Predictors of social media users' intention to donate online towards international NGOs in the fake news era

The paper investigates how exposure to and sharing of fake news on social media affects social media users’ intention to donate online to international NGOs (INGOs). Framed by the Theory of Planned Behaviour (TPB), the study addresses a research gap on fake news effects in the non-profit sector. It proposes and tests a conceptual model linking social media usage, brand reputation, fake news sharing, perceived risks of online donations, attitudes toward online donations, brand image, and intention to donate online. The study aims to clarify how external factors in social media environments shape attitudes and intentions toward online giving to INGOs.

The literature review grounds the model in TPB, where intention is shaped by attitude, subjective norms, and perceived behavioral control. Prior research shows extrinsic factors (project, organization, platform information) influence attitudes and donation intentions; attitudes and perceived behavioral control often predict intention to donate. The review highlights social media’s role in facilitating e-philanthropy and brand-building for NGOs, while simultaneously enabling rapid diffusion of misinformation. Fake news can harm organizations’ reputations, trust, and donor relationships. From this base, ten hypotheses were developed: H1: Social media usage positively influences sharing fake news about INGOs; H2: INGO brand reputation positively influences sharing fake news; H3: Sharing fake news positively influences perceived risks of online donations; H4: Sharing fake news positively influences attitude toward donating online; H5: Sharing fake news positively influences intention to donate online; H6: Perceived risks positively influence intention to donate online; H7: Attitude toward donating online positively influences perceived risks; H8: Attitude toward donating online positively influences intention to donate online; H9: Attitude toward donating online positively influences INGO brand image; H10: INGO brand image positively influences intention to donate online.

Design: Quantitative, cross-sectional survey analyzed with PLS-SEM (SmartPLS 3.0). Sampling and data collection: Snowball convenience sampling among Romanian social media users during summer 2023. The questionnaire was distributed via email to student groups and across common Romanian social platforms (Facebook, Instagram, LinkedIn). A filter ensured inclusion of respondents who had donated online to NGOs/INGOs in the prior 12 months. Final sample: n = 1629 (61.8% women; 66.4% urban; 87.7% high school/tertiary educated; 85.1% aged 18–50). Measures: All constructs measured on 5-point Likert scales (1 = strongly disagree to 5 = strongly agree). Constructs and item sources included: brand reputation of INGOs (BRINGO; Bennett & Gabriel, 2003); sharing fake news about INGOs (SFINGO; Chadwick & Vaccari, 2019); social media usage (SMU; Kumar & Pansari, 2016); perceived risks of online donations to INGOs (PRODINGO; Lang, 2018); attitude toward donating online to INGOs (ADONGO; Treiblmaier & Pollach, 2006); intention to donate online to INGOs (IDONGO; Bilgin & Kethüda, 2022); brand image of INGOs (BIINGO; Bilgin & Kethüda, 2022). Analysis steps: Two-stage PLS-SEM. Measurement model: Item loadings exceeded 0.70; internal consistency and convergent validity were satisfactory (composite reliability > 0.80; AVE > 0.50). Discriminant validity supported via Fornell–Larcker and HTMT criteria. Common method/collinearity checks: All indicator VIFs < 5 (max 3.720 for PRODINGO2). Structural model: Inner VIFs < 5 (max 2.212). Model fit acceptable: SRMR = 0.048 (< 0.08). Explanatory power: R2 = 0.547 for perceived risks (PRODINGO) explained by SFINGO and ADONGO; R2 = 0.562 for intention (IDONGO) explained by PRODINGO, SFINGO, ADONGO, and BIINGO. Hypotheses assessed via bootstrapping with significance tests and confidence intervals.

All 10 hypothesized paths were statistically significant. Key path coefficients (β), t-values, and p-values: - H1 (SMU → SFINGO): β = 0.240, t = 9.938, p < 0.001 (positive, medium). - H2 (BRINGO → SFINGO): β = 0.111, t = 5.183, p < 0.001 (positive, small-to-medium). - H3 (SFINGO → PRODINGO): β = 0.131, t = 6.956, p < 0.001 (positive, small). - H4 (SFINGO → ADONGO): β = 0.170, t = 6.342, p < 0.001 (positive, small-to-medium). - H5 (SFINGO → IDONGO): β = 0.038, t = 2.245, p = 0.025 (positive, very small). - H6 (PRODINGO → IDONGO): β = 0.080, t = 4.479, p < 0.001 (positive, small). - H7 (ADONGO → PRODINGO): β = 0.706, t = 48.524, p < 0.001 (strong positive). - H8 (ADONGO → IDONGO): β = -0.063, t = 2.426, p = 0.016 (negative, very small). - H9 (ADONGO → BIINGO): β = 0.073, t = 2.615, p = 0.009 (positive, very small). - H10 (BIINGO → IDONGO): β = 0.743, t = 50.115, p < 0.001 (very strong positive). Model fit and explanatory power: SRMR = 0.048; R2(PRODINGO) = 0.547; R2(IDONGO) = 0.562. Interpretation highlights: Social media usage and INGO brand reputation increase fake news sharing. Fake news sharing increases perceived risks and influences attitudes toward online donation, with a slight positive direct effect on intention. Attitude strongly increases perceived risks and slightly enhances brand image, but unexpectedly shows a small negative direct effect on intention. Brand image is the strongest predictor of intention to donate online.

The findings support the extended TPB framework by showing how external social media dynamics (usage and brand reputation) shape fake news sharing, which then influences key psychological determinants of donation behavior. Social media usage is a double-edged sword: it supports outreach but increases fake news sharing, which elevates perceived risks and shapes attitudes. Brand reputation also relates positively to fake news sharing, underscoring a complex brand–misinformation interplay. Attitude toward online donations strongly heightens perceived risks and modestly improves brand image; however, its small negative direct effect on intention suggests intervening concerns (e.g., risk perceptions or contextual factors) may dampen conversion from positive attitude to actual intention. Crucially, brand image exerts a very strong positive effect on intention, indicating brand-building and credibility are central levers for driving online donations. Overall, the results clarify how misinformation processes intersect with donor cognition and intention, offering actionable insights for INGOs to mitigate fake news effects and enhance donation outcomes.

This study extends TPB to include fake news sharing as a key external factor influencing online donation intentions to INGOs. Using a large donor sample in Romania and PLS-SEM, the research demonstrates that social media usage and brand reputation fuel fake news sharing, which affects perceived risks, attitudes, and ultimately intention. The strongest direct driver of intention is brand image, while attitude shows a small negative direct effect on intention but a strong positive effect on perceived risks. Contributions include: (1) integrating fake news dynamics within TPB to explain online donation intention; (2) providing empirical evidence of the brand reputation–fake news sharing link in the non-profit context; and (3) highlighting brand image as a critical determinant of intention. Practical implications for INGOs: monitor and counter misinformation, proactively communicate transparent performance and secure donation processes, strengthen brand reputation and image, and craft targeted campaigns to build trust. Future research directions include: probability sampling and multi-country replications; longitudinal designs; comparing INGO vs. local/regional/national NGOs; segmenting by cause area; distinguishing intentional vs. accidental fake news sharing; incorporating donation page features (design, usability, security), technology perceptions (usefulness, ease-of-use), and donor flow states; and examining generational differences.

- Geographic and sampling limits: Single-country study (Romania) with convenience snowball sampling; results may not generalize broadly. - Cross-sectional design: Limits causal inference and may not capture dynamic interactions over time. - Scope of organizations: Focused on international NGOs only. - Cause-agnostic evaluation: Respondents rated their preferred INGOs regardless of specific causes, which may confound cause-specific effects. - Fake news behavior not differentiated: Did not distinguish intentional from accidental sharing. - Potential omitted variables: Did not include donation page features, technology perceptions, or flow experiences that may affect intention.