The proliferation of social media has fundamentally altered information consumption, with billions of users spending significant time daily on these platforms. Traditional advertising struggles with declining visibility due to ad blockers, pushing marketers towards social media influencer (SMI) collaborations. SMIs, distinct from traditional celebrities, are perceived as influential individuals who have cultivated engaged audiences through consistent, valuable content creation. They function as third-party endorsers, leveraging established relationships with stakeholders to influence others. Influencer marketing offers various objectives, including brand awareness enhancement, increased product interest, and direct sales, ultimately impacting brand value and revenue. While research on influencer marketing has expanded, gaps remain regarding the factors shaping parasocial relationships (PSRs) between SMIs and followers, the impacts of these relationships on consumer responses, and the role of consumers' persuasion knowledge. This study addresses these gaps by investigating what drives PSR formation, how PSRs influence brand credibility and purchase intention, and how persuasion knowledge moderates these relationships.

The elaboration likelihood model (ELM) of persuasion underpins this research, suggesting that high-involvement individuals focus on message arguments (central route), while low-involvement individuals are swayed by peripheral cues like source characteristics (peripheral route). SMIs provide both informative and entertaining content, and authenticity fosters trust, enhancing their persuasive power. Homophily, or perceived similarity between SMI and follower, can also stimulate relationship development and information sharing. Parasocial relationships (PSRs), unilateral relationships formed with media personalities, are crucial in influencer marketing, fostering feelings of belonging and companionship. While PSRs have been studied in traditional celebrity contexts, their application to SMIs is a growing area of research. Previous studies highlight the importance of SMI credibility, user characteristics, and content characteristics in building PSRs. The impact of persuasion knowledge—consumers' understanding of persuasion techniques—is another under-researched area. This study therefore focuses on the antecedents of PSRs (informative and entertainment content value, authenticity, and homophily), the impact of PSRs on brand credibility and purchase intention, and the moderating role of persuasion knowledge.

A questionnaire survey was designed based on existing literature, measuring constructs like informative and entertainment value, authenticity, homophily, PSR, persuasion knowledge, brand credibility, and purchase intention using 7-point Likert scales. The questionnaire, initially piloted with 50 respondents, was distributed online to 800 respondents in China, resulting in 216 valid responses after screening for relevant experience and attention checks. The sample was predominantly female (57.4%) and aged 18-25 (57.8%). Partial least squares structural equation modeling (PLS-SEM) using SmartPLS 3.3.9 was employed for data analysis, chosen for its suitability with smaller sample sizes and complex models, and its ability to handle latent variables effectively. Harman's single-factor test was used to address potential common method bias, and the measurement model was assessed for reliability and validity. The structural model was evaluated using fit indices like SRMR and NFI, and bootstrapping analysis was performed to test the hypotheses.

The results supported the hypotheses that informative content value (β = 0.174, p < 0.05), influencer authenticity (β = 0.309, p < 0.01), and homophily (β = 0.230, p < 0.05) positively influenced PSRs. Entertainment value showed no significant relationship with PSRs. PSRs positively influenced brand credibility (β = 0.550, p < 0.001) and purchase intention (β = 0.490, p < 0.001). Followers' persuasion knowledge negatively moderated the relationship between PSRs and brand credibility (β = -0.077, p < 0.05), but not between PSRs and purchase intention. Effect sizes indicated large effects of PSRs on brand credibility and medium effects on purchase intention, while effects of content value, authenticity, and homophily on PSRs were small. The blindfolding procedure confirmed the model's predictive relevance (Q2 > 0 for PSR, brand credibility, and purchase intention). Control variables (age and gender) were not significant.

The study findings support the importance of informative content, authenticity, and homophily in establishing strong PSRs between SMIs and their followers. These PSRs, in turn, significantly impact followers' brand perceptions and purchase behaviors. The absence of a significant relationship between entertainment value and PSRs contrasts with some previous research, suggesting a potential contextual influence or that informative content might be prioritized over entertainment. The negative moderating effect of persuasion knowledge on the PSR-brand credibility relationship suggests that increased awareness of persuasion techniques may lead to a decrease in brand credibility when PSRs are established. However, this effect doesn’t apply to purchase intention, suggesting that other factors like trust in influencers override the negative influence of persuasion knowledge on purchase decisions.

This research contributes by integrating ELM, the advertising value model, and persuasion knowledge models to explain the effectiveness of influencer marketing through the lens of PSRs. It highlights the importance of informative content, authenticity, and homophily in SMI selection and the significant role of PSRs in shaping brand perceptions and purchase decisions. Future research could explore the entertainment value further, examine other influencing factors, investigate additional moderators, and consider different SMI categories.

The study's limitations include its focus on a Chinese sample, which may limit generalizability. The relatively small sample size also needs to be considered. Future research should address this by using larger, more diverse samples across different cultures and platforms. Additionally, the study focused on a specific set of variables; other variables could also be studied.