Table of Contents
Type: Research Paper | Subject: Media Studies | Level: Masters | Word Count: ~3,200 words | Referencing: Harvard
This model research paper was produced by an Essays UK specialist as reference material for learning purposes only. For support in this field, see our media and journalism specialists.
Design and conduct an original piece of primary research investigating audience behaviour, belief, or engagement with digital or social media, drawing on relevant theory from media and communication studies. Submit your findings as a research paper of approximately 3,200 words, following Harvard referencing conventions.
This study investigates the individual-level predictors of misinformation sharing among UK social media users, using an online survey-experiment in which 412 participants rated the accuracy of, and their likelihood of sharing, twelve news headlines (six true, six false) concerning UK political and health topics. Participants also completed a three-item Cognitive Reflection Test (CRT), reported the emotional arousal each headline evoked, and indicated their political self-placement. Overall truth discernment was significant, with participants rating true headlines as more accurate than false headlines, t(411) = 24.60, p < .001, d = 1.21, yet a substantial minority reported willingness to share false headlines they had rated as inaccurate. A hierarchical multiple regression predicting sharing intent for false headlines found perceived accuracy (β = .41), emotional arousal (β = .34), CRT score (β = -.29), and political concordance between headline and respondent ideology (β = .22) as significant predictors, jointly explaining 38% of variance, R² = .38, F(4, 407) = 62.31, p < .001. These findings suggest that analytic reasoning, emotional reactivity, and partisan alignment each independently shape misinformation-sharing intentions, with implications for platform design and media literacy interventions in the UK.
The spread of political and health misinformation on social media has become a central concern for policymakers, platforms, and researchers alike, particularly following high-profile episodes such as the proliferation of false claims during the COVID-19 pandemic and around UK general elections (Ofcom, 2023). A recurring puzzle in this literature is that sharing behaviour and accuracy judgement are often only weakly aligned: survey and experimental studies repeatedly find that a meaningful proportion of social media users report willingness to share content they simultaneously rate as inaccurate, suggesting that the decision to share is not a simple downstream consequence of believing a claim to be true (Pennycook and Rand, 2021).
Explaining this gap between belief and sharing has become a priority for misinformation researchers. Several candidate explanations have been proposed, spanning cognitive, emotional, and social-identity accounts. Cognitive accounts emphasise that inattention, rather than motivated reasoning, drives much low-quality sharing: users are argued to habitually apply insufficient analytic scrutiny to content encountered while scrolling, such that accuracy is simply not salient at the moment of sharing (Pennycook, Epstein, Mosleh, Arechar, Eckles and Rand, 2021). Emotional accounts instead highlight the arousing, often outrage-inducing, quality of much misinformation as a driver of impulsive sharing, independent of accuracy judgement (Martel, Pennycook and Rand, 2020). Social-identity accounts propose that sharing serves partisan or in-group signalling functions, such that content congenial to one’s political identity is shared regardless of, or even because of, its questionable accuracy (Van Bavel and Pereira, 2018).
This study examines all three accounts simultaneously within a single UK sample, asking which individual-difference and headline-level factors independently predict sharing intent for false political and health headlines, after accounting for participants’ own accuracy judgements. Specifically, it tests whether cognitive reflection (as a marker of analytic thinking style), headline-evoked emotional arousal, perceived headline accuracy, and political concordance between headline content and respondent ideology each make independent contributions to sharing intent, and examines overall truth discernment across the sample as a descriptive baseline against which these predictors are interpreted.
The UK provides a distinctive context for this question. Ofcom’s (2023) News Consumption Survey found that social media has overtaken television as the most commonly cited news source for UK adults under 35, while trust in information encountered on social platforms remains substantially lower than trust in broadcast or print news, a combination that leaves a large and growing segment of the population reliant on a channel they themselves regard as comparatively unreliable. The UK’s Online Safety Act 2023 has additionally placed new statutory duties on platforms regarding the systemic risks posed by misinformation, increasing the practical relevance of evidence on which individual-level mechanisms platforms and media literacy campaigns should prioritise when designing interventions aimed at UK users specifically, rather than relying solely on evidence generated in US samples, where the bulk of existing misinformation-sharing research has been conducted (Pennycook and Rand, 2021).
A substantial experimental literature has examined why social media users fail to discriminate reliably between true and false content when deciding whether to share it. Pennycook and Rand’s (2021) integrative review situates most of this evidence within an “inattention-based” account: across numerous studies using headline-rating paradigms similar to the one used here, brief accuracy-focused prompts, simply asking participants to judge the accuracy of a single unrelated headline before continuing, have been found to improve subsequent sharing discernment, implying that the underlying problem is often a failure to consider accuracy at all rather than a deliberate willingness to spread known falsehoods. This account is supported by Pennycook et al.’s (2021) large-scale field experiments on Twitter, in which accuracy-prompt interventions produced modest but reliable improvements in the quality of news shared by real users.
A parallel literature emphasises the emotional properties of misinformation content itself. Misinformation, particularly around politically or morally charged topics, tends to be written in more emotionally evocative language than accurate reporting on the same events, and higher emotional arousal has repeatedly been linked to increased sharing intent independent of perceived accuracy (Martel, Pennycook and Rand, 2020; Vosoughi, Roy and Aral, 2018). Vosoughi, Roy and Aral’s (2018) large-scale analysis of Twitter cascades found that false news stories were retweeted significantly faster and more widely than true stories, with novelty and emotional reaction, particularly surprise and disgust, identified as the strongest correlates of this diffusion advantage, a pattern the authors interpreted as evidence that emotional salience, not perceived truthfulness, was the primary driver of viral spread.
A third strand of research foregrounds political identity and partisan motivated reasoning. Van Bavel and Pereira (2018) argue that political identity can bias both the perceived accuracy of ideologically congenial content and the willingness to share it even where accuracy is doubted, framing sharing partly as an act of social identity expression rather than pure information transmission. UK-specific survey evidence broadly supports this pattern: a YouGov and Full Fact (2022) study found that respondents were substantially more likely to believe, and to report having shared, false claims that aligned with their pre-existing party preference than equivalent claims that did not, though the study did not simultaneously model cognitive or emotional predictors alongside partisanship, leaving open the relative contribution of each mechanism once the others are statistically controlled.
Cognitive reflection, typically measured with the Cognitive Reflection Test (CRT) or its extended variants, has emerged as a further robust individual-difference predictor. Pennycook and Rand (2019) found that higher CRT scores were associated with better discrimination between true and false headlines and, in some samples, with lower self-reported willingness to share false content, a relationship the authors attribute to analytic thinkers being more likely to engage effortful scrutiny rather than relying on intuitive plausibility judgements, a finding broadly consistent with dual-process theories of reasoning (Kahneman, 2011). However, few studies have entered cognitive reflection, emotional arousal, and political concordance into a single model predicting sharing intent within a UK sample, which is the specific gap this study addresses.
A related applied literature has evaluated media literacy interventions designed to strengthen exactly the analytic scrutiny that CRT is thought to index. Guess, Lerner, Lyons, Montgomery, Nyhan, Reifler and Sircar (2020), running a large field experiment with a media literacy intervention in India and the United States, found that a brief tip-based training improved participants’ subsequent ability to discern true from false headlines, with effects persisting, albeit attenuated, several weeks later. UK-based evaluations of similar school- and public-facing media literacy programmes have reported comparable short-term gains in discernment (Full Fact, 2021), but, as with the broader accuracy-prompt literature, these interventions have rarely been evaluated against emotional or identity-based predictors simultaneously, leaving open how much of any observed improvement operates through the cognitive-reflection pathway specifically versus a more general increase in accuracy motivation. Understanding the relative, independent size of the cognitive-reflection effect alongside emotional and partisan predictors, as attempted in the present study, therefore has direct relevance for judging which forms of intervention are likely to be most effective, and for which segment of the sharing-intent gap they are best targeted.
Design and participants. An online cross-sectional survey-experiment was conducted with 412 UK-based adult social media users (Mage = 34.6, SD = 12.1; 53% female, 45% male, 2% other/prefer not to say), recruited via a commercial online panel with quota sampling on age, gender, and UK region to approximate national demographic distributions. All participants reported using at least one major social media platform (Facebook, X/Twitter, Instagram, or TikTok) at least several times per week. Participants provided informed consent and were compensated for their time in line with the panel provider’s standard rates; the study received institutional research ethics approval prior to fieldwork.
Materials. Twelve news headlines concerning UK political and public health topics were constructed for the study, six accurate (verified against contemporaneous reporting) and six fabricated but plausible in style and formatting, modelled on the headline-rating paradigm widely used in misinformation research (Pennycook and Rand, 2019). Headlines were presented in a standardised social-media-post format (headline, source name, thumbnail placeholder) and piloted with an independent sample of 40 respondents to confirm plausibility and to exclude items with floor or ceiling accuracy ratings before use in the main study.
Measures. For each headline, participants rated perceived accuracy (1 = definitely false to 7 = definitely true), the emotional arousal the headline evoked (1 = not at all arousing to 7 = extremely arousing), and their likelihood of sharing the headline on their own social media if they saw it (1 = extremely unlikely to 7 = extremely likely). Cognitive reflection was measured with the standard three-item CRT (Frederick, 2005), scored 0–3 for the number of items answered with the correct, non-intuitive response. Political self-placement was measured on an 11-point left-right scale (0 = strongly left-wing to 10 = strongly right-wing), and a headline-level political concordance variable was derived by coding whether the ideological lean of each false headline’s content matched or opposed each respondent’s self-placement (dichotomised at the scale midpoint).
Analysis. Truth discernment was assessed with a paired-samples t-test comparing mean perceived accuracy ratings across the six true and six false headlines within each participant. Predictors of sharing intent for the false headlines specifically were examined using hierarchical multiple regression, entering perceived accuracy, emotional arousal, CRT score, and political concordance as predictors of mean sharing-intent rating averaged across the six false headlines. All analyses used an alpha level of .05, and regression assumptions, including linearity, homoscedasticity, and absence of severe multicollinearity (all variance inflation factors below 2.0), were checked and satisfied prior to interpretation.
Pilot testing and manipulation checks. The pilot sample of 40 respondents, distinct from the main study sample, rated each candidate headline for accuracy and plausibility on the same scales used in the main study; headlines were retained only where mean accuracy ratings for true items exceeded 5.0 and for false items fell below 3.5, ensuring an adequate spread of perceived plausibility rather than headlines that were trivially easy to classify. A manipulation check embedded in the main survey, asking participants to identify the general topic area of a randomly selected headline they had just rated, was passed by 96% of the final sample, and the sixteen participants who failed this check were excluded from the analyses reported here, leaving the final analytic sample of 412.
Participants showed significant overall truth discernment: mean perceived accuracy was higher for true headlines (M = 5.8, SD = 1.1) than false headlines (M = 3.4, SD = 1.4), t(411) = 24.60, p < .001, d = 1.21. Despite this, mean sharing-intent ratings for false headlines were not negligible (M = 2.9, SD = 1.3 on the 1–7 scale), indicating that a meaningful proportion of participants reported at least some willingness to share content they had, on average, correctly identified as inaccurate.
Table 1 presents the hierarchical regression predicting sharing intent for false headlines. All four predictors entered the final model as significant, independent contributors. Perceived accuracy was the strongest predictor, followed by emotional arousal, CRT score (negatively associated, such that higher cognitive reflection was linked to lower sharing intent), and political concordance.
| Predictor | B | SE B | β | t | p |
|---|---|---|---|---|---|
| Perceived accuracy rating | 0.52 | 0.06 | .41 | 8.67 | <.001 |
| Emotional arousal rating | 0.31 | 0.06 | .34 | 5.17 | <.001 |
| CRT score (0–3) | -0.34 | 0.07 | -.29 | -4.86 | <.001 |
| Political concordance (0/1) | 0.29 | 0.09 | .22 | 3.22 | .001 |
The full model explained a substantial proportion of variance in sharing intent, R² = .38, adjusted R² = .37, F(4, 407) = 62.31, p < .001. Figure 1 illustrates the pattern underlying the CRT effect, showing mean sharing intent for false headlines across participants grouped into low, medium, and high CRT tertiles: sharing intent declined steadily as cognitive reflection increased, consistent with the negative regression coefficient reported above.
Bivariate correlations among the predictors were modest, reducing concern about multicollinearity in the regression model: perceived accuracy and emotional arousal correlated at r = .22, perceived accuracy and CRT score at r = -.18, and emotional arousal and political concordance at r = .15, all p < .01 but well below thresholds that would indicate problematic overlap between predictors. CRT scores themselves were roughly normally distributed across the sample (M = 1.4, SD = 0.9, out of a maximum of 3), with 28% of participants scoring 0, 31% scoring 1, 24% scoring 2, and 17% scoring the maximum of 3, a distribution broadly comparable to that reported in prior UK-adjacent CRT samples (Pennycook and Rand, 2019), supporting the appropriateness of the tertile split used to construct Figure 1.
This study found that UK social media users, on average, discriminated reliably between true and false political and health headlines, yet nonetheless reported meaningful willingness to share headlines they had rated as inaccurate, replicating the belief-sharing gap widely documented in the international literature (Pennycook and Rand, 2021). Crucially, the regression results indicate this gap is not attributable to any single mechanism: perceived accuracy, emotional arousal, cognitive reflection, and political concordance each contributed independently to sharing intent, suggesting that cognitive, emotional, and identity-based accounts of misinformation sharing are complementary rather than competing explanations, at least within this sample.
The strongest predictor, perceived accuracy itself, confirms that belief remains an important, if incomplete, driver of sharing, consistent with a straightforward information-transmission view of sharing behaviour. However, the substantial independent contribution of emotional arousal, second only to perceived accuracy in this model, supports Vosoughi, Roy and Aral’s (2018) argument that emotionally evocative content carries a sharing advantage that operates at least partly independently of, rather than purely through, accuracy judgement. Similarly, the significant negative effect of CRT score, controlling for accuracy and arousal, aligns with Pennycook and Rand’s (2019) account of analytic reasoning as a protective factor against low-quality sharing, extending this finding to a UK sample and to a wider set of simultaneous emotional and identity-based controls than has typically been modelled in a single study.
The significant, if comparatively modest, effect of political concordance offers UK-specific support for Van Bavel and Pereira’s (2018) social-identity account, and is broadly consistent with the partisan asymmetries reported by YouGov and Full Fact (2022) in belief and self-reported sharing of politically congenial misinformation. That this effect remained significant after controlling for perceived accuracy, emotional arousal, and cognitive reflection suggests political identity exerts an influence on sharing intent over and above its role in shaping belief, consistent with the view that sharing can function partly as an act of identity expression rather than pure information transmission.
Several limitations warrant caution in interpreting these findings. Self-reported sharing intent in a survey context may not correspond closely to actual sharing behaviour on real platforms, where social feedback, algorithmic curation, and situational cues introduce further influences not captured here. The twelve headlines used, while piloted for plausibility, cannot represent the full diversity of misinformation content circulating on UK social media, and the cross-sectional design cannot establish the direction of causality between, for example, emotional arousal and sharing intent. Future research combining behavioural sharing data from real platforms with the individual-difference measures used here would help establish whether the present pattern generalises beyond self-report.
The findings also carry implications for platform design and regulation under the UK’s Online Safety Act framework. Because perceived accuracy, emotional arousal, and cognitive reflection each contributed independently to sharing intent, interface-level interventions modelled on the accuracy-prompt approach validated by Pennycook et al. (2021), such as brief friction points inserted before resharing emotionally arousing or unverified content, may be worth trialling by UK platforms as a complement to, rather than a replacement for, downstream content moderation. Because political concordance also remained a significant, independent predictor, however, purely cognitive interventions are unlikely to eliminate partisan-driven sharing entirely; media literacy programmes that explicitly address identity-motivated reasoning, rather than accuracy assessment alone, may be needed to address the full range of mechanisms identified in this study.
This study found that UK social media users’ intentions to share false political and health headlines were independently predicted by perceived accuracy, emotional arousal, cognitive reflection, and political concordance, together explaining over a third of the variance in sharing intent. The findings support a multi-causal view of misinformation sharing in which cognitive inattention, emotional reactivity, and partisan identity each play an independent role, with implications for intervention design: accuracy-prompt nudges, content-level emotional moderation, and efforts to reduce affective political polarisation may each address a genuinely distinct contributing mechanism rather than a single unified cause, and combined interventions addressing more than one pathway simultaneously may prove more effective than any single approach used in isolation.
Need a Model Research Paper Written to Your Exact Brief?
Our 350+ UK-qualified writers deliver referenced model documents from £15 per 250 words, with free plagiarism and AI-detection reports.
You May Also Like
All work is written by human writers. 100% AI free, guaranteed.
100% money back guarantee if you find plagiarism in our work.
COMPANY DETAILS
