "The effects of explainability and causability on perception, trust, and acceptance: Implications for explainable AI," International Journal of Human-computer Studies, vol. 146, p. 102551, 2021. https://doi.org/10.1016/j.ijhcs.2020.102551 https://creativecommons.org/licenses/by/4.0/ https://doi.org/10.30996/jmm17.v10i1.8690 https://doi.org/10.47577/tssj.v55i1.10651 https://doi.org/10.1177/0008125619859317 https://doi.org/10.1007/s10551-022-05048-7 https://doi.org/10.1007/s10551-022-05048-7 https://doi.org/10.21272/mmi.2023.1-15 https://doi.org/10.3389/fpsyg.2022.856663 https://doi.org/10.1108/yc-02-2016-00580 https://doi.org/10.54097/mk0f1454 https://doi.org/10.1002/cb.2290 https://doi.org/10.1111/ijcs.12658 https://doi.org/10.1037/0022-3514.51.5.1032 https://doi.org/10.1016/j.ijhcs.2020.102551 1242 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate [17] C. Meske, E. Bunde, J. Schneider, and M. Gersch, "Explainable artificial intelligence: Objectives, stakeholders, and future research opportunities," Information Systems Management, vol. 39, no. 1, pp. 53-63, 2022. https://doi.org/10.1080/10580530.2020.1849465 [18] J. Gerlings, M. S. Jensen, and A. Shollo, "Explainable AI, but explainable to whom? An exploratory case study of xAI in healthcare," Handbook of Artificial Intelligence in Healthcare: Vol 2: Practicalities and Prospects, pp. 169-198, 2022. https://doi.org/10.1007/978-3-030-83620-7_7 [19] D. Shin and Y. J. Park, "Role of fairness, accountability, and transparency in algorithmic affordance," Computers in Human Behavior, vol. 98, pp. 277-284, 2019. https://doi.org/10.1016/j.chb.2019.04.019 [20] J. Y. Tsai, S. Egelman, L. Cranor, and A. Acquisti, "The effect of online privacy information on purchasing behavior: An experimental study," Information Systems Research, vol. 22, no. 2, pp. 254-268, 2011. https://doi.org/10.1287/isre.1090.0260 [21] X. Chen, J. Sun, and H. Liu, "Balancing web personalization and consumer privacy concerns: Mechanisms of consumer trust and reactance," Journal of Consumer Behaviour, vol. 21, no. 3, pp. 572-582, 2022. https://doi.org/10.1002/cb.1947 [22] T. Hongsuchon and J. Li, "Accessing the influence of consumer participation on purchase intention toward community group buying platform," Frontiers in Psychology, vol. 13, p. 887959, 2022. https://doi.org/10.3389/fpsyg.2022.887959 [23] T. Schürmann, N. Gerber, and P. Gerber, "Benefits of formalized computational modeling for understanding user behavior in online privacy research," Journal of Intellectual Capital, vol. 21, no. 3, pp. 431-458, 2020. https://doi.org/10.1108/jic-05-2019-0126 [24] Z. Sun, G. Zang, Z. Wang, S. Ge, W. Liu, and K. Wang, "Determining factors affecting the user's intention to disclose privacy in online health communities: A dual-calculus model," Frontiers in Public Health, vol. 11, p. 1109093, 2023. https://doi.org/10.3389/fpubh.2023.1109093 [25] A. P. M. Richarde, P. H. M. Prado, and J. B. Ferreira, "Sinais de Privacidade: Explorando Relações entre Cookies e Intenção de Compra On-Line," Revista de Administração Contemporânea, vol. 27, p. e220311, 2023. [26] C. E. Tucker, "Social networks, personalized advertising, and privacy controls," Journal of Marketing Research, vol. 51, no. 5, pp. 546-562, 2014. https://doi.org/10.1509/jmr.10.0355 [27] D. A. Cooper, T. Yalcin, C. Nistor, M. Macrini, and E. Pehlivan, "Privacy considerations for online advertising: A stakeholder’s perspective to programmatic advertising," Journal of Consumer Marketing, vol. 40, no. 2, pp. 235-247, 2023. https://doi.org/10.1108/jcm-04-2021-4577 [28] M. J. Culnan and P. K. Armstrong, "Information privacy concerns, procedural fairness, and impersonal trust: An empirical investigation," Organization Science, vol. 10, no. 1, pp. 104-115, 1999. https://doi.org/10.1287/orsc.10.1.104 [29] A. Holzinger, G. Langs, H. Denk, K. Zatloukal, and H. Müller, "Causability and explainability of artificial intelligence in medicine," Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, vol. 9, no. 4, p. e1312, 2019. https://doi.org/10.1002/widm.1312 [30] S. Y. Ho and D. Bodoff, "The effects of web personalization on user attitude and behavior," MIS Quarterly, vol. 38, no. 2, pp. 497-A10, 2014. https://doi.org/10.25300/misq/2014/38.2.08 [31] L. F. Lina and A. Setiyanto, "Privacy concerns in personalized advertising effectiveness on social media," Sriwijaya International Journal of Dynamic Economics and Business, pp. 147-156, 2021. https://doi.org/10.29259/sijdeb.v1i2.147- 156 [32] I. van Ooijen, C. M. Segijn, and S. J. Opree, "Privacy cynicism and its role in privacy decision-making," Communication Research, vol. 51, no. 2, pp. 146-177, 2024. https://doi.org/10.1177/00936502211060984 [33] R. Genç and G. B. Turna, "The mediating effect of attitude towards online advertising in the influence of social media addiction on online purchase intention," Business & Management Studies: An International Journal, vol. 11, no. 2, pp. 511-531, 2023. https://doi.org/10.15295/bmij.v11i2.2228 [34] I. Melati, T. Indira Budiwan, and H. Sarjono, "The influence of celebrity endorsement on consumer’s attitude: A study case of smartphone brands in Jakarta, Indonesia," International Journal of Engineering and Technology, vol. 7, no. 4, pp. 30-32, 2018. [35] K.-C. Yang, C.-H. Huang, C. Yang, and S. Y. Yang, "Consumer attitudes toward online video advertisement: YouTube as a platform," Kybernetes, vol. 46, no. 5, pp. 840-853, 2017. https://doi.org/10.1108/k-03-2016-0038 [36] R. A. Evans et al., "Impact of COVID-19 on immunocompromised populations during the Omicron era: Insights from the observational population-based INFORM study," The Lancet Regional Health–Europe, vol. 35, 2023. https://doi.org/10.1016/j.lanepe.2023.100747 [37] G. Huang and Z. Wang, "The Application of Virtual Reality Technology in the Coordination and Interaction of Regional Economy and Culture in the Sustainable Development of Ecotourism," Mathematical Problems in Engineering, vol. 2022, 2022. [38] G. Vilone and L. Longo, "Explainable artificial intelligence: A systematic review," arXiv preprint arXiv:2006.00093, 2020. https://doi.org/10.1080/10580530.2020.1849465 https://doi.org/10.1007/978-3-030-83620-7_7 https://doi.org/10.1016/j.chb.2019.04.019 https://doi.org/10.1287/isre.1090.0260 https://doi.org/10.1002/cb.1947 https://doi.org/10.3389/fpsyg.2022.887959 https://doi.org/10.1108/jic-05-2019-0126 https://doi.org/10.3389/fpubh.2023.1109093 https://doi.org/10.1509/jmr.10.0355 https://doi.org/10.1108/jcm-04-2021-4577 https://doi.org/10.1287/orsc.10.1.104 https://doi.org/10.1002/widm.1312 https://doi.org/10.25300/misq/2014/38.2.08 https://doi.org/10.29259/sijdeb.v1i2.147-156 https://doi.org/10.29259/sijdeb.v1i2.147-156 https://doi.org/10.1177/00936502211060984 https://doi.org/10.15295/bmij.v11i2.2228 https://doi.org/10.1108/k-03-2016-0038 https://doi.org/10.1016/j.lanepe.2023.100747 1243 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate [39] C. M. Segijn, H. A. Voorveld, and K. A. Vakeel, "The role of ad sequence and privacy concerns in personalized advertising: An eye-tracking study into synced advertising effects," Journal of Advertising, vol. 50, no. 3, pp. 320-329, 2021. https://doi.org/10.1080/00913367.2020.1870586 [40] J. F. Hair Jr, M. L. D. da Silva Gabriel, and V. K. Patel, "Modelagem de Equações Estruturais Baseada em Covariância (CB-SEM) com o AMOS: Orientações sobre a sua aplicação como uma Ferramenta de Pesquisa de Marketing," REMark-Revista Brasileira de Marketing, vol. 13, no. 2, pp. 44-55, 2014. https://doi.org/10.5585/remark.v13i2.2718 [41] J. F. Hair Jr, M. LDS Gabriel, D. d. Silva, and S. Braga, "Development and validation of attitudes measurement scales: Fundamental and practical aspects," RAUSP Management Journal, vol. 54, no. 4, pp. 490-507, 2019. https://doi.org/10.1108/rausp-05-2019-0098 [42] C. Fornell and D. F. Larcker, "Evaluating structural equation models with unobservable variables and measurement error," Journal of Marketing Research, vol. 18, no. 1, pp. 39-50, 1981. https://doi.org/10.1177/002224378101800104 [43] J. Van Doorn and J. C. Hoekstra, "Customization of online advertising: The role of intrusiveness," Marketing Letters, vol. 24, pp. 339-351, 2013. https://doi.org/10.1007/s11002-012-9222-1 [44] T. Shanahan, T. P. Tran, and E. C. Taylor, "Getting to know you: Social media personalization as a means of enhancing brand loyalty and perceived quality," Journal of Retailing and Consumer Services, vol. 47, pp. 57-65, 2019. https://doi.org/10.1016/j.jretconser.2018.10.007 [45] C. A. Lin and T. Kim, "Predicting user response to sponsored advertising on social media via the technology acceptance model," Computers in Human Behavior, vol. 64, pp. 710-718, 2016. https://doi.org/10.1016/j.chb.2016.07.027 [46] R. Hoffman, G. Klein, S. Mueller, M. Jalaeian, and C. Tate, "The stakeholder playbook for explaining AI systems," 2021. https://doi.org/10.31234/osf.io/9pqez [47] R. R. Hoffman, "A taxonomy of emergent trusting in the human–machine relationship," Cognitive Systems Engineering, pp. 137-164, 2017. https://doi.org/10.1201/9781315572529-8 [48] R. P. Singh and N. Banerjee, "Exploring the influence of celebrity credibility on brand attitude, advertisement attitude and purchase intention," Global Business Review, vol. 19, no. 6, pp. 1622-1639, 2018. https://doi.org/10.1177/0972150918794974 [49] M. A. Shareef, B. Mukerji, M. A. A. Alryalat, A. Wright, and Y. K. Dwivedi, "Advertisements on Facebook: Identifying the persuasive elements in the development of positive attitudes in consumers," Journal of Retailing and Consumer Services, vol. 43, pp. 258-268, 2018. https://doi.org/10.1016/j.jretconser.2018.04.006 [50] I. Erkan and C. Evans, "The influence of eWOM in social media on consumers’ purchase intentions: An extended approach to information adoption," Computers in Human Behavior, vol. 61, pp. 47-55, 2016. https://doi.org/10.1016/j.chb.2016.03.003 [51] A. J. Kim and E. Ko, "Do social media marketing activities enhance customer equity? An empirical study of luxury fashion brand," Journal of Business Research, vol. 65, no. 10, pp. 1480-1486, 2012. https://doi.org/10.1016/j.jbusres.2011.10.014 [52] S.-C. Chu and Y. Kim, "Determinants of consumer engagement in electronic word-of-mouth (eWOM) in social networking sites," International Journal of Advertising, vol. 30, no. 1, pp. 47-75, 2011. https://doi.org/10.2501/ija-30- 1-047-075 Appendix 1. Instruments development. No Variable Instruments References 1 Perceived of Personalization (PP) PP1:I believe the ad is customized to my needs [44] PP2: Overall, this ad is suitable for my situation PP3: This ad makes me feel like I'm the only customer PP4. I think this ad allows me to order products specifically designed for me PP5: This ad offers purchasing recommendations tailored to my needs 2 Privacy Concerns (PC) PC1: Can users feel secure about providing sensitive information to Facebook? [45] PC2: Do users feel secure about posting personal information on someone's personal Facebook? PC3: Do users feel secure about sending personal information via Facebook messenger service? PC4: Do users feel secure about keeping someone's personal information on Facebook confidential? https://doi.org/10.1080/00913367.2020.1870586 https://doi.org/10.5585/remark.v13i2.2718 https://doi.org/10.1108/rausp-05-2019-0098 https://doi.org/10.1177/002224378101800104 https://doi.org/10.1007/s11002-012-9222-1 https://doi.org/10.1016/j.jretconser.2018.10.007 https://doi.org/10.1016/j.chb.2016.07.027 https://doi.org/10.31234/osf.io/9pqez https://doi.org/10.1201/9781315572529-8 https://doi.org/10.1177/0972150918794974 https://doi.org/10.1016/j.jretconser.2018.04.006 https://doi.org/10.1016/j.chb.2016.03.003 https://doi.org/10.1016/j.jbusres.2011.10.014 https://doi.org/10.2501/ija-30-1-047-075 https://doi.org/10.2501/ija-30-1-047-075 1244 Edelweiss Applied Science and Technology ISSN: 2576-8484 Vol. 9, No. 2: 1228-1244, 2025 DOI: 10.55214/25768484.v9i2.4745 © 2025 by the authors; licensee Learning Gate 3 Perceived Quality of XAI (PQ) PQ1: I feel that the tool works well. [46] [47] PQ2: I feel secure that when I rely on the tool, I will get the correct answer. PQ3: The tool is effective in that it works very quickly. PQ4: This tool is very reliable. I can trust that it is always right. 4 Attitude Advertising (AA) AA1: Interesting advertisement [48] [49] AA2: Advertising is trusted AA3: The advertisement has attracted attention AA4: I like to see product information on my Facebook 5 Purchase Decision (PD) PD1: I am interested in the content brands post on social media. [50] [51] [52] PD2: Promotions on social media motivate me to buy products. PD3: I will to buy products that are advertised on social media. PD4: Positive reviews on social media influence my purchasing decisions. PD5: Comments on social media influence my purchasing decisions.