The Stimulus-Organism-Response (SOR) framework has been widely applied in e-commerce and social commerce research to explain how external stimuli influence consumer behavior through internal psychological states. Stimuli trigger emotions such as pleasure and arousal, which influence whether the individual approaches or avoids the stimulus [8]. Subsequent use of the concept extended these emotional parameters with more generalizable cognitive and affective constructs in social media and e-commerce settings, such as trust and engagement [9]-[12]. Furthermore, in social media advertising, attributes such as information, entertainment, credibility, or source-related stimuli might influence consumer attitude, behavioral intention, and perceived value [6][13][14].

Six advertising attributes are relevant to the TikTok Shop context: 1) informativeness, or the usefulness of product information; 2) entertainment, or the enjoyment gained from an advertisement; 3) visual appeal, or the attractiveness of its presentation; 4) credibility, or how trustworthy the advertisement appears; 5) influencer persuasiveness, or how convincing an endorsement is; and 6) content clarity, or how clearly product features are communicated. These traits have previously been associated with consumer attitudes, behavioral intentions, and perceived value in social media advertising [4][13][14].

Consumer engagement is defined as the psychological state that reflects active behavioral interaction with, and a positive emotional response to, a stimulus [11]. In the social media commerce context, consumer engagement is illustrated by actively watching product videos in advertisements, interacting with those advertisements through likes, comments, and shares, and spending time investigating featured products. Prior studies identify consumer engagement as an important psychological response in online/social advertising contexts [12][15].

In social commerce contexts, where uncertainty is higher due to increased information asymmetry between buyers and sellers, consumer trust plays an important role. Consumer trust is defined as the willingness to rely on an exchange partner based on positive expectations about its actions [16]. Consumer trust in TikTok Shop refers to reliance, belief in product information, perceived transactional security, and confidence in the seller. TikTok Shop-specific research also identifies trust as an important factor in purchase intention [1]. Studies have shown consumer trust predicts purchase and continuance intention in social commerce and live-streaming commerce contexts [10][17][18]. Trust therefore represents beliefs about reliability, safety, and information credibility, whereas purchase intention represents a respondent’s stated likelihood of making a future purchase.

Based on the literature above, we hypothesize as follows:

H1: The six TikTok Shop advertising attributes are positively associated with consumer engagement.

H2: The six TikTok Shop advertising attributes are positively associated with consumer trust.

H3: Consumer engagement is positively associated with trust in TikTok Shop.

H4: Consumer engagement is positively associated with Gen Z purchase intention.

H5: Trust is positively associated with Gen Z purchase intention.

H6: Mediating effect of consumer engagement and trust on the relationship between stimulus attributes of TikTok Shop advertisements and Gen Z purchase intention.

The study uses a quantitative, cross-sectional survey design. The survey was administered through Google Forms and distributed using purposive online outreach through Instagram, personal networks, and other public online channels. A total of 270 responses were received. Eligibility was assessed using self-reported screening questions asking whether respondents identified as Generation Z (born between 1997-2012), used TikTok, and had seen product advertisements on TikTok Shop. Of the 270 responses, 59 did not identify as Gen Z, 20 identified as Gen Z but did not use TikTok, and 1 used TikTok but had not seen TikTok Shop product advertisements. The remaining 190 respondents formed the final sample for analysis. The survey was not restricted to a single geographic market, and respondent location was not collected.

The sample included 116 males (61.1%) and 72 females (37.9%); 2 individuals identified as none (1.1%). The mean age of respondents was 20.23 years (SD = 5.82), with a range of 13 to 40 years. Regarding academic qualifications, 55.3% reported high school education, 20.5% college-level education, and the remainder held graduate degrees or other qualifications; 70.5% reported daily TikTok use, and 55.8% confirmed they shop on e-platforms frequently every week. Eligibility was based on the screening responses rather than independent verification of demographic information.

All the items were operationalized using a 5-point Likert scale anchored from 1 to 5 (1 = strongly disagree; 5 = strongly agree). The survey items were developed specifically for this study from the construct definitions and literature reviewed above rather than directly adopted from an existing validated scale. The questionnaire was reviewed with the faculty co-author and pilot-tested before final distribution; feedback on item wording and clarity was used to refine the questionnaire. The stimulus component included six distinct single-item advertising attributes: informativeness, entertainment, visual appeal, credibility, influencer persuasiveness, and content clarity. The 4 items of consumer engagement are active watching, interacting with content, investigating the product, and felt engagement, whereas the 4 items of trust included product trust, seller reliability, safety of purchase, and product information trustworthiness. Purchase intention was measured using the 2 items that directly assessed intent to purchase and likelihood of buying recommended products. The additional items on considering TikTok Shop when shopping online and recommending purchases from TikTok Shop were treated as separate platform-consideration and advocacy outcomes and were not included in the purchase-intention composite.

Descriptive statistics, Cronbach’s alpha reliability analysis, and Pearson correlations were used in the analysis using the software Python (pandas v1.5; scipy v1.10). Composite scores were calculated as the mean of the items for engagement, trust, and purchase intention; the six advertising stimulus attributes were analyzed separately rather than treated as a validated unidimensional construct. Internal consistency was evaluated using Cronbach’s alpha [19]. Discriminant validity between trust and purchase intention was assessed using the heterotrait-monotrait ratio (HTMT) [20]. Pearson correlations examined the bivariate relationships used to test H1 - H5.

To address H6, we performed separate regression-based serial mediation analyses for each of the six advertising attributes, with engagement and trust as sequential mediators and purchase intention as the outcome.

We calculated 95% confidence intervals for the direct and specific indirect effects using 10,000 bootstrap resamples [21].

For advertising stimuli, informativeness received the highest rating (M = 3.28, SD = 1.04), closely followed by Entertainment (M = 3.24, SD = 0.97), while Content Clarity received the lowest rating (M = 2.58, SD = 1.15). Regarding engagement, Perceived Engagement has the highest mean score (M = 3.30, SD = 0.87), whereas Active Interaction has received a conspicuously low mean rating (M = 1.58, SD = 0.85). The trust scale found perceived purchase safety to have the highest rating among sub-scale measurements (M = 3.34, SD = 0.98). The two purchase-intention items also had relatively high mean scores, with Intent to Purchase the highest (M = 3.74, SD = 1.34). (See Table 1)

Table 1. Descriptive statistics for survey items (N = 190).

Note. PI = Purchase Intention. Platform consideration and advocacy were analyzed separately and were not included in the PI composite.

The three multi-item constructs showed good internal reliability above the 0.70 threshold [19]. Trust showed the largest scale reliability (α = 0.925), followed by purchase intention (α = 0.885) and engagement (α = 0.811). Because the six advertising attributes were modeled separately, no alpha was used to claim that they formed a unidimensional construct. The HTMT value between trust and purchase intention was 0.947, above the 0.90 benchmark [20], indicating that discriminant validity between these two measures was not fully established. (See Table 2)

Table 2. Multi-item construct reliability and composite means.

All six advertising attributes were positively associated with engagement (r = 0.351 - 0.677, all p < 0.001), supporting H1, and all six were positively associated with trust (r = 0.266 - 0.706, all p < 0.001), supporting H2. Table 3 reports the attribute-level associations.

Table 3. Advertising attribute correlations with engagement and trust (N = 190).

Note. ***p < 0.001.

The association between engagement and trust supports H3 (r = 0.623).

The positive association between engagement and purchase intention supports H4 (r = 0.566). Trust showed the strongest association with purchase intention (r = 0.865), supporting H5. Table 4 presents the correlations among engagement, trust, and purchase intention. H6 is reported in the serial mediation analysis in Table 5.

Table 4. Pearson correlation matrix for engagement, trust, and purchase intention (N = 190).

Note. PI = Purchase Intention. ***p < 0.001.

Table 5. Summary of hypothesis testing results.

*Note. H6 was tested using separate serial mediation analyses for the six advertising attributes with 10,000 bootstrap resamples. The sequential indirect effect was significant for five of the six attributes (see Table 5).

Serial mediation was estimated separately for each advertising attribute. Table 6 reports the direct effect and each specific indirect effect with 95% bootstrap confidence intervals. The sequential indirect path through engagement and trust was significant for entertainment, visual appeal, credibility, influencer persuasiveness, and content clarity because their confidence intervals excluded zero. The sequential path for informativeness was not significant because its confidence interval included zero. H6 was therefore partially supported. Influencer persuasiveness and content clarity also showed significant negative direct effects on purchase intention despite positive total indirect effects. This pattern is consistent with competitive or inconsistent mediation, suggesting that their positive indirect effects on purchase intention operate primarily through engagement and trust rather than through a positive direct pathway. Visual appeal showed a more complex pattern: its trust-only indirect effect was negative and nonsignificant, while its sequential indirect effect through engagement and trust was comparatively strong and significant, suggesting that engagement may be important in translating visual appeal into trust-related purchase intention. (See Table 6)

Table 6. Serial mediation effects by advertising attribute.

Note. Values are unstandardized B coefficients with 95% bootstrap confidence intervals in brackets. Intervals that exclude zero indicate a statistically significant effect.

First, this study contributes to theory extension by applying the S-O-R framework to TikTok Shop, a relatively novel short-form social commerce channel, and by verifying its relevance across emerging platforms. This empirical confirmation extends the SOR framework’s broad applicability to new communication channel types. At the attribute level, all six advertising attributes were positively associated with both engagement and trust, supporting and expanding existing findings on social media advertising effects [13][14].

In particular, the notably strong link between trust and purchase intention (r = 0.865) reiterates the significance of trust in unfamiliar, high-risk e-commerce contexts and is consistent with TikTok Shop-specific evidence [1][10][17][18].

Because trust showed the strongest relationship with purchase intention among the variables studied, this suggests that trust may be particularly important in TikTok Shop purchase intention. Furthermore, the findings emphasize the importance of evaluating Gen Z social commerce behavior through a trusting lens.

The strongest relationship found between trust and purchase intention highlights the need for marketing activities targeting Gen Z on TikTok Shop to prioritize building trust and credibility, alongside entertainment- or visual-focused efforts. Specifically, brands should provide transparent product information, highlight seller credibility, offer secure payment processes, and use authentic rather than purely promotional influencer endorsements.

The relatively lower scores for Content Clarity (M = 2.58) and Active Interaction (M = 1.58) suggest key areas for improvement. Advertisers on TikTok Shop need to balance high entertainment value with more substantial and clear information in their marketing campaigns. Increased efforts through interactive product demonstrations, engaging Q & A sessions, and prompt responses to comments can promote greater interaction and eventual purchase among consumers.

Future research can extend this line of inquiry by addressing limitations in our study, such as the cross-sectional design that restricts drawing definitive causal inferences between S-O-R constructs. A longitudinal design or experimental methods could better establish temporal order and causality in stimulus-organism-response relationships. Despite these causality limitations, correlational findings can still provide support.

Most respondents in our sample were high school-educated male Gen Z users; consequently, their perceptions may not generalize to all demographics.

This self-selection into the sample makes it crucial for studies of social media consumption to ensure a wide and varied age and demographic distribution among participants. Generational eligibility and TikTok exposure were determined through self-report, and some reported ages were inconsistent with the stated Gen Z birth-year range; therefore, demographic responses were not independently verified. Respondent geography was also not collected. The researcher-developed measures were reviewed and pilot-tested but were not directly adopted from previously validated scales. In addition, the HTMT value between trust and purchase intention (0.947) exceeded the 0.90 benchmark, so discriminant validity was not fully established and the strong trust-purchase intention relationship should be interpreted cautiously. Future research should validate more sharply differentiated measures and could also explore moderating factors such as prior purchase experience, product category, and social media literacy across a wider and more diverse group of Gen Z consumers.