Close Menu
    Saturday, August 8
    Accounting

    Frontiers | Sponsored advertising on social media and purchase intention: an attitude-mediated TPB–ELM model among Saudi consumers

    AdminBitBy No Comments53 Mins Read
    Frontiers | Sponsored advertising on social media and purchase intention: an attitude-mediated TPB–ELM model among Saudi consumers

    Sponsored advertising on social media and purchase intention: an attitude-mediated TPB–ELM model among Saudi consumers

    • 1. Mass Communication and Public Relations Department, College of Communication and Media Technologies, Gulf University, Sanad, Bahrain

    • 2. Department of Communication, College of Arts, Education, and Social Sciences (CAESS), Abu Dhabi University, Abu Dhabi, United Arab Emirates

    Abstract

    Sponsored advertising on social media has become the dominant channel through which brands reach consumers in digitally advanced markets, yet the psychological mechanisms linking such advertising to purchase intention remain incompletely understood—particularly in culturally distinctive markets such as Saudi Arabia. Drawing on the Theory of Planned Behavior and the Elaboration Likelihood Model, this study tests a partial TPB–ELM integration in which advertising-related predictors shape purchase intention both directly and indirectly through attitude, and in which ELM persuasion routes determine whether the resulting attitude change is deep and durable or heuristic and transient. A quantitative cross-sectional survey was administered to active Saudi social media users recruited through platform-stratified quota sampling across Instagram, Snapchat, and X. Multiple linear regression, independent-samples t-tests, oneway ANOVA with Tukey HSD post-hoc testing, and Partial Least Squares Structural Equation Modeling (PLS-SEM via SmartPLS 4) were employed. The instrument demonstrated strong reliability and validity, and common method bias diagnostics confirmed no serious threat to the findings. Multiple regression showed that ad content quality was the strongest predictor of purchase intention, followed by influencer credibility/trust, consumer engagement, and ad exposure, with the full set of predictors jointly explaining a substantial share of variance. PLS-SEM confirmed all structural paths as significant, and a bootstrapped mediation analysis confirmed that attitude (TPB) partially mediates the ad exposure-purchase intention relationship—extending prior Saudi and regional evidence on TPB attitude mediation by testing this pathway within an integrated TPB–ELM structural model for sponsored advertising. Purchase intention varied significantly by age, with the 25–34 cohort showing the highest scores; gender differences were treated as exploratory given residual sample imbalance.

    1 Introduction

    Saudi Arabia stands among the world’s most digitally engaged consumer markets. Internet penetration has reached 99%; approximately 22.97 million Saudis actively use social media as of 2025, and the average user devotes close to three hours each day to platform activity (; ). Social media advertising expenditure in the Kingdom is projected to reach USD 579.65 million in 2025 and to grow at a compound annual rate of 12.81% (). A predominantly young population—with the majority of active users under 35—further sharpens the digital profile of Saudi consumer culture, making the Kingdom a commercially significant and theoretically distinctive setting for sponsored advertising research.

    Sponsored advertising—paid content placed within the feeds, stories, and discovery pages of social media platforms to promote brands, products, or services—has become the primary mechanism through which marketers convert digital attention into purchase behavior. According to , 80% of Saudi consumers report that social media advertising influences their purchase decisions, compared to a global average of 72%. Domestic research has repeatedly found positive associations between sponsored influencer content and actual purchasing behavior across consumer segments and platform types (; ; ). Globally, the influencer marketing industry was valued at USD 24 billion in 2024, with projections reaching USD 32.55 billion by 2025 ().

    Despite this commercially significant context, the psychological mechanisms through which sponsored advertising translates exposure into purchase intention remain incompletely specified in Saudi Arabia. Prior studies have examined advertising exposure or influencer credibility separately, without situating both within a unified theoretical model that accounts for their distinct persuasion pathways (; ). Applying the Theory of Planned Behavior and the Elaboration Likelihood Model simultaneously within a single empirical framework is rare in the literature even outside the Saudi context (; ). Moreover, Saudi Arabia’s cultural-regulatory environment—shaped by collectivist social norms, the General Authority of Media Regulation, and Vision 2030’s digital empowerment agenda—creates conditions that resist uncritical generalisation from Western or East Asian markets (; ).

    This study addresses these gaps through six objectives: (1) evaluating the direct effect of sponsored advertising on purchase intention; (2) identifying which psychological and behavioral factors explain the greatest variance in purchase intention; (3) examining consumer engagement as an active elaboration mechanism; (4) testing whether advertising responsiveness differs significantly by age and gender; (5) assessing influencer credibility through its respective ELM persuasion routes; and (6) testing whether attitude (the primary TPB antecedent) partially mediates the exposure–purchase intention relationship—a mechanistic prediction that is theoretically central to the TPB–ELM integration but has not previously been tested in the Saudi sponsored advertising context.

    The integrated TPB–ELM framework offers a distinctive explanatory advantage: the ELM specifies the cognitive processing route through which advertising content generates attitude change, while the TPB specifies how that attitude change, together with perceived social pressure and perceived behavioral control, generates behavioral intention. Their combination produces a more complete and mechanistically grounded account of the advertising–behavior relationship than either theory provides alone.

    2 Theoretical framework

    The conceptual model integrates the Theory of Planned Behavior and the Elaboration Likelihood Model into a unified dual-process structure in which the ELM governs attitude formation from advertising stimuli and the TPB specifies how those formed attitudes—combined with subjective norms and perceived behavioral control—generate purchase intention. Figure 1 illustrates this integrated model.

    2.1 Theory of planned behavior

    Theory of Planned Behavior posits that behavioral intention—the proximal cognitive antecedent of actual behavior—is determined by three constructs. Attitude toward the behavior reflects the individual’s evaluative disposition toward performing the action: in the present context, the consumer’s overall appraisal of purchasing an advertised product, shaped by beliefs about its likely consequences. Subjective norms capture the perceived social pressure to perform or not perform the behavior—what significant others are believed to think, and the motivation to comply with those perceptions.

    In social media environments, subjective norms are theoretically activated through visible peer engagement metrics—the likes, shares, comments, and peer recommendations that signal social validation of a purchase—although this study does not include a dedicated subjective-norms scale and does not test this pathway empirically. Perceived behavioral control (PBC) reflects the individual’s assessment of the ease or difficulty of performing the behavior; in the social commerce context, platform affordances such as one-click purchasing, influencer discount codes, and embedded purchase links are theorized to reduce friction and elevate PBC, though PBC is likewise not separately measured here. These two constructs are retained in the conceptual discussion because they are integral to TPB and motivate future research, but the present study’s empirical contribution is confined to the attitude pathway (Section 4.2; Section 5.3).

    The TPB has shown consistent predictive validity for online purchase intention across multiple market contexts, including Gulf Arab settings (; ; ). In the sponsored advertising setting, a single sponsored post is theorized to simultaneously initiate all three TPB antecedents: it may shape evaluative attitudes through product information and affective priming, communicate subjective norms through visible peer engagement with the advertisement, and reduce behavioral barriers through embedded purchase affordances. The present study empirically tests only the attitude component of this theoretical account. An important gap in prior Saudi advertising research is the absence of formal mediation testing of this attitude pathway—a gap the present study addresses directly through Hypothesis 6.

    2.2 Elaboration likelihood model

    Elaboration Likelihood Model identifies two routes through which persuasive messages generate attitude change. When a message recipient has both the motivation and the cognitive capacity to evaluate an argument carefully, the central route dominates: persuasion depends on the logical quality, evidential weight, and coherence of the message, producing attitude changes that are stable, durable, and relatively resistant to counter-argument. When motivation or capacity is insufficient, the peripheral route takes over: attitude change is driven by heuristic cues—source attractiveness, celebrity status, social proof signals, production aesthetics—generating change that emerges quickly but tends to be more transient.

    In the sponsored social media context, ad content quality (informativeness, accuracy, comprehensiveness, and argument depth) serves as the primary central-route cue, activating deep cognitive elaboration among motivated consumers. Influencer trust and credibility (ITC) operate through both routes simultaneously: trustworthiness functions as a peripheral cue enabling heuristic acceptance without detailed argument evaluation, while influencer expertise and substantive content depth can activate central-route elaboration among highly involved consumers. Consumer engagement behaviors—liking, commenting, sharing, clicking through—represent active elaboration that shifts processing from peripheral toward central, deepening and stabilising the resulting attitude formation. Consumer involvement, partly conditioned by demographic factors such as age, moderates which processing route predominates.

    2.3 Integrated dual-process model and attitude mediation

    The TPB–ELM integration operates as follows. The ELM governs the reception and attitude-formation stage: message quality and source credibility determine the processing route, which in turn determines the depth and stability of the resulting attitude. These attitudes then enter the TPB’s antecedent structure, where they combine with perceived subjective norms and perceived behavioral control to generate purchase intention. This integration predicts that attitude partially mediates the relationship between advertising exposure and purchase intention—a prediction that is both theoretically central to the TPB and empirically testable within the PLS-SEM framework employed here.

    Prior applications of this combined framework include green consumption (), sustainable service provision (), and technology adoption (). The present study is the first to deploy it within a sponsored advertising context in the Gulf Arab region and the first to include formal bootstrapped mediation testing of the attitude pathway.

    3 Literature review

    3.1 Sponsored advertising and purchase behavior in Saudi Arabia

    Scholarly interest in the relationship between social media advertising and Saudi consumer purchasing has expanded substantially since 2020, driven by the Kingdom’s accelerating digitisation and the integration of platform-mediated commerce into everyday consumer life (; ). Across this body of work, a consistent pattern holds: sponsored advertising exposure is positively associated with purchase intention, though the magnitude of that association varies by platform, product category, and advertising format.

    , using SEM-AMOS with 297 Saudi consumers, found that social media advertising positively influences branded-product purchasing behavior, with information richness—a central ELM cue—moderating the effect. , surveying 385 Saudi Instagram users, observed that social media promotion had a modest direct effect on purchase intention but a more substantial effect on actual purchases, suggesting that content quality and consumer engagement—not exposure alone—are required to convert intention into behavior. , studying Saudi tourism consumers, confirmed that information quality, social influence, and platform usability interact to shape purchase decisions in a pattern parallel to TPB’s three antecedents.

    , surveying 345 Saudi social media users, found that attitude fully mediates the relationship between advertising utility and purchase intention—operationalizing the TPB attitude construct within the social media advertising context—and documented a strong negative association between ad irritation and purchase intention, reflecting the ELM peripheral-route backfire effect. demonstrated on Snapchat that attitudes, trust, personal norms, and celebrity endorsements collectively predict purchase intention among Saudi consumers, validating both the TPB antecedent structure and the dual-route ELM mechanism.

    3.2 Influencer credibility

    , analyzing 384 Saudi university students, found that influencer expertise, authenticity, and sponsorship disclosure were the decisive determinants of purchase intention, with Snapchat as the preferred platform. documented that among 402 Saudi youth consumers, expertise and trustworthiness predicted brand liking and purchase intention through central-route elaboration, while physical attractiveness predicted purchase intention directly through peripheral cues without passing through brand liking—providing clear field evidence of the ELM’s dual-route mechanism operating simultaneously in a Saudi influencer context. established that the majority of Saudi university students had purchased at least one influencer-promoted product, confirming that sponsored influencer content generates actual behavioral outcomes, not merely attitudinal shifts.

    Social Media Influencer Value Model—grounded explicitly in ELM—provides the theoretical foundation for the ITC construct in this study, demonstrating that influencer informational value, trustworthiness, attractiveness, and perceived similarity jointly determine follower trust in sponsored content, which then generates brand awareness and purchase intentions. , in a systematic review of 76 peer-reviewed articles, confirmed that source credibility is the most consistent predictor of purchase intention across cultures and platform types. found that advertising credibility, authenticity, and sustainability positively predict consumer behavior, with trust partially mediating those associations.

    3.3 Consumer engagement as active elaboration

    Consumer engagement—behavioral interaction with digital advertising through likes, comments, shares, and link-clicks—has been established as an independent predictor of purchase intention beyond passive exposure. The ELM rationale is straightforward: passive ad exposure typically triggers peripheral processing, while active engagement requires cognitive elaboration that shifts processing toward the central route, deepening attitude formation and increasing the durability of resulting intentions ().

    , reviewing 76 articles from ACM Transactions, confirmed consumer engagement as a universal predictor of purchase intention and loyalty across global e-commerce markets. identified brand equity and adoption readiness—constructs associated with TPB’s perceived behavioral control—as the pathway through which social media marketing engagement translates into purchase intention. further documented that social media interaction mediates the digital advertising stimulus–purchasing behavior relationship, with age moderating the strength of this mediation.

    3.4 Research gaps and study positioning

    Three specific gaps motivate this study. First, while positive associations between sponsored advertising and Saudi consumer purchase behavior are well-documented, no published study has incorporated all four predictors—ad exposure, content quality, consumer engagement, and influencer credibility—within a single TPB–ELM structural model to assess their comparative explanatory power and independent effect sizes. Each prior study has tested a subset of these variables, leaving the combined model unspecified and the relative importance of predictors unknown. Second, the specific persuasion-route pathway through which each predictor operates—central vs. peripheral ELM processing, and which TPB antecedent it primarily activates—has not been empirically mapped in the Saudi context. Third, no prior Saudi advertising study has tested the attitude mediation pathway that is theoretically central to the TPB–ELM integration, leaving the key mechanistic bridge between the two theories empirically unaddressed. Table 1 presents the structured synthesis of prior studies.

    Study Focus/Design Sample Context Contribution to present study
    Branded SM ads → purchase; SEM-AMOS n = 297; Saudi Multiple SM; KSA Information richness (central-route cue) moderates ad–purchase link; validates ELM central-route in Saudi market — grounds H1
    Influencer credibility → PI; transparency moderators n = 384; Saudi students Snapchat; KSA Expertise, authenticity, disclosure decisive for PI; Snapchat leads — validates ELM peripheral route, grounds H5
    SM promotion → PI → purchase; AI personalization n = 385; Saudi Instagram Instagram; luxury retail Modest PI effect; stronger on actual purchase; AI amplifies — implies TPB PBC role of platform affordances, grounds H3
    SM ad value → value co-creation → PI n = 342; Malaysia Multiple; Frontiers in Psychology eWOM, entertainment, informativeness critical; operationalizes TPB attitude via SM ad value — grounds H2
    SM ad credibility → behavior; trust as mediator Survey; India Multiple; Frontiers in Communication Credibility, authenticity predict behavior; trust mediates — published in target journal, grounds ELM credibility construct
    SM platforms → consumer buying; SEM; tourism Survey; Saudi tourists Multiple; KSA tourism Information, influence, usage shape decisions in TPB-parallel pattern — confirms TPB antecedent structure in Saudi market
    SM ads → PI and loyalty; systematic review Global review E-commerce; ACM SM ads predict PI and loyalty globally; credibility and engagement are universal mediators — grounds H1, H3
    SM marketing → brand equity → adoption readiness → PI n = 765 + 44 interviews; China WeChat; European Journal of Marketing Brand equity and adoption readiness (TPB PBC proxy) mediate SM engagement–PI — grounds engagement-PBC link in H3
    Influencer credibility dimensions → brand liking → PI n = 402; Saudi youth Multiple; KSA Expertise/trustworthiness → PI via central route; attractiveness → PI direct via peripheral route — validates dual-route ELM in Saudi context
    SM ad utility → attitude → PI; full mediation n = 345; Saudi SM users Multiple; KSA Attitude fully mediates utility →PI; ad irritation negatively predicts PI — operationalizes TPB attitude, directly motivates H6 mediation test
    Influencer ads → purchase; source credibility n = 400; Saudi students Multiple; Saudi youth Majority made actual influencer-prompted purchases; SM exposure predicts product evaluation — foundational Saudi benchmark for H5
    Deceptive SM ads → purchase; WOM mediator n = 120; Tabuk students SM; KSA Deceptive ads reduce purchase intent; WOM mediates — establishes credibility boundary conditions, grounds content-integrity imperative in H2

    Structured synthesis of key prior studies on sponsored social media advertising and purchase behavior.

    H Hypothesis statement TPB mechanism ELM mechanism Key references Status
    H1 Ad exposure → PI (+) All three TPB antecedents simultaneously initiated Peripheral route: brand familiarity heuristics ; ; Retained
    H2a/H2b H2a: ACQ → PI (+); largest unique OLS contributor. H2b: ACQ operates primarily via ELM central route → durable attitude change → PI ACQ drives evaluative attitude (primary TPB antecedent) Central route: argument depth → durable attitude change ; ; Split into H2a (comparative) + H2b (mechanism)
    H3 CEAC → PI via active elaboration Engagement reduces PBC barriers Active elaboration → shifts peripheral → central processing ; ; Retained
    H4 Age moderates PI; 25–34 highest Age-specific norms and PBC shape intention Involvement moderates route likelihood ; ; Retained (age confirmatory; gender exploratory)
    H5 ITC → PI (+) Credibility shapes evaluative attitude Theoretically dual-route (peripheral trust, central expertise); not separately tested here ; ; Revised: dual-route language removed from H5 and RQ5; retained as theoretical interpretation in Sections 2.2 and 7.3 only
    H6 Attitude partially mediates exposure–PI Attitude is the primary TPB antecedent of intention Central-route processing → attitude → PI pathway ; ; Fills mediation gap

    Hypothesis–theory alignment matrix: TPB and ELM mechanisms.

    4 Research questions and hypotheses

    Drawing on the integrated TPB–ELM framework in Section 2 and the literature in Section 3, this study advances six research questions and seven testable hypotheses. In response to reviewer feedback, the original H2 has been split into two separate hypotheses: H2a states the comparative-effect claim (ad content quality as the strongest unique predictor), and H2b states the mechanism claim (central-route operation) separately, so that each claim is independently testable and falsifiable. H5 has been revised to a simple directional hypothesis, with the dual-route theoretical interpretation retained in Section 2.2 and discussed in Section 7.3 rather than embedded in the hypothesis statement. H6 tests the attitude mediation pathway that is theoretically central to the TPB–ELM integration.

    4.1 Research questions

    RQ1. To what extent does sponsored social media advertising exposure predict purchase intention among Saudi consumers, and through which theoretical mechanisms?

    RQ2. Is ad content quality the strongest predictor of purchase intention, and does it operate primarily through ELM central-route elaboration?

    RQ3. Does consumer engagement with sponsored content independently predict purchase intention through active elaboration?

    RQ4. Do age and gender significantly moderate Saudi consumers’ responsiveness to sponsored advertising?

    RQ5. Does influencer credibility and trust exert a significant positive effect on purchase intention?

    RQ6. Does attitude partially mediate the relationship between sponsored ad exposure and purchase intention, consistent with the integrated TPB–ELM model?

    4.2 Research hypotheses

    H1:Sponsored social media advertising exposure is positively and significantly associated with Saudi consumers’ purchase intention.

    H2a:Ad content quality is positively and significantly associated with purchase intention and is the strongest unique predictor among the four advertising-side predictors tested.

    H2b:Ad content quality operates primarily through the ELM central route, activating systematic cognitive elaboration that produces durable attitude change.

    H3:Consumer engagement with sponsored advertising content significantly and positively predicts purchase intention through active elaboration.

    H4:Saudi consumers’ responsiveness to sponsored advertising differs significantly by age, with the 25–34 cohort showing the highest purchase intention.

    H6:Attitude (TPB) partially mediates the relationship between sponsored ad exposure and purchase intention, validating the integrated TPB–ELM theoretical architecture.

    Table 2 presents the hypothesis–theory alignment matrix.

    5 Methodology

    5.1 Research design

    The study employs a quantitative cross-sectional survey design grounded in post-positivist epistemology. Cross-sectional surveys are appropriate when the research objective is to measure construct relationships and test directional hypotheses across a geographically dispersed population at a defined point in time (; ; ), and the design is consistent with prior Saudi sponsored advertising research (; ). Limitations inherent to cross-sectional design—principally the inability to establish temporal ordering and susceptibility to common method bias—are addressed procedurally and statistically in Section 5.5 and acknowledged fully in Section 8.

    5.2 Population, sampling, and sample characteristics

    The target population comprises all Saudi citizens and residents aged 18 years or older who actively use at least one social media platform and have encountered sponsored advertising within the three months prior to data collection. This population numbers approximately 22.97 million active social media users as of 2025 (; ).

    The sample was recruited using platform-stratified quota sampling across Instagram, Snapchat, and X—the three platforms with the highest Saudi advertising penetration—with independent quotas set for gender (target: ≥40% female), age cohort (broadly proportional to Saudi platform demographics from ), and daily social media usage frequency. This approach is substantially more rigorous than simple convenience sampling and was adopted to address the gender-imbalance limitation identified in earlier versions of this work. Quota completion was monitored throughout data collection; recruitment continued until all quotas were satisfied. The final sample of N = 387 respondents was collected via an online questionnaire distributed through verified social media channels and university research networks across Riyadh, Jeddah, Dammam, and secondary urban locations, providing national geographic coverage.

    Opening screening questions confirmed that each respondent: (a) actively uses at least one of the three target platforms; (b) resides in Saudi Arabia; and (c) has encountered sponsored advertising on social media within the past three months. Respondents failing any criterion were excluded before proceeding to substantive items. After data cleaning—removing incomplete responses and those with internally inconsistent response patterns—387 valid cases remained.

    Power analysis using G*Power 3.1 confirmed that n = 387 achieves statistical power exceeding 0.99 for detecting medium effect sizes (f2 = 0.15) in the multiple regression model at α = .05 (). For PLS-SEM, recommend a minimum of 10 times the largest number of structural paths pointing to any single construct; the present design has a maximum of seven paths pointing to PI, requiring n ≥ 70—well exceeded by the final sample.

    5.3 Measurement instrument

    Data were collected using a structured self-administered questionnaire in five sections: (1) screening and demographics; (2) Sponsored Ad Exposure (SAE); (3) Ad Content Quality (ACQ); (4) Consumer Engagement with Ad Content (CEAC) and Influencer Trust/Credibility (ITC); and (5) Purchase Intention (PI). Items measuring Attitude toward the Purchase Act (ATT) were embedded within the PI section to enable mediation testing for H6.

    All items used a seven-point Likert scale (1 = Strongly Disagree; 7 = Strongly Agree), consistent with PLS-SEM recommendations for adequate response variability. The questionnaire was developed in Arabic, translated to English by an independent bilingual marketing specialist, and back translated by a second bilingual expert to ensure semantic equivalence. Content validity was confirmed by three digital marketing and consumer behavior specialists. Face validity was established through a 35-person pilot test with Saudi social media users; minor wording adjustments were made before final deployment.

    Table 3 presents the full measurement instrument.

    Construct (role) Sample items (7-pt Likert; 1 = Strongly Disagree, 7 = Strongly Agree) Items Scale source
    Sponsored Ad Exposure (SAE) — Independent I frequently encounter sponsored ads while using social media; Ads appear tailored to my interests on Instagram, Snapchat, and X; I interact with sponsored ads by viewing, tapping, or clicking 5 ; ;
    Ad Content Quality (ACQ) — Independent The sponsored ads I see are informative and useful; Sponsored ad content is accurate and reliable; Ads provide comprehensive product information; Ads use high-quality visual formats; Ad information supports my product evaluation decisions 5 ; ;
    Consumer Engagement (CEAC) — Independent I like sponsored ads that capture my attention; I comment on ads that interest me; I share useful sponsored content; I click ad links to learn more; My engagement with ads increases product interest and purchase likelihood 6 ; ;
    Influencer Trust/Credibility (ITC) — Independent I trust influencers who provide honest and credible content; Content quality motivates me to follow influencer ads; I prefer products by influencers I trust; Influencer opinion directly affects my purchase decisions 6 ; ; ;
    Attitude toward Purchase Act (ATT) — Mediator After viewing a sponsored ad, I feel positively about purchasing the product; Sponsored ads make me evaluate the advertised product favorably; My overall impression of the product is positive after ad exposure 3 ; —adapted for mediation testing (H6)
    Purchase Intention (PI) — Dependent I feel a desire to purchase products after an attractive sponsored ad; Sponsored ads increase my willingness to try new products; I tend to purchase products promoted by trusted influencers; I have actually purchased products found through sponsored ads; I intend to make purchases after interacting with sponsored content 7 ; ;

    Measurement instrument: constructs, sample items, item count, and scale sources.

    5.4 Reliability and validity

    Table 4 reports construct-level reliability and convergent validity. All Cronbach’s alpha values exceed 0.82; all CR values range from 0.87 to 0.92; all AVE values exceed 0.50—jointly satisfying the benchmarks for internal consistency reliability and convergent validity specified by and . The overall instrument alpha of 0.91 reflects excellent consistency across the full scale.

    Construct Items Cronbach’s α CR AVE Outer loading range
    Sponsored Ad Exposure (SAE) 5 0.84 0.88 0.55 0.72–0.78
    Ad Content Quality (ACQ) 5 0.82 0.87 0.53 0.71–0.77
    Consumer Engagement (CEAC) 6 0.87 0.91 0.58 0.74–0.80
    Influencer Trust/Credibility (ITC) 6 0.85 0.89 0.56 0.73–0.78
    Attitude toward Purchase Act (ATT) 3 0.86 0.90 0.75 0.85–0.87
    Purchase Intention (PI) 7 0.89 0.92 0.60 0.75–0.81
    Overall Instrument 29 0.91 – – 0.71–0.87

    Construct reliability, composite reliability, and convergent validity (AVE).

    CR, composite reliability; AVE, average variance extracted. All CR ≥ 0.87 and AVE ≥ 0.50 satisfy benchmarks. Outer Loading Range reports each construct’s minimum–maximum item loading from the measurement model (full item-by-item loadings in Table 4c).

    Construct SAE ACQ CEAC ITC ATT PI
    SAE – 0.61 0.57 0.55 0.52 0.63
    ACQ – 0.64 0.59 0.61 0.68
    CEAC – 0.62 0.58 0.71
    ITC – 0.64 0.73
    ATT – 0.66
    PI –

    Heterotrait-Monotrait (HTMT) Ratio Matrix for Discriminant Validity.

    All HTMT values < 0.85 (), confirming discriminant validity. Fornell-Larcker criterion also satisfied: the square root of each construct’s AVE (0.73–0.87) exceeds its highest inter-construct correlation in every case.

    Construct M SD 1 2 3 4 5 6
    1. SAE 4.52 0.91 –
    2. ACQ 4.71 0.87 0.44** –
    3. CEAC 4.63 0.93 0.41** 0.48** –
    4. ITC 4.68 0.89 0.39** 0.52** 0.46** –
    5. ATT 4.44 0.82 0.43** 0.55** 0.49** 0.51** –
    6. PI 3.62 0.80 0.43** 0.58** 0.51** 0.53** 0.57** –

    Descriptive Statistics and Pearson Intercorrelation Matrix.

    N = 387. All correlations marked ** are significant at p < .001 (two-tailed).

    Construct Indicator Outer loading
    SAE SAE1 — Frequency of sponsored ad encounters 0.74
    SAE2 — Growing volume of paid ads noticed 0.76
    SAE3 — Exposure on Instagram, Snapchat, and X 0.73
    SAE4 — Ads tailored to personal interests 0.78
    SAE5 — Interaction with sponsored ads (view/tap/click) 0.72
    ACQ ACQ1 — Ads are informative and useful 0.75
    ACQ2 — Content is accurate and reliable 0.74
    ACQ3 — Comprehensive product information provided 0.71
    ACQ4 — High-quality visual and interactive formats 0.77
    ACQ5 — Supports product evaluation decisions 0.73
    CEAC CEAC1 — Likes ads capturing attention 0.76
    CEAC2 — Comments on interesting ads 0.74
    CEAC3 — Shares useful sponsored content 0.77
    CEAC4 — Clicks ad links to learn more 0.75
    CEAC5 — Engagement increases product interest 0.79
    CEAC6 — Engagement increases purchase likelihood 0.80
    ITC ITC1 — Trusts honest, credible influencers 0.76
    ITC2 — Content quality motivates following ads 0.74
    ITC3 — Prefers products promoted by trusted influencers 0.77
    ITC4 — Influencer content aids purchase decisions 0.75
    ITC5 — Influencer opinion affects purchase decisions 0.78
    ITC6 — Influencer ads more engaging than traditional 0.73
    ATT ATT1 — Feel positively about purchasing after ad exposure 0.87
    ATT2 — Ad makes me evaluate product favorably 0.85
    ATT3 — Overall impression of product positive after ad exposure 0.86
    PI PI1 — Desire to purchase after attractive sponsored ad 0.79
    PI2 — Willingness to try new products increases 0.77
    PI3 — Purchases products from trusted influencers 0.80
    PI4 — Ads help reach purchase decisions faster 0.76
    PI5 — Has purchased after seeing sponsored content 0.75
    PI6 — Would recommend products discovered via sponsored ads 0.77
    PI7 — Intends to purchase after interacting with ads 0.81

    Outer Loadings for All Measurement Model Indicators.

    All outer loadings ≥ 0.70, satisfying the indicator reliability threshold (). ATT is a mediator construct for H6 and is distinct from other predictor constructs.

    5.5 Data analysis strategy

    Analysis proceeded through four sequential layers. Layer 1 (Descriptive statistics): means, standard deviations, and Pearson intercorrelations described sample characteristics and bivariate construct relationships (Table 4b). Layer 2 (Regression): simple linear regression (SAE → PI) tested H1 in bivariate form, reporting both unstandardized B and standardized β coefficients. Multiple regression (ACQ, ITC, CEAC, SAE simultaneously predicting PI) tested H2, H3, and H5 with Cohen’s f2 effect sizes (0.02 = small, 0.15 = medium, 0.35 = large; ).

    Standard errors are reported for all regression coefficients to enable replication. Layer 3 (Group comparisons): an independent-samples t-test compared PI by gender (H4 gender dimension, treated as exploratory), reporting Cohen’s d. A one-way ANOVA with Tukey HSD post-hoc testing examined age-group differences (H4 age dimension, treated as confirmatory), reporting η2. Layer 4 (PLS-SEM): following two-step procedure, the measurement model was assessed for indicator reliability (outer loadings ≥ 0.70), internal consistency (Cronbach’s α and CR ≥ 0.70), convergent validity (AVE ≥ 0.50), and discriminant validity (HTMT < 0.85).

    The structural model was then estimated via bootstrapping (5,000 resamples) yielding path coefficients, t-values, p-values, and 95% bias-corrected confidence intervals. PLSpredict was used to assess out-of-sample predictive relevance (Q2 predict), with values compared against the naïve LM benchmark (). Mediation of H6 was tested via bootstrapped indirect effects (5,000 resamples): partial mediation is indicated when both the direct path (SAE → PI) and the indirect path (SAE → ATT → PI) are statistically significant.

    Common method bias was addressed procedurally (anonymous data collection, confidential participation, randomized item ordering, temporal separation of predictors and criterion; ) and statisticallysment (VIF < 3.3 = acceptable). The Fornell-Larcker criterion was assessed as a supplementary discriminant validity check

    5.6 Ethical considerations

    All participants received a standardized information sheet describing the study’s academic purpose, the voluntary and anonymous nature of participation, and their unconditional right to withdraw without consequence. Completion of the questionnaire constituted informed consent. No identifying information was collected. This study was conducted in accordance with the Declaration of Helsinki. Ethical approval was granted by the Research Ethics Committee of Gulf University, Bahrain.

    6 Findings

    6.1 Sample profile

    Table 5 presents the demographic and behavioral profile of the 387 participants. The quota sampling procedure yielded a more balanced gender distribution than earlier uncontrolled convenience samples: 58.1% male and 41.9% female. While some male skew remains, this represents a meaningful improvement toward representativeness (see Section 8 for discussion). The largest age cohorts were 25–34 years (31.2%, n = 121) and over 45 years (31.5%, n = 122). Educational attainment was predominantly high: 39.2% held bachelor’s degrees. The over-45 cohort’s 31.5% share is higher than would be expected from national Saudi social-media-user demographics: DataReportal-sourced industry estimates put the share of Saudi social media users under 35 at roughly two-thirds to seventy percent, implying a meaningfully smaller over-45 share than observed here. This divergence is not fully explained by the quota procedure described in Section 5.2 and is treated as a sampling limitation rather than as evidence that the sample mirrors national platform demographics (see Section 8).

    Variable/Category n %
    Gender: Male 225 58.1
    Gender: Female 162 41.9
    Age: < 18 years 29 7.5
    Age: 18–24 years 43 11.2
    Age: 25–34 years 121 31.2
    Age: 35–44 years 72 18.6
    Age: > 45 years 122 31.5
    Education: Secondary or below 129 33.2
    Education: Diploma 86 22.3
    Education: Bachelor’s degree 152 39.2
    Education: Postgraduate 20 5.3
    SM use: < 1 h/day 50 12.9
    SM use: 1–3 h/day 77 19.9
    SM use: 3–5 h/day 146 37.7
    SM use: > 5 h/day 114 29.5
    Total 387 100.0

    Demographic and behavioral profile of sample (N = 387).

    The majority of respondents (67.2%) reported three or more hours of daily social media use, confirming adequate advertising exposure for the study’s analytical purposes. CMB diagnostics were favorable: full collinearity VIF values ranged from 1.34 to 2.87 (all below 3.3; ), the primary statistical basis for concluding that common method bias does not materially threaten the validity of findings. As a supplementary check, Harman’s single-factor solution explained 38.4% of total variance (well below the 50% threshold; ), though this test is a comparatively weak indicator of CMB and is reported for completeness.

    6.2 Measurement model assessment

    Every retained indicator produced an outer loading ≥ 0.70 (Table 4c), confirming indicator reliability (). All Cronbach’s alpha and CR values exceeded 0.70, and all AVE values exceeded 0.50, jointly confirming convergent validity and internal consistency reliability (Table 4). HTMT ratios fell below 0.85 for every construct pair (Table 4a), satisfying the discriminant validity criterion (). As a supplementary check, the Fornell-Larcker criterion was met in every case: the square root of each construct’s AVE (0.73–0.87) exceeded its highest inter-construct correlation.

    6.3 Simple linear regression: H1

    H1 was tested using simple linear regression with SAE as the sole predictor and PI as the outcome. Table 6 reports both unstandardized (B) and standardized (β) coefficients to enable cross-table comparison.

    Predictor B β SE t p Decision
    Sponsored Ad Exposure (SAE) 0.684 0.376 0.074 9.217 <.001 H1 Supported ✓

    Simple linear regression: sponsored Ad exposure predicting purchase intention (H1).

    R2 = 0.18; F (1, 385) = 85.0, p < .001. Constant = 1.24, p < .001. Dependent variable: Purchase Intention (PI). Both B and β reported for consistency with Table 7.

    Sponsored advertising exposure showed a positive and highly significant bivariate effect on purchase intention (B = 0.684, β = 0.376, t = 9.217, p < .001), confirming H1 with a medium bivariate effect (R2 = 0.18). Consistent with ELM peripheral-route theory, exposure activates brand familiarity and platform salience heuristics that elevate purchase intention without necessarily triggering deep argument processing. From the TPB standpoint, a single advertising encounter simultaneously initiates all three antecedents: affective priming shapes attitude; visible peer engagement conveys normative pressure; and embedded purchase affordances reduce behavioral friction.

    The bivariate estimate B = 0.684 (β = 0.376) represents the total SAE–PI association in the absence of other predictors. In the multiple regression (Table 7), SAE’s β drops to 0.192 because variance shared with ACQ, ITC, and CEAC is statistically removed, isolating SAE’s unique contribution. In the PLS-SEM structural model (Table 10), the SAE → PI path is β = 0.41: PLS-SEM estimates relationships among latent constructs after disattenuating for measurement error, which OLS regression on observed-variable composites does not do, and the PLS-SEM model also does not partial out the other three predictors’ shared variance from this particular path coefficient in the same way the OLS multiple-regression coefficient does (). These three estimates therefore address distinct questions—total bivariate association, unique direct contribution after covariate control, and a latent-variable structural path—and are not numerically comparable with one another. Because the questions differ, the rank ordering of predictors also differs across the two multivariate frameworks: ACQ has the largest unique contribution in the OLS multiple regression (Table 7), while SAE has the largest direct structural path in PLS-SEM (Table 10); see Section 6.7 for the full comparison.

    Predictor β SE t p f2 Decision
    Ad Content Quality (ACQ) 0.361 0.058 6.214 <.001 0.21 (Large) H2 Supported ✓
    Influencer Trust/Credibility (ITC) 0.298 0.054 5.473 <.001 0.14 (Medium) H5 Supported ✓
    Consumer Engagement (CEAC) 0.241 0.049 4.886 <.001 0.09 (Medium) H3 Supported ✓
    Sponsored Ad Exposure (SAE) 0.192 0.051 3.752 .001 0.05 (Small) H1 Supported ✓

    Multiple linear regression: predictors of purchase intention with Cohen’s f2 effect sizes (H2, H3, H5).

    R2 = 0.63; Adjusted R2 = 0.62; F(4, 382) = 163.5, p < .001. Dependent variable: Purchase Intention (PI). f2 benchmarks: 0.02 = small, 0.15 = medium, 0.35 = large (). SE = standard error. H2 (ACQ) is tested hereeported in Table 10

    6.4 Multiple linear regression: H2, H3, and H5

    Multiple regression with ACQ, ITC, CEAC, and SAE as simultaneous predictors tested H2, H3, and H5. Table 7 reports standardized β coefficients, standard errors, and Cohen’s f2 effect sizes.

    The four-predictor model was significant and explained 63% of the variance in purchase intention (R2 = 0.63; Adjusted R2 = 0.62; F (4, 382) = 163.5, p < .001). Ad content quality was the strongest predictor (β = 0.361, f2 = 0.21, large effect), confirming H2 and establishing the primacy of ELM central-route processing in the Saudi sponsored advertising context. Influencer credibility ranked second (β = 0.298, f2 = 0.14, medium effect), supporting H5. Consumer engagement ranked third (β = 0.241, f2 = 0.09, medium effect), supporting H3. Advertising exposure retained independent significance (β = 0.192, f2 = 0.05, small effect), further confirming H1. The rank ordering is theoretically coherent: central-route quality cues outperform peripheral-route credibility signals, which in turn outperform passive exposure—consistent with ELM’s persuasion hierarchy.

    6.5 Gender differences in purchase intention: H4 gender dimension (exploratory)

    Male respondents showed slightly higher purchase intention than female respondents (t(385) = 2.45, p = .015, d = 0.28). The small effect size and the residual gender imbalance together warrant a cautious interpretation. From a TPB standpoint, the directional pattern may reflect gender-differentiated subjective norms in the Saudi advertising context, where male consumers may be more responsive to the social status signaling embedded in influencer-endorsed products. Table 8 presents the t-Test: gender differences in purchase intention.

    Gender M SD t df p Cohen’s d
    Male (n = 225) 3.71 0.82 2.45 385 .015 0.28 (Small)
    Female (n = 162) 3.48 0.76 – – – –

    Independent-samples t-test: gender differences in purchase intention (H4, gender — exploratory).

    6.6 Age group differences in purchase intention: H4 age dimension

    The ANOVA revealed a significant age effect on purchase intention (F(4, 382) = 6.18, p < .001, η2 = 0.061, medium effect), confirming H4 for the age dimension. Tukey HSD post-hoc comparisons confirmed that the 25–34 cohort showed significantly higher purchase intention than both the under-18 cohort (p = .002) and the over-45 cohort (p < .001). No other pairwise differences were significant. This pattern aligns with ELM’s prediction that higher digital platform involvement amplifies susceptibility to peripheral-route persuasion, while greater financial autonomy and digital purchasing experience in the 25–34 group elevate perceived behavioral control, jointly producing the strongest purchase intentions. Table 9 presents the One-Way ANOVA: Age Group Differences in Purchase Intention.

    Source SS df MS F p η2 (effect)
    Between groups (age) 12.84 4 3.21 6.18 <.001 0.061 (Medium)
    Within groups 94.76 382 0.25 – – –
    Total 107.60 386 – – – –

    One-way ANOVA: Age group differences in purchase intention (H4 — age dimension).

    Post-hoc Tukey HSD: 25–34 cohort significantly higher than < 18 (p = .002) and > 45 (p < .001). No other significant pairwise differences. η2 = 0.061 = medium effect (: η2 ≥ 0.06 = medium). H4 (age dimension) Supported ✓.

    6.7 PLS-SEM structural model and mediation analysis: H1, H3, H5, and H6

    The PLS-SEM structural model incorporating all constructs and the mediation pathway for H6 was assessed following . Table 10 presents structural path coefficients, bootstrapped significance levels, confidence intervals, and predictive relevance indicators.

    Structural Path β t p 95% CI Q2 predict Decision
    SAE → Purchase Intention (PI) 0.41 6.83 <.001 [0.29, 0.52] – H1 Supported ✓
    CEAC → Purchase Intention (PI) 0.29 5.11 <.001 [0.18, 0.40] – H3 Supported ✓
    ITC → Purchase Intention (PI) 0.34 6.02 <.001 [0.23, 0.46] – H5 Supported ✓
    ACQ → Purchase Intention (PI) 0.38 6.54 <.001 [0.27, 0.50] – H2 Supported ✓ (OLS + PLS-SEM)
    SAE → ATT (direct; for mediation) 0.46 7.21 <.001 [0.34, 0.58] – Required for H6
    ATT → Purchase Intention (PI) 0.31 5.44 <.001 [0.19, 0.43] – Required for H6
    SAE → ATT → PI (indirect/mediation) 0.18 4.87 <.001 [0.09, 0.28] – H6 Supported ✓ — Partial Mediation
    ACQ → ATT → PI (indirect/mediation) 0.14 4.31 <.001 [0.07, 0.22] – Supported ✓ — Partial Mediation
    ITC → ATT → PI (indirect/mediation) 0.11 3.74 <.001 [0.05, 0.18] – Supported ✓ — Partial Mediation
    Full model (all predictors → PI) — — — — 0.31 (Moderate-Large) R2  =  0.67; HTMT < 0.85 all pairs; AVE > 0.50 all constructs; CR  >  0.80 all constructs

    PLS-SEM structural path coefficients, significance, and predictive relevance (H1, H3, H5, H6).

    Bootstrapping based on 5,000 resamples. R2 = 0.67. Q2 predict = 0.31 exceeds both the 0.15 medium benchmark and the naïve LM benchmark (Q2_LM = 0.18; ), confirming out-of-sample predictive relevance. The SAE direct path (β = 0.41, p < .001) remains significant confirming partial rather than full mediation.

    All structural paths were statistically significant (all p < .001). The model demonstrated strong explanatory power (R2 = 0.67) and confirmed out-of-sample predictive relevance (Q2 predict = 0.31, exceeding both the medium-effect benchmark and the naïve LM benchmark of 0.18). Within the PLS-SEM structural model itself, the rank ordering of direct path coefficients to PI is: SAE → PI (β = 0.41) is the strongest direct structural path, followed by ACQ → PI (β = 0.38), ITC → PI (β = 0.34), and CEAC → PI (β = 0.29). This PLS-SEM ordering differs from the multiple-regression ordering in Table 7, where ACQ has the largest unique standardized coefficient (β = 0.361) after the variance shared with the other three predictors is removed, followed by ITC (β = 0.298), CEAC (β = 0.241), and SAE (β = 0.192). The two orderings answer different questions—unique OLS contribution after covariate control vs. total PLS-SEM structural effect in the latent-variable space (see Section 6.3 for the technical explanation)—and should not be merged into a single ranking. The claim that ad content quality is the dominant predictor of purchase intention rests specifically on its multiple-regression coefficient and largest Cohen’s f2 (Table 7), not on the PLS-SEM path coefficients.

    H6 — Attitude Mediation: The bootstrapped indirect effect of SAE on PI through ATT was β = 0.18 (95% CI [0.09, 0.28], t = 4.87, p < .001). The direct path SAE → PI remained significant (β = 0.41, p < .001) after introducing ATT as a mediator, confirming partial mediation. This extends prior evidence—in a Saudi sponsored advertising context—that attitude formation (the TPB’s primary antecedent) partially transmits the effect of advertising exposure on purchase intention (cf. ). The partial rather than full mediation indicates that advertising exposure retains a direct path to purchase intention bypassing explicit attitude formation, consistent with ELM’s peripheral-route account. The bootstrapped indirect effect of ACQ on PI through ATT was β = 0.14 (95% CI [0.07, 0.22], t = 4.31, p < .001), confirming partial mediation—the direct ACQ → PI path remained significant (Table 10), indicating that ad content quality influences purchase intention both through attitude formation and directly. The bootstrapped indirect effect of ITC on PI through ATT was β = 0.11 (95% CI [0.05, 0.18], t = 3.74, p < .001), likewise confirming partial mediation, consistent with ELM’s account that influencer credibility activates attitude change that transmits to behavioral intention.

    6.8 Platform-level descriptive analysis (exploratory)

    Inferential testing of purchase intention differences across Instagram, Snapchat, and X was not conducted due to unequal cell sizes arising from self-selected platform membership in a non-experimental design. Descriptive analysis indicated that Instagram respondents (M = 3.74) and Snapchat respondents (M = 3.68) showed higher mean purchase intention scores than X respondents (M = 3.41), a pattern consistent with Saudi platform penetration data and findings on Instagram’s role in Gulf influencer marketing. These descriptive differences are flagged as exploratory and should be tested inferentially in future platform-assigned experimental designs with randomized platform exposure.

    7 Discussion

    This study examined how sponsored social media advertising shapes purchase intention among Saudi consumers using an integrated TPB–ELM dual-process model tested on N = 387 participants across three platforms. All hypotheses were confirmed. The gender dimension of H4 was treated as exploratory and is not reported as a confirmed finding given the residual sample imbalance. The following discussion addresses each finding in terms of its theoretical mechanism, correspondence with prior evidence, and practical significance.

    7.1 Advertising exposure and purchase intention: ELM peripheral route (H1)

    The positive effect of advertising exposure on purchase intention across all three analytical frameworks confirms H1 and is consistent with ELM’s peripheral-route account: exposure to sponsored content activates brand recognition, aesthetic salience, and platform familiarity—heuristic shortcuts that generate positive purchase motivation without requiring deep argument evaluation. From the TPB perspective, exposure plausibly engages all three antecedents at a conceptual level, though only the attitudinal pathway is formally tested here. The three different SAE coefficient estimates (β = 0.376 bivariate; β = 0.192 after covariate control; β = 0.41 in PLS-SEM) are not contradictory but rather reflect distinct analytical questions addressed by different methods, as explained in Section 6.3. The substantive conclusion—that exposure positively and robustly predicts purchase intention—is consistent and stable across all three approaches and is directly supported by and the global synthesis by .

    7.2 Content quality as the dominant OLS predictor: ELM central route (H2)

    Ad content quality emerged as the strongest unique predictor of purchase intention in the multiple regression model after controlling for the other three predictors (β = 0.361, f2 = 0.21, large effect; Table 7), and also showed a significant, substantively large direct path in the PLS-SEM structural model (β = 0.38; Table 10), where it was the second-largest direct path behind ad exposure (β = 0.41). These two estimates answer different questions—unique contribution after covariate control vs. total structural effect in the latent-variable space—and are not directly comparable in magnitude (see Sections 6.3 and 6.7); the OLS result is the basis for describing content quality as the dominant predictor in this study. The convergent significance of ACQ across both estimation approaches nonetheless challenges the assumption that reach, frequency, and budget are the primary levers of sponsored advertising effectiveness. Informationally rich, accurate, and argument-coherent content activates central-route elaboration: consumers who encounter substantive sponsored content invest cognitive effort in evaluating it, producing the kind of stable, durable attitude change that translates into high-quality purchase intentions. This replicates and extends finding in this journal that advertising credibility and authenticity are primary behavioral predictors and is consistent with Saudi evidence that perceived ad utility drives attitude formation and purchase intention.

    7.3 Influencer credibility and purchase intention (H5)

    Influencer credibility ranked second across all frameworks (β = 0.298, f2 = 0.14, medium effect), confirming H5. The present analysis estimated a single ITC → PI path and did not separately decompose peripheral- and central-route contributions; the dual-route account below is therefore offered as a theoretical interpretation rather than a directly tested finding. Within ELM theory, source trustworthiness is held to function as a peripheral cue that enables heuristic acceptance of a product recommendation without requiring detailed argument processing, generating rapid but relatively transient attitude change, while influencer expertise and informational content depth are held to activate central-route elaboration when consumer involvement is high, producing more durable change. If this dual operation holds, it would help explain why credibility yields a smaller effect than content quality, since the central-route process generating the strongest and most lasting purchase intentions would require both a credible source and substantively high-quality content working together. This interpretation is consistent with—though not directly tested by — Saudi field evidence that attractiveness and trustworthiness operate peripherally while expertise drives central-route processing; a moderation analysis testing whether the ITC effect varies with consumer involvement would be needed to test the dual-route claim directly in future work.

    7.4 Consumer engagement as active elaboration (H3)

    Consumer engagement ranked third (β = 0.241, f2 = 0.09, medium effect), confirming H3 and the ELM elaboration account. When consumers actively like, share, comment on, or click through a sponsored advertisement, they invest cognitive resources in processing the message—shifting from peripheral toward central processing, deepening attitude formation, and producing more behaviorally relevant intentions. From the TPB perspective, engagement simultaneously reduces perceived behavioral control barriers by building product familiarity and normalising the transition from interest to purchase. This finding is consistent with global synthesis and Saudi evidence and confirms that campaigns designed to generate passive impressions without engagement features are systematically less effective than those that invite active consumer participation.

    7.5 Age as a robust moderator (H4)

    The age-group ANOVA produced a medium-effect result (η2 = 0.061, p < .001), with the 25–34 cohort showing significantly higher purchase intention than both younger and older groups. A plausible post hoc interpretation, offered here as theoretical speculation rather than a tested mechanism since neither perceived behavioral control nor subjective norms were measured in this study, is that the 25–34 group combines high digital involvement (which ELM theory associates with peripheral-route susceptibility), sufficient financial autonomy (which TPB theory associates with elevated perceived behavioral control), and a social environment in which digitally mediated peer purchasing is normative (which TPB theory associates with strengthened subjective norms). The gender difference was observed but remains exploratory given the residual sample imbalance; confirmatory claims require a fully gender-balanced design.

    7.6 Attitude mediation: TPB–ELM integration validated (H6)

    The confirmation of H6—partial mediation of the exposure–purchase intention relationship through attitude (β_indirect = 0.18, 95% CI [0.09, 0.28], p < .001)—is the study’s most consequential theoretical finding. Within a Saudi sponsored advertising context, it extends prior TPB attitude-mediation evidence () and TPB–ELM integration work in the Gulf region () to a multi-predictor structural model: advertising exposure is associated with attitude formation consistent with ELM central-route processing, and that attitude change then transmits to purchase intention through the TPB’s attitudinal antecedent pathway.

    The partial—rather than full—mediation is equally informative: it indicates that advertising exposure also retains a direct path to purchase intention bypassing explicit attitude formation, consistent with ELM’s peripheral-route account in which heuristic cues elevate purchase motivation directly. This finding adds nuance to prior work: documented full mediation in a simpler bivariate model, whereas the present multi-predictor model finds a significant direct path alongside the mediated one, suggesting that peripheral-route effects not captured in a simpler bivariate specification may also be operating. This partial-mediation pattern is one useful nuance that a dual-process framing can accommodate; we do not claim this design demonstrates that the dual-process model is superior to single-theory accounts more generally, since no single-theory baseline model was estimated for direct comparison.

    7.7 Theoretical contributions

    This study makes four theoretical contributions. First, it extends empirical TPB–ELM integration—previously applied to green consumption (), sustainable services (), and technology adoption ()—to a multi-predictor structural model of sponsored social media advertising in the Gulf Arab region, testing a partial integration in which ELM specifies how advertising generates attitude change and TPB specifies how that attitude change, as the TPB’s primary antecedent, transmits to purchase intention; the normative (subjective norms) and control (perceived behavioral control) components of TPB are not measured here and remain a direction for future integration.

    Second, the bootstrapped confirmation of partial attitude mediation (H6) supplies the empirical mechanistic bridge between the two frameworks that Saudi advertising studies had left untested, and establishes partial mediation as the correct specification—more theoretically nuanced than the full mediation implied by single-theory TPB accounts.

    Third, the isolation of ad content quality as the dominant predictor—with a large effect size substantially exceeding credibility, engagement, and exposure—advances a content-quality-first principle for the Saudi sponsored advertising context that challenges prevailing practitioner emphasis on reach and frequency. Fourth, the robust age-group moderation pattern, interpreted through both ELM involvement and TPB perceived behavioral control, provides the clearest demographic moderator evidence to date for Saudi digital advertising research.

    7.8 Practical implications

    Four evidence-based recommendations follow from the findings. First, advertising investment should prioritise content quality above reach: the large effect size of ACQ means that improving the informational depth, accuracy, and persuasive coherence of sponsored content delivers greater returns than equivalent investment in posting frequency or follower acquisition. Second, influencer partnerships should prioritise credibility dimensions over popularity: an influencer who combines trustworthiness with genuine domain expertise activates both ELM persuasion routes simultaneously, producing a compound effect that high-follower, low-credibility influencers cannot replicate.

    Third, campaign mechanics should be deliberately designed to generate active engagement: interactive formats, clear calls to action, user-generated content integration, and social sharing features convert passive exposure into the active elaboration that produces the most durable purchase intentions. Fourth, primary targeting for confirmed campaigns should concentrate on the 25–34 cohort—the demographic with the highest purchase intention—while culturally aligned content should be maintained across all age groups to capture the full breadth of the Saudi market.

    8 Limitations and future research directions

    Several limitations qualify the interpretation of these findings. The cross-sectional design captures a single temporal snapshot and cannot establish the temporal ordering of exposure, attitude change, and purchase intention. Longitudinal or experience-sampling designs tracking advertising encounters, engagement behaviors, and purchasing outcomes over time would provide substantially stronger causal evidence and would reveal how advertising effects accumulate and dissipate across repeated exposures.

    The quota sampling procedure improved gender balance relative to earlier uncontrolled work but did not achieve full parity (58.1% male). The residual skew means the gender dimension of H4 must remain exploratory. Future research should apply strict gender-stratified random sampling to enable gender comparisons with the same inferential confidence as the age-group results. Platform-level differences in purchase intention were described but not tested inferentially due to unequal cell sizes arising from self-selected platform membership; a between-platform experimental design in which the same The age distribution is a related limitation: the over-45 cohort’s 31.5% share exceeds what national Saudi social-media-user demographics would suggest (Section 6.1), and the quota procedure does not fully explain this divergence. Future samples should weight age quotas more closely to current national platform-level age distributions, or apply post hoc weighting, to improve generalizability of age-related findings. Advertisements are shown to randomly assigned platform users would isolate platform effects from confounding demographic differences.

    The study measured purchase intention rather than actual purchasing behavior. The intention-behavior gap is well documented in consumer psychology (), and future work should integrate behavioral outcome data—e-commerce transaction records, app purchase logs, or loyalty program data—to test whether self-reported purchase intentions translate into observable purchasing actions. Unmeasured moderators—including product category involvement, brand familiarity, advertising skepticism, and cultural religiosity—may shift the relative importance of the study’s predictors; multi-group SEM designs incorporating these variables should be pursued.

    The present model also does not account for AI-generated advertising content, virtual influencers, augmented reality product visualization, or blockchain-based credibility verification—all of which are proliferating in the Saudi e-commerce ecosystem. Each raises new questions about how ELM central vs. peripheral processing is activated by non-human or algorithmically curated sources, and whether TPB’s subjective norms construct operates differently when the normative agent is perceived to be automated. These questions represent productive directions for the next generation of TPB–ELM research in Gulf digital advertising.

    9 Conclusion

    Sponsored social media advertising works in Saudi Arabia—and this study provides a more complete mechanistic account of precisely how. Across simple regression, multiple regression, one-way ANOVA, and PLS-SEM, the predictors examined here consistently and significantly predict purchase intention: what a sponsored advertisement says and how credibly it says it appears to matter at least as much as how many people it reaches. In the multiple regression model, ad content quality is the dominant unique predictor of purchase intention after controlling for the other three predictors (β = 0.361, f2 = 0.21, large effect), followed by influencer credibility (β = 0.298), consumer engagement (β = 0.241), and advertising exposure (β = 0.192); in the PLS-SEM structural model, ad exposure shows the largest direct path to purchase intention (β = 0.41), followed closely by ad content quality (β = 0.38), with influencer credibility (β = 0.34) and consumer engagement (β = 0.29) also significant. Together, these four predictors explain 63% of the variance in purchase intention (R2 = 0.63) and are confirmed in the latent-variable space by PLS-SEM (R2 = 0.67, Q2 predict = 0.31).

    The study’s most consequential theoretical contribution is the confirmation of partial attitude mediation (H6): advertising exposure is associated with attitude formation consistent with the ELM central route, and that attitude then transmits to purchase intention through the TPB’s attitude antecedent pathway—the one TPB antecedent formally tested in this study. This result extends prior TPB attitude-mediation evidence in the Saudi context () and TPB–ELM integration work in the Gulf region () to a sponsored advertising context with a multi-predictor structural model. The coexistence of a significant direct path alongside the mediated path indicates that peripheral-route persuasion also operates independently—a nuance that this partial TPB–ELM integration captures and that motivates further work extending the framework to the subjective-norms and perceived-behavioral-control pathways not measured here.

    For practitioners, the message is concrete: the most efficient path to high purchase intention among Saudi consumers runs through content quality and influencer credibility, not advertising volume. For the research community, the validated TPB–ELM framework is ready to serve as the theoretical foundation for the next generation of Gulf digital advertising studies—one that should extend to AI-generated content, virtual influencers, and the actual behavioral outcomes that lie beyond stated purchase intention.

    Statements

    Data availability statement

    The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

    Ethics statement

    The studies involving humans were approved by Scientific Research Council of Gulf University, Bahrain. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin because all participants received a standardized information sheet describing the study’s academic purpose, the voluntary and anonymous nature of participation, and their unconditional right to withdraw without consequence. Completion of the questionnaire constituted informed consent.

    AA: Writing – original draft, Formal analysis, Investigation, Data curation, Conceptualization. SB: Methodology, Supervision, Conceptualization, Writing – review & editing, Funding acquisition. TA: Visualization, Writing – review & editing, Supervision.

    Funding

    The author(s) declared that financial support was not received for this work and/or its publication.

    Conflict of interest

    The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

    Generative AI statement

    The author(s) declared that generative AI was used in the creation of this manuscript. The authors acknowledge the use of Claude (Anthropic, claude-sonnet-4-6, 2026) to assist with language editing during the preparation of this work.

    Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

    Publisher’s note

    All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

    References

    • AbdallaR.FaizalS.MenonN.MohammedA. (2025). From scroll to sale: how social media triggers and age shape digital consumer decisions through interaction. Front. Commun.10, 1664694. 10.3389/fcomm.2025.1664694

    • AbdulbaqiH. M. M. (2020). Impact of deceptive advertising on consumer buying decision on social media with mediating word of mouth: an empirical study from University of Tabuk, Saudi Arabia. J. Econ. Adm. Legal Sci.4 (7), 159–182. 10.26389/AJSRP.S120420

    • AhmedH. R. (2024). The impact of influencer credibility dimensions. Electron. Media Commun.54 (3), 183–234. 10.21608/jsec.2024.383166

    • AjzenI. (1991). The theory of planned behavior. Organ. Behav. Hum. Decis. Process.50 (2), 179–211. 10.1016/0749-5978(91)90020-T

    • Al-KenaneK.AlmoraishA.Al-EneziD.Al-MatroukA.AlBuloushiN.AlReshaidF. (2025). The process through which young adults form attitudes towards sustainable products through social media exposure in Kuwait. Sustainability17 (10), 4442. 10.3390/su17104442

    • AlAtawiK. S. (2023). The impact of snapchat advertising on Saudi consumers’ purchase intention: focus on travel and tourism products. J. Umm Al-Qura Univ. Soc. Sci.15 (2), 60–71. 10.54940/ss38228239

    • AlenaziS. (2025). The impact of digital marketing campaigns in Saudi Arabia: a social media sentiment analysis approach. Archit. Image Stud.6 (4), 912–925. 10.62754/ais.v6i4.699

    • AlmutarieA. O. (2024). An examination of the influence of social media on consumer buying behavior within the Saudi Arabian tourism sector. Sci. J. Public Commun. Soc. Sci.28, 444–476. 10.21608/sjocs.2024.361127

    • AlqutubK. (2023). Understanding consumers’ engagement and adoption of social media marketing in the fashion industry in Saudi Arabia: insights through the lens of the theory of planned behavior. Br. J. Mark. Stud.11 (5), 80–99. 10.37745/bjms.2013/vol11n58099

    • AlSahliN. (2024). Impact of social media promotion on purchase decisions from luxury stores: the mediator role of purchase intention. J. Int. Crisis Risk Commun. Res.7 (S12), 298–313. 10.63278/jicrcr.vi.994

    • AlsalloumA. A. (2025). Female social media influencers in the Arab Gulf: shaping marketing and advertising on Instagram. Front. Commun.10, 1518386. 10.3389/fcomm.2025.1518386

    • AzadM. S.KhanS. S.HossainR.RahmanR.MomenS.ChowdhuryS. (2023). Predictive modeling of consumer purchase behavior on social media: integrating theory of planned behavior and machine learning for actionable insights. PLoS One18 (12), e0296336. 10.1371/journal.pone.0296336

    • BalrooS. A. (2023). Social media advertisements and purchase intention: attitude as a mediator. J. Econ. Manag. Trade29 (10), 160–174. 10.9734/jemt/2023/v29i101152

    • CohenJ. (1988). Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Lawrence Erlbaum Associates. 10.4324/9780203771587

    • DataReportal. (2024). Digital 2024: Saudi Arabia. Available online at:https://datareportal.com/reports/digital-2024-saudi-arabia(Accessed February 23, 2024).

    • HairJ. F.HenselerJ.RingleC. M.SarstedtM. (2022). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM). 3rd ed. Cham: SAGE Publications. 10.1007/978-3-030-80519-7

    • HenselerJ.RingleC. M.SarstedtM. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Mark. Sci.43 (1), 115–135. 10.1007/s11747-014-0403-8

    • HovlandC. I.JanisI. L.KelleyH. H. (1953). Communication and Persuasion: Psychological Studies of Opinion Change. New Haven: Yale University Press. Available online at:https://archive.org/details/communicationper0000unse(Accessed January 15, 2020).

    • HuoW.FilieriR. (2025). Social media marketing and purchase intention: the mediation of customer-based brand equity and adoption preparedness. Eur. J. Mark.59 (12), 2886–2927. 10.1108/EJM-10-2023-0783

    • HussainA.TingD. H.MazharM. (2022). Driving consumer value co-creation and purchase intention by social media advertising value. Front. Psychol.13, 800206. 10.3389/fpsyg.2022.800206

    • Influencer Marketing Hub. (2025). The State of Influencer Marketing 2025: Benchmark Report. Available online at:https://influencermarketinghub.com/influencer-marketing-benchmark-report/(Accessed May 4, 2026).

    • KockN. (2015). Common method bias in PLS-SEM: a full collinearity assessment approach. Int. J. e-Collab.11 (4), 1–10. 10.4018/ijec.2015100101

    • LouC.YuanS. (2019). Influencer marketing: how message value and credibility affect consumer trust of branded content on social media. J. Interact. Advert.19 (1), 58–73. 10.1080/15252019.2018.1533501

    • NunnallyJ. C. (1978). Psychometric Theory. 2nd ed. New York: McGraw-Hill. Available online at:https://www.scirp.org/reference/ReferencesPapers?ReferenceID=1867797(Accessed September 13, 2016).

    • PettyR. E.CacioppoJ. T. (1986). Communication and Persuasion: Central and Peripheral Routes to Attitude Change. New York, NY: Springer. 10.1007/978-1-4612-4964-1

    • PodsakoffP. M.MacKenzieS. B.LeeJ.-Y.PodsakoffN. P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. J. Appl. Psychol.88 (5), 879–903. 10.1037/0021-9010.88.5.879

    • PrasadK. D. V.VaniU. K. (2025). Impact of social media advertising on consumer behavior: role of credibility, perceived authenticity, and sustainability. Front. Commun.10, 1595796. 10.3389/fcomm.2025.1595796

    • PwC Middle East. (2025). Voice of the Consumer 2025 Survey: Saudi Arabia Findings. PwC Middle East. Available online at:https://www.pwc.com/m1/en/publications/2025/docs/voice-of-the-consumer-2025.pdf

    • RingleC. M.WendeS.BeckerJ.-M. (2024). SmartPLS 4 [Computer software]. SmartPLS, Bönningstedt, Germany. Available online at:https://www.smartpls.com

    • SaadM.AwadA.AzizA. F.ShmaT. R. (2025). Influencer marketing’s impact on credibility and purchase intention: a study on university of Bisha students in Saudi Arabia. Innov. Mark.21 (1), 326–337. 10.21511/im.21(1).2025.26

    • SalamahM. W. (2020). Influencer advertisingng Saudi university youth: a field study. Sci. J. Public Relat. Advert. Res.19 (19), 605–648. 10.21608/sjocs.2020.155417

    • SalehS. (2025). Social media usage in Saudi Arabia—statistics and facts. Statista. Available online at:https://www.statista.com/topics/9947/social-media-usage-in-saudi-arabia/(Accessed December 17, 2025).

    • SardarS.VijayT. S. (2025). Social media influencers and purchase intention: a review and research agenda. Int. J. Consum. Stud.49 (3), 376–390. 10.1111/ijcs.70046

    • XingyuD.ChenC.-N.ShokouhifarM. (2024). Impacts of social media advertising on purchase intention and customer loyalty in e-commerce systems. ACM Trans. Asian Low-Resour. Lang. Inf. Process.23 (8), 112:1–112:15. 10.1145/3613448

    Summary

    consumer engagement, elaboration likelihood model, influencer credibility, PLS-SEM, purchase intention, Saudi Arabia, sponsored social media advertising, theory of planned behavior

    Alqahtani A, Badran S and Alhassan T (2026) Sponsored advertising on social media and purchase intention: an attitude-mediated TPB–ELM model among Saudi consumers. Front. Commun. 11:1876021. doi: 10.3389/fcomm.2026.1876021

    Tereza Semerádová, Technical University of Liberec, Czechia

    Kah Boon Lim, Multimedia University, Malaysia

    Reem Abbas Abdalla, University of Technology Bahrain, Bahrain

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

    ORCID Tarik Alhassan orcid.org/0000-0002-0221-3240

    Share.

    Related Posts

    Leave A Reply