How shades of truth and age affect responses to COVID-19 (Mis)information: randomized survey experiment among WhatsApp users in UK and Brazil

The study addresses how individuals of different ages process and respond to COVID-19 misinformation on WhatsApp, particularly across varying ‘shades of truth’ (full falsity, partial falsity, full truth), and whether corrective information from an authoritative source (WHO) mitigates misinformation effects. Motivated by the COVID-19 ‘infodemic’ and widespread misinformation (e.g., false cures like garlic), the research seeks to clarify mixed evidence on age-related susceptibility to misinformation and evaluate the efficacy of corrective messaging during a health crisis. Conducted in the UK and Brazil—contexts with high WhatsApp usage and significant pandemic burden—the study explores three sets of questions: (1) associations between age and misinformation beliefs, perceived credibility, and intention-to-share; (2) effects of (mis)information type on these outcomes; and (3) the impact of WHO-branded corrective information on beliefs, credibility, and sharing intentions. The goal is to inform evidence-based infodemic management and audience-targeted interventions.

The paper synthesizes research on: (1) Age and misinformation susceptibility, noting competing evidence—some studies suggest older adults are more vulnerable due to cognitive decline, source monitoring errors, and limited digital literacy, while others show older adults can leverage knowledge to resist misinformation and are no more susceptible to the illusory truth effect than younger adults. (2) Forms of misinformation: fully false versus partially false/incomplete information, with partial truths potentially more insidious due to mixing accurate and inaccurate elements. (3) Corrective information efficacy: corrections can sometimes backfire but are often effective when combining retractions with alternative explanations or factual elaboration; source credibility can matter in crises, with expert/agency sources improving corrections. This groundwork motivates testing age differences, message ‘shades of truth,’ and WHO-sourced corrective content within a social media context (WhatsApp).

Design: Two randomized online experiments (UK and Brazil) using a 2 (Age: 18–54 vs 55+) × 3 ((Mis)information type: full falsity, partial falsity, full truth) × 2 (Exposure: initial vs corrective) mixed design. Age and (mis)information type were between-subjects; Exposure was within-subjects. Participants: N=1454 (UK n=725; Brazil n=729). Quota sampling via Qualtrics ensured even distribution across stimuli and age groups; inclusion criteria: 18+ and WhatsApp users. Data collection: May 26–June 4 (UK) and June 9 (Brazil), 2020. Ethics approval from Northumbria University; informed consent obtained. Stimuli: Exposure 1 presented WhatsApp-style forwards framed on a smartphone mock-up. All included a constant claimed source (#WHO) and a share request line, with content manipulated to reflect (a) full falsity (completely inaccurate garlic cure claim), (b) partial falsity (mix of accurate and inaccurate elements), or (c) full truth (accurate statement regarding garlic not curing COVID-19). Exposure 2 presented a WHO EPI-WIN “myth busters” infographic debunking the garlic cure claim, including WHO branding. Procedure: Participants first completed COVID-19 knowledge items, then were randomly assigned to one of the three Exposure 1 stimuli for a minimum of 90s (Brazil: M=114.79, SD=51.46; UK: M=109.10, SD=45.10). After measures, all viewed the WHO corrective stimulus for at least 90s (Brazil: M=114.16, SD=60.93; UK: M=113.24, SD=58.69) and repeated the measures. Completion time: Brazil 5–40 min; UK 4–21 min. Measures: (1) COVID-19 knowledge: five true/false items sourced from WHO (e.g., common symptoms, incubation period, transmission, antibiotic ineffectiveness, multiple masks). A knowledge index was computed per country. (2) Misinformation belief: perceived accuracy of “Garlic can cure me of the coronavirus (COVID-19)” on a 5-point scale (1= completely inaccurate, 5= completely accurate). Reported country-level means: UK after misinfo M=2.16 (SD=1.48), after correction M=2.21 (SD=1.49); Brazil after misinfo M=1.96 (SD=1.43), after correction M=1.91 (SD=1.35). (3) Message credibility: three-item scale (accurate, believable, authentic) on 5-point scale (Appelman & Sundar, 2016) after each exposure (UK α=0.95 Exposure 1 M=2.53 SD=1.44; α=0.96 Exposure 2 M=3.27 SD=1.45; Brazil α=0.95 Exposure 1 M=1.95 SD=1.25; α=0.97 Exposure 2 M=2.85 SD=1.53). (4) Intention-to-share: six items (friends, immediate family, extended family, colleagues, strangers, nobody [reverse]) on 5-point likelihood scale after each exposure (UK α=0.88 Exposure 1 M=2.39 SD=1.28; α=0.88 Exposure 2 M=2.85 SD=1.26; Brazil α=0.88 Exposure 1 M=1.93 SD=1.10; α=0.92 Exposure 2 M=2.71 SD=1.36). Analyses: Descriptive statistics and t-tests for profiles/knowledge. General linear models (GLM) with repeated measures tested main effects (Age, (Mis)information type, Exposure) and interactions (two-way, three-way) on misinformation belief, message credibility, and intention-to-share. Bonferroni corrections applied to relevant tests.

Sample and knowledge: UK n=725; Brazil n=729. COVID-19 knowledge was high in UK (M=4.15 SD=0.93) and medium in Brazil (M=3.25 SD=0.67). Older adults had higher knowledge than younger adults in both countries (UK older M=4.43 SD=0.74 vs younger M=3.85 SD=1.01, p<.05; Brazil older M=3.29 SD=0.63 vs younger M=3.19 SD=0.70, p<.05). Misinformation belief (garlic cure): - UK: Main Age effect: younger > older in belief accuracy [F(1,719)=337.14, p<.001, η²=0.32]. (Mis)information type and Exposure main effects ns. Interaction Age×Exposure significant [F(2,719)=7.45, p=.007, η²=0.01]: belief increased among older adults after corrective exposure (backfire). - Brazil: Main Age effect: younger > older [F(1,723)=4.17, p=.042, η²=0.01]. (Mis)information type effect: beliefs stronger in full-truth vs full-falsity conditions [F(2,723)=4.83, p=.008, η²=0.01]. Exposure main effect ns. Interactions: Age×Exposure [F(2,723)=4.92, p=.027, η²=0.01]: older adults’ belief decreased post-correction (trend; p=.033, not significant after Bonferroni). (Mis)information type×Exposure [F(2,723)=5.62, p=.004, η²=0.02]: belief increased in full-truth group after corrective exposure. Perceived credibility: - UK: Main effects: Age (younger > older) [F(1,719)=231.86, p<.001, η²=0.24]; (Mis)information type [F(2,719)=16.12, p<.001, η²=0.04]: full truth > full falsity and partial falsity; Exposure [F(1,719)=171.37, p<.001, η²=0.19]: corrective > initial. Interactions: Age×Exposure [F(2,719)=38.72, p<.001, η²=0.05] and (Mis)information type×Exposure [F(2,719)=14.10, p<.001, η²=0.04]; corrective perceived as more credible across groups, with larger gains among older adults. - Brazil: Main effects: Age (younger > older) [F(1,723)=12.80, p<.001, η²=0.02]; (Mis)information type [F(2,723)=11.65, p<.001, η²=0.03]: full falsity < partial falsity and full truth; Exposure [F(1,723)=218.30, p<.001, η²=0.23]: corrective > initial. Interactions: Age×Exposure [F(2,723)=10.08, p=.002, η²=0.01], (Mis)information type×Exposure [F(2,723)=14.23, p<.001, η²=0.04], and Age×(Mis)information type [F(2,723)=5.55, p=.004, η²=0.02]. Intention-to-share: - UK: Main effects: Age (younger > older) [F(1,719)=488.79, p<.001, η²=0.41]; (Mis)information type [F(2,719)=3.27, p=.039, η²=0.01]: full truth > full falsity; Exposure [F(1,719)=111.91, p<.001, η²=0.14]: corrective > initial. Interactions: Age×Exposure [F(2,719)=25.87, p<.001, η²=0.04] and (Mis)information type×Exposure [F(2,719)=4.35, p=.013, η²=0.01]; both ages more likely to share corrective info, with larger increase among older adults; strongest increases in partial falsity and full truth groups. - Brazil: Main effects: Age (younger > older) [F(1,723)=9.10, p=.003, η²=0.01]; (Mis)information type [F(2,723)=4.49, p=.012, η²=0.01]: full truth > full falsity; Exposure [F(1,723)=229.47, p<.001, η²=0.24]: corrective > initial. Interactions: Age×Exposure [F(2,723)=7.32, p=.007, η²=0.01] (larger effect in younger), (Mis)information type×Exposure [F(2,723)=12.38, p<.001, η²=0.03], and Age×(Mis)information type [F(2,723)=6.08, p=.002, η²=0.02] with younger adults more likely to share full/partial falsity. Overall patterns: Younger adults exhibited higher misinformation belief, perceived credibility, and sharing intentions. Corrective information from WHO consistently increased credibility and intention-to-share accurate information across groups and countries; effects on misinformation belief were mixed, including a potential backfire among older UK adults.

The findings address the research questions by showing that age is a key determinant of how WhatsApp users interpret and act on COVID-19 information. Younger adults (18–54) were more susceptible to believing misinformation, perceived messages as more credible, and showed higher sharing intentions than older adults, despite older adults holding higher COVID-19 knowledge. Exposure to different ‘shades of truth’ mattered: fully true messages garnered higher credibility and sharing intention than fully or partially false messages, indicating audiences can differentiate accuracy. However, partial truths remain problematic for entrenching beliefs and encouraging sharing. Corrective information from the WHO effectively increased message credibility and sharing intentions across groups and countries, and generally reduced misinformation beliefs, though not uniformly; notably, older UK adults exhibited a backfire effect, whereby beliefs in the false garlic claim increased after correction. These results underscore the necessity of tailored, evidence-based corrective strategies, considering age-specific vulnerabilities and potential backfire, and suggest that authoritative, well-branded corrective content (e.g., WHO infographics) can improve credibility and sharing of accurate information.

This study contributes evidence on how age and message ‘shades of truth’ shape beliefs, credibility perceptions, and sharing intentions related to COVID-19 (mis)information on WhatsApp, and demonstrates the overall utility of WHO-sourced corrective information in enhancing credibility and promoting sharing of accurate content. Key contributions include: (1) identifying younger adults as more vulnerable to misinformation belief and more likely to share; (2) documenting that full truth content is rated more credible and more shareable than false content; and (3) revealing potential age-specific backfire risks following corrections among older adults in the UK. Future research should: (a) test broader misinformation types (e.g., conspiracy theories, hoaxes) across modalities; (b) evaluate corrective strategies at scale with diverse, representative samples; (c) investigate mechanisms linking beliefs and credibility to sharing; (d) refine audience segmentation approaches for targeted interventions; and (e) pre-test corrective messages to minimize backfire effects.

- Single misinformation topic (garlic as a cure) limits generalizability across diverse misinformation themes (e.g., conspiracies, policy misinformation). - Stimuli were presented as WhatsApp-style forwards within a survey environment rather than in-platform, limiting ecological validity; ethical constraints prevented real-world dissemination of stimuli. - Participants were exposed to claim content multiple times (stimuli and measurement items), which may have contributed to familiarity/backfire effects, especially among older adults. - Oversampling of adults 55+ means samples are not population-representative of the UK and Brazil. - Cultural and contextual differences were not tested via cross-country statistical comparisons; findings are descriptive across countries.