The COVID-19 pandemic, formally declared by the World Health Organization (WHO) on March 11, 2020, triggered one of the most consequential public health crises in recent history ([14]). Within weeks, governments worldwide imposed sweeping quarantine measures while simultaneously confronting a parallel crisis: an “infodemic”—a flood of accurate and inaccurate information that complicated public understanding and behavioral compliance ([18]; [43]). Social media platforms emerged as the primary arena in which health guidance, misinformation, fear, and community solidarity competed for public attention and shaped behavioral responses ([12]; [37]).

In the Philippines, the Enhanced Community Quarantine (ECQ) was announced on March 13, 2020, initially covering the National Capital Region (NCR) before expanding to nearby provinces as local transmission accelerated ([44]). Metro Manila, as the country’s most densely populated region and undisputed pandemic epicenter, experienced the most stringent enforcement of quarantine protocols and the most intense public scrutiny of compliance with these protocols—much of it conducted on social media platforms ([35]). This digital-behavioral intersection makes Metro Manila a uniquely appropriate context for examining how social media moderates adherence behaviors during public health emergencies.

The Philippines consistently ranks among the world’s highest social media engagement nations; Filipinos spend an average of more than five hours daily on social networking platforms, a figure that increased substantially during pandemic-era lockdowns as in-person interaction ceased and digital channels became the primary mode of information access and social connection ([22]; [31]). Against this backdrop, the influence of social media on COVID-19 health behaviors acquired particular empirical salience. Yet few studies have empirically tested social media’s moderating—rather than its direct predictive—role in shaping the relationship between behavioral intentions and actual quarantine compliance in the Metro Manila context.

The present study addresses this gap by extending the Theory of Planned Behavior ([2]) to include social media as a moderating variable, testing this extended model with data gathered during the critical ECQ-to-GCQ transition period. The study contributes to three areas of scholarship: 1) The empirical literature on social media and health behavior compliance during infectious disease crises; 2) TPB theory by demonstrating social media’s conditioning role on attitudinal and control-related behavioral predictors; 3) Digital health communication practice by providing evidence-based insights relevant to future pandemic response design in high-social-media-use, collectivist societies.

This study extends [2] Theory of Planned Behavior by incorporating social media as a moderating variable between the framework’s core cognitive constructs and quarantine adherence. The original TPB posits that attitude, subjective norm, and perceived behavioral control determine behavioral intention and, subsequently, actual behavior. In the present adaptation, perceived behavioral control is decomposed into its two constituent components—self-efficacy and controllability—following established precedent in health behavior research ([35]). Critically, the study departs from prior applications that position media as a direct behavioral antecedent or mediating variable, instead treating social media as a contextual amplifier that conditions the strength of relationships between pre-existing behavioral antecedents and compliance outcomes.

The theoretical rationale for treating social media as a moderator rather than a mediator or direct predictor warrants explicit justification. As a mediator, social media would transmit the effect of TPB antecedents to adherence—implying that attitudinal or control-based influences must pass through social media exposure to affect behavior. As a direct predictor, social media would independently generate compliance regardless of pre-existing cognitive orientations. Neither specification is theoretically supportable in this context: residents’ attitudes, subjective norms, and control beliefs develop prior to and independently of their social media environments; exposure alone cannot produce compliance without antecedent motivational states ([20]). The moderating specification, by contrast, positions social media as a contextual amplifier that strengthens or attenuates the behavioral translation of pre-existing cognitions—a role supported by emerging empirical evidence that the behavioral effects of TPB antecedents vary meaningfully across levels of digital health information engagement ([49]). [4], in a UK pandemic compliance study, demonstrated that higher social media health campaign engagement intensified the relationship between risk-protective attitudes and actual preventive behaviors, consistent with a moderating rather than mediating role. [48] similarly found in a cross-national COVID-19 compliance study that social media moderation of TPB constructs was more theoretically and statistically defensible than mediation or direct prediction models, particularly where behavioral antecedents are strongly pre-determined by personal experience and community context. The moderating role is further grounded in the information environment literature: in a digital ecosystem where behavioral norms, risk perceptions, and compliance guidance are primarily communicated through social platforms, individuals’ attitudinal orientations and perceived control beliefs do not operate in isolation from their social media environment ([12]; [28]). Social media exposure may reinforce positive attitudes, increase normative salience, and provide practical resources that enhance perceived controllability, thereby amplifying pathways from behavioral antecedents to compliance behaviors.

Figure 1 presents the study’s conceptual framework. The model positions social media (SM) as a moderating variable between attitude (ATT), subjective norms (SN), self-efficacy (SE), controllability (CONT), and adherence to quarantine protocols (ATQP). Direct paths (H1-H4) retain the original TPB structure, while moderation paths (H5a-H5d) represent the study’s central theoretical contribution.

Figure 1. Conceptual framework: extended theory of planned behavior with social media as moderating variable.

This study employed a quantitative, predictive research design to examine the moderating role of social media campaigns in the relationship between TPB constructs and COVID-19 quarantine adherence among Metro Manila residents. Predictive research design is appropriate when the objective is to assess directional relationships between specified antecedents and behavioral outcomes while controlling for confounding influences ([21]).

Primary data were gathered through an online, self-administered survey distributed across Metro Manila’s 16 cities and one municipality from August 15 to September 20, 2020. This period corresponds to the government-mandated transition between the Enhanced Community Quarantine (ECQ) and General Community Quarantine (GCQ), capturing behavioral compliance at a moment of regulatory flux when residents depended heavily on social media for updated quarantine protocol information.

Convenience sampling yielded 413 usable responses. The survey was distributed through online channels given the ECQ-period restriction on face-to-face contact. Specifically, the survey link was disseminated via Facebook community groups organized by barangay and city, institutional pages of De La Salle University and partner organizations, professional network chains through respondents’ workplaces, and snowball referral through initial respondents’ personal networks. Inclusion criteria required respondents to be Metro Manila residents during the August-September 2020 data collection period, aged 21 - 40 years, and active social media users (defined as daily platform use for any purpose). The restriction to respondents aged 21 - 40 was deliberate: this cohort constitutes the primary economically active, internet-connected population segment in Metro Manila and represents the highest social media engagement demographic in the Philippines ([22]). This age bracket was also most directly subject to work-from-home mandates and quarantine pass enforcement during the ECQ period, making their compliance behaviors both theoretically and practically significant. Researchers acknowledge that this restriction limits generalizability to younger (below 21) and older (above 40) Metro Manila populations, for whom social media’s moderating effects on quarantine compliance may differ. The sample comprised 217 males (52%) and 196 females (48%) respondents. Occupationally, 84% were employed and 16% were students. Educational attainment was predominantly at the college graduate level (69%), with 13% holding master’s degrees. Monthly income was concentrated in the PHP 21,000 - 30,000 range (69%). Regarding social media use, 81% of respondents reported spending four to seven hours daily on social media platforms, with 58% primarily using these platforms for news and information consumption—a pattern consistent with pandemic-era digital behavior documented in similar contexts ([22]; [31]).

The survey instrument comprised seven sections covering attitude, subjective norm, self-efficacy, controllability, social media use, adherence to quarantine protocols, and demographic characteristics. All construct items were measured using a five-point Likert scale (1 = Strongly Disagree to 5 = Strongly Agree). For this study, social media campaigns are operationally defined as structured, purposive health communication initiatives disseminated by government agencies (e.g., the Department of Health, local government units), public health authorities, or credible community organizations through social networking platforms—primarily Facebook, which served as the dominant platform for COVID-19 public health communication in the Philippines during this period. Respondents’ engagement with social media campaigns was measured as a composite of two dimensions: a) exposure, operationalized as the frequency of encountering COVID-19-specific public health guidance content on social media; b) engagement, operationalized as the degree of active behavioral interaction with that content, including reading in full, liking, sharing, and following embedded guidance. This construct is explicitly distinguished from general social media use—referring to overall platform time regardless of content type or purpose—and from passive incidental exposure to health-adjacent content not produced by credible health authorities. All TPB constructs (attitude, subjective norm, self-efficacy, controllability) and quarantine adherence were measured reflectively, consistent with standard practice in TPB research where items are theorized as interchangeable manifestations of a latent construct rather than its formative causal indicators ([21]). Scale items were adapted from the Handbook of Marketing Scales ([6]) and supplemented with domain-specific items developed through a systematic review of the COVID-19 health behavior literature. Appendix A provides the full item wording and source references for all constructs. The instrument underwent pre-testing with a pilot sample to assess item clarity and preliminary reliability. Participation was voluntary and anonymous, and informed consent was obtained from all respondents prior to survey participation. No personally identifiable information was collected.

Data analysis employed Partial Least Squares Structural Equation Modeling (PLS-SEM) via SmartPLS 3.0 ([36]). PLS-SEM is particularly well-suited for this study’s complexity, given its capacity to simultaneously model multiple moderating relationships, handle non-normal data distributions, and yield reliable estimates with moderate sample sizes ([21]). A two-stage analysis approach was followed: first evaluating the measurement model for reliability and convergent/discriminant validity, then assessing the structural model’s predictive relationships and moderation paths.

Table 4 presents the structural model results via bootstrapping in SmartPLS. The significance threshold was set at p < 0.01 ([38]). Attitude significantly predicted quarantine adherence (H1: β = 0.518, t = 4.695, p < 0.001), supporting the hypothesis and aligning with extensive prior research confirming attitude as a robust behavioral antecedent across health compliance contexts ([1]; [29]; [39]; [47]; [49]). Controllability similarly demonstrated significant direct effects on adherence (H4: β = 0.242, t = 3.789, p < 0.001), confirming that structural enabling conditions play a material role in compliance behavior independently of attitudinal and normative factors. Subjective norm (H2: β = −0.002, p = 0.934) and self-efficacy (H3: β = 0.074, p = 0.100) did not significantly predict adherence. These null findings are theoretically meaningful and may reflect several complementary explanations. One plausible account draws on Self-Determination Theory ([15]), which distinguishes behaviors arising from genuine personal valuation versus those performed in response to external expectations—suggesting compliance may have been driven more by autonomous motivations than by normative pressure or capability assessments alone. However, additional mechanisms warrant consideration. The non-significant subjective norm effect may partly reflect measurement constraints: ECQ-era social isolation reduced visibility of in-person normative behavioral cues that standard subjective norm items were designed to capture. The non-significant self-efficacy effect may similarly reflect ceiling conditions, with near-universal public awareness of quarantine protocols—repeatedly amplified through social media—rendering individual capability assessments less behaviorally discriminating at this stage of the pandemic. Enforcement intensity during the ECQ-to-GCQ transition, characterized by barangay-level checkpoints, legal penalties for violations, and visible community monitoring, may further have compressed self-efficacy’s behavioral variance by making compliance a structural expectation rather than a purely capability-dependent choice. Risk perception, which was not directly measured in this study, represents another plausible driver of autonomous compliance behavior that merits examination in future research. These alternative accounts suggest the null direct effects of subjective norm and self-efficacy should be interpreted with appropriate caution rather than as definitive evidence of intrinsic motivation dominance; future studies incorporating motivational regulation scales would clarify the underlying mechanisms. This pattern is broadly consistent with Self-Determination Theory, which distinguishes between behaviors arising from genuine personal valuation versus those performed in response to external expectations.

Table 4 also presents the moderation results for H5a-H5d, testing whether social media (SM) significantly moderates the relationships between each TPB construct and adherence to quarantine protocols (ATQP). Remarkably, all four moderation hypotheses were supported, indicating that social media consistently and significantly conditioned the influence of attitude, subjective norm, self-efficacy, and controllability on quarantine compliance.

Table 4. Structural model results—direct and moderation path.

Note: **p < 0.01. β = standardized path coefficient; SD = Standard Deviation.

To further specify the nature of the moderation effects, conditional effects were examined by interpreting the direction and magnitude of each interaction coefficient across varying levels of social media engagement. Consistent with all positive and significant interaction terms in Table 4, each TPB-to-adherence path strengthens as social media engagement increases from low (−1 SD below the mean) to high (+1 SD above the mean) levels. At high social media engagement levels, the controllability-adherence path is most steeply amplified (H5d: β-interaction = 0.601), meaning that residents who were both highly engaged with social media campaigns and high in controllability demonstrated the strongest quarantine adherence in the entire model. Similarly, the self-efficacy-adherence and attitude-adherence paths show meaningful positive slope increases at higher social media engagement (H5c: β-interaction = 0.497; H5a: β-interaction = 0.492), confirming that both capability-based and evaluative antecedents are behaviorally activated more fully for high social media users. The subjective norm-adherence path, while showing the smallest interaction effect (H5b: β-interaction = 0.391), still demonstrates a meaningful positive slope at higher engagement levels—indicating that normative pressure acquires behavioral relevance primarily under conditions of active social media campaign engagement. Critically, the subjective norm and self-efficacy direct paths remained non-significant (H2, H3), yet their moderated slopes are positive at high social media levels, confirming that these pathways are activated rather than merely amplified by social media. This conditional pattern—where dormant direct paths become behaviorally operative through social media moderation—represents the study’s most theoretically distinctive finding and underscores the importance of high-quality, accessible social media campaigns in mobilizing the full range of behavioral antecedents available to public health communicators. Social media most powerfully moderated the controllability-adherence relationship (H5d: β = 0.601, t = 7.146, p < 0.001), the strongest path in the entire model. This suggests that social media functioned as an information infrastructure during the pandemic, providing residents with real-time access to resource guides, assistance programs, and updated protocol information that strengthened their perceived structural capacity to comply. The stability of this effect across bootstrapping iterations (SD = 0.084) further confirms its robustness.

The moderation of the attitude-adherence relationship (H5a: β = 0.492, t = 3.870, p < 0.001) indicates that social media reinforced pre-existing positive orientations toward quarantine, translating favorable attitudes into actual compliance behavior more effectively than attitude alone predicted. This is consistent with evidence that health media exposure amplifies attitudinal pathways to protective behavior ([27]; [23]).

Theoretically significant are the supported moderation effects for self-efficacy (H5c: β = 0.497, t = 4.425, p < 0.001) and subjective norm (H5b: β = 0.391, t = 3.633, p < 0.001)—both of which had no significant direct effects on adherence. Social media effectively activated these otherwise dormant pathways, suggesting it does not merely reinforce existing behavioral dispositions but creates new ones. Digital platforms enhanced self-efficacy by exposing residents to practical compliance guidance and peer behavioral modeling, while making community normative expectations visible and salient in ways that face-to-face interaction during lockdown could not ([9]; [17]). The hierarchy of moderation strength—controllability, self-efficacy, attitude, subjective norm—implies that social media’s most critical behavioral function during the pandemic was structural enablement rather than attitudinal reinforcement alone, a distinction with direct implications for public health campaign design.

Taken together, these findings position social media as a universal behavioral activator across both attitudinal and structural pathways, with implications discussed in the sections that follow.

This study provides evidence that social media campaigns functioned as powerful behavioral amplifiers of quarantine adherence among Metro Manila residents during the 2020 ECQ-to-GCQ transition. The extended TPB model confirms that attitude and controllability are robust direct predictors of compliance, while social media universally strengthened all four TPB-adherence pathways—including the subjective norm and self-efficacy pathways that showed no significant direct effects. This finding reveals that social media’s behavioral influence extends beyond reinforcing pre-existing dispositions: it activates dormant pathways and creates compliance-enabling conditions that would not otherwise emerge.

The study’s central theoretical contribution is the empirical demonstration that social media conditions—rather than independently determining—the behavioral influence of cognitive and structural antecedents. The hierarchy of moderation strength, with controllability yielding the strongest effect, further suggests that social media’s most critical pandemic function was structural enablement rather than attitudinal persuasion alone. This distinction carries direct implications for how public health authorities design and deploy digital campaigns during future health emergencies—prioritizing resource information and compliance facilitation alongside motivational messaging.

For public health authorities and government communicators, findings recommend that social media investment prioritize content that simultaneously reinforces positive behavioral attitudes and addresses structural barriers to compliance. LGU-level social media accounts and national health agencies would benefit from integrating real-time resource guides—detailing food assistance programs, financial support mechanisms, and updated quarantine protocol tiers—within their campaign content, directly targeting the controllability pathway identified in this study.

For marketing practitioners operating in post-pandemic consumer markets, the sustained association between social media engagement and health-protective behavioral attitudes in this study reflects a broader shift in Filipino consumer expectations. Social media strategies that position brands within health-protective, community-enabling frameworks may find stronger resonance with audiences conditioned by pandemic-era digital health communication than those relying exclusively on traditional commercial messaging frames.

This study’s cross-sectional design limits directional inference beyond the predictive associations identified through PLS-SEM. The convenience sample, while sufficiently large for model estimation, was restricted to Metro Manila residents aged 21 - 40 with internet access, likely underrepresenting older residents, those without digital connectivity, and more economically and educationally diverse populations—segments for whom social media’s moderating influence on quarantine adherence may differ considerably.

Future research should employ longitudinal designs to examine how social media’s moderating role shifts across pandemic phases, from initial lockdown through vaccine rollout and endemic normalization. Platform-specific analyses distinguishing Facebook, YouTube, and short-form video platforms would deepen theoretical understanding, given that the present study treats social media engagement without differentiating by platform or content quality. Incorporating misinformation exposure as a parallel moderating variable would address this gap directly. Regional comparative studies across Philippine provinces would further extend the model beyond the Metro Manila context and allow examination of geographic and cultural variation in the social media-TPB-adherence relationship.

This manuscript is a revised and updated version of the article originally published in Advances in Journalism and Communication (Vol. 9, 2021, pp. 85-101; DOI: 10.4236/ajc.2021.93007). The revision ameliorates theoretical perspectives and remedies the discussion of post-pandemic insights. The original dataset (N = 413 Metro Manila residents surveyed August-September 2020), research design, and statistical findings are retained unchanged. The updated version also addresses similar observations identified in the earlier version, which arose inadvertently and have since been carefully reviewed and corrected through improved citation, paraphrasing, and attribution. These revisions aim to strengthen the clarity, academic rigor, and scholarly contribution of the study. The author declares no conflict of interest.

All constructs were measured on a five-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree) and modeled reflectively. Items are listed below with their source references.

Attitude toward Quarantine Protocols (ATT)—adapted from [1] and [33]

ATT1. I believe that following quarantine protocols is an effective way to prevent COVID-19 from spreading.

ATT2. I think that quarantine protocols are necessary given the current state of the pandemic.

ATT3. Complying with quarantine measures is a responsible action that I support.

Subjective Norm (SN)—adapted from [2] and [46]

SN1. People who are important to me expect me to comply with quarantine protocols.

SN2. My family members think I should follow quarantine measures.

SN3. Most people in my community are complying with quarantine protocols.

Self-Efficacy (SE)—adapted from [2] and [35]

SE1. I am confident that I can follow quarantine protocols even when it is inconvenient.

SE2. I have the ability to stay at home and avoid unnecessary travel as required by quarantine rules.

SE3. I feel capable of adhering to quarantine protocols for an extended period.

Controllability (CONT)—adapted from [2] and [3]

CONT1. I have access to sufficient food and basic necessities that allow me to stay at home.

CONT2. My living situation makes it possible for me to comply with quarantine protocols.

CONT3. External conditions (financial, social, logistical) make it feasible for me to adhere to quarantine protocols.

Social Media Campaign Engagement (SM)—adapted from [30] and [9]

SM1. I regularly encounter COVID-19 health campaign posts from government or health authorities on social media.

SM2. I actively read or watch COVID-19 prevention content shared by health agencies on social media.

SM3. I share COVID-19 health campaign information from credible sources on my social media accounts.

Adherence to Quarantine Protocols (ATQP)—adapted from [35] and [3]

ATQP1. I stay at home except when going out for essential needs, as required by quarantine protocols.

ATQP2. I wear a face mask and face shield when going outside, as required by quarantine rules.

ATQP3. I observe physical distancing and avoid mass gatherings in accordance with quarantine protocols.

ATQP4. I comply with curfew hours and travel restrictions imposed under quarantine guidelines.