The Impact of Prebunking Interventions Against Misinformation on Discrimination Ability and Criterion: An IPD Network Meta-Analysis

The Impact of Prebunking Interventions Against Misinformation on Discrimination Ability and Criterion: An IPD Network Meta-Analysis | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article The Impact of Prebunking Interventions Against Misinformation on Discrimination Ability and Criterion: An IPD Network Meta-Analysis Xiaojun Sun, Xuqing Bai, Bizhong Chen, Gengfeng Niu, Peipei Mao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6660774/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Prebunking interventions against misinformation have been widely studied, yet few have effectively distinguished between discrimination ability and discrimination criterion using signal detection theory. This study is the first to systematically analyze the effects of prebunking interventions on discrimination ability and discrimination criterion using network meta-analysis based on individual participant data from 30 independent experiments (N = 30,530). Results indicate that prebunking interventions such as media literacy training, inoculation strategies, and writing letters to elders did not enhance discrimination ability but instead led to stricter discrimination criterion, making individuals more likely to judge information as false. Accuracy prompts, feedback, and bias awareness interventions had no significant impact on either discrimination ability or criterion. In contrast, financial incentives significantly improved discrimination ability without altering the discrimination criterion, thereby avoiding negative spillover effects. Further analysis revealed that after the intervention, males, older individuals, those with higher education, and those with greater analytical thinking showed improved discrimination ability but adopted a stricter criterion. Meanwhile, individuals in Asia applied more lenient criterion, whereas those in Europe and Oceania were more stringent. Extended analysis showed that improvements in discrimination ability became evident after three to four weeks but were accompanied by a stricter discrimination criterion. We emphasize the need for future research to employ data analysis approaches grounded in signal detection theory, consider targeted interventions informed by demographic factors, and conduct long-term follow-ups to evaluate the sustained effectiveness of prebunking interventions. Social science/Psychology/Human behaviour Humanities/Cultural and media studies Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Misinformation has posed significant threats to critical societal issues, including politics, health, and climate. Such misinformation not only disseminates inaccurate content among individuals 1 but also induces negative emotion 2 and irrational behaviors 3 . Moreover, misinformation can exacerbate group polarization 4 , incite social unrest 5 , and even contribute to geopolitical conflicts 6 . Given these severe consequences, combating misinformation is of paramount importance. Researchers across various fields have made considerable efforts to combat misinformation, which can generally be categorized into two types: prebunking interventions and debunking interventions 7 . Debunking interventions, which correct misinformation after exposure (e.g., fact-checking labels), have been extensively studied, and their overall effectiveness has been systematically evaluated in several meta-analyses 8 , 9 . However, the continued influence of misinformation—despite corrections—raises questions about the limitations of this approach, such as delayed exposure, repetition effects, and the first-mover advantage enjoyed by false content 10 . In contrast, prebunking interventions seek to reduce susceptibility to misinformation prior to exposure, using strategies like media literacy 11 , inoculation strategies 12 , and financial incentives 13 . Despite their growing application, the mechanisms underlying their effects remain less well understood, with three key issues yet to be fully addressed. First, the critical question is whether prebunking interventions are actually effective. Even including meta-analytic studies 14 , most of the early studies of prebunking interventions found that the interventions were effective. Nevertheless, with further research, some scholars began to recognize and acknowledge that the effects of these preventive interventions are limited 15 , 16 . This significant shift in perspective can be attributed to a change in how intervention effectiveness is measured. Earlier studies often assessed the effectiveness of prebunking interventions by measuring the accuracy of correctly identifying true and false information 12 , or by calculating the difference in accuracy between true and false information 17 . The main limitation of these measures is that they cannot determine whether the intervention's effectiveness is due to an improvement in discrimination ability or a stricter discrimination criterion 18 . Therefore, researchers have started to advocate for the use of signal detection theory 16 , 18 , 19 to differentiate between individuals' discrimination ability and discrimination criterion as a measure of their ability to judge information 1 , 20 or as an indicator of intervention effectiveness 15 . Second, an important accompanying issue is the potential negative spillover effects of prebunking interventions. Negative spillover effects refer to individuals becoming more likely to judge both true and false information as false, as a consequence of adopting a stricter discrimination criterion 21 . Previous research has shown that interventions, such as media literacy 11 , 22 and inoculation strategies 23 , reduce individuals' trust in false information while simultaneously undermining their trust in true information. Such negative spillovers may discourage the discrimination and dissemination of true information and undermine the original intent of the prebunking interventions. Unfortunately, there is a lack of extensive evidence within the signal detection theory framework to evaluate both the effectiveness of prebunking interventions (discrimination ability) and their potential negative spillover effects (discrimination criterion), which are crucial for combating misinformation. Third, although prebunking interventions have broad applicability, their effectiveness may be influenced by individual characteristics (e.g., age) and study characteristics (e.g., topic). According to dual-process theory 24 , individuals rely on two systems for information processing: System 1, which is fast, intuitive, and automatic, and System 2, which is slow, analytical, and effortful. In prebunking interventions, interventions such as accuracy prompts and feedback are primarily designed to encourage individuals to shift their processing from System 1 to System 2, thereby enhancing the effectiveness of the intervention. However, individual characteristics such as age, gender, education, analytical thinking (Cognitive Reflection Test, CRT), and region may influence the extent to which System 1 and System 2 are engaged in information processing. Gender. Compared to female, male is more likely to engage System 2 in information processing 25 , suggesting that interventions may be more effective for male. However, this could also increase the likelihood of overthinking, making their discrimination criterion stricter 26 . Age. Older individuals often possess higher crystallized intelligence, which makes them more likely to correctly discriminate information after intervention. However, their decision-making may be more cautious, leading to a stricter discrimination criterion 27 , 28 . In contrast, adolescents, whose cognitive control abilities are not fully developed, are more likely to rely on System 1 processing 25 , even after interventions, resulting in lower discrimination ability and a looser discrimination criterion. Education. Higher education levels are typically associated with stronger critical thinking and information evaluation skills, making individuals more likely to engage System 2 for deeper analysis. As a result, those with higher education are better able to distinguish signals from noise after an intervention, leading to improved discrimination ability 26 . Analytical thinking . Analytical thinking (CRT) is a key measure of an individual's tendency for reflective thinking 29 . Compared to individuals with low analytical thinking, those with high analytical thinking are more likely to engage System 2 processing after an intervention, thereby enhancing their discrimination ability. Region. Regional differences may influence individuals' trust mechanisms and discrimination criterion through variations in media environments and social norms, leading to cross-cultural heterogeneity in the effectiveness of prebunking interventions 11 . In addition to individual differences, study characteristics may also affect the effectiveness of prebunking interventions, including measurement phases, topic, images, social media indicators, consistency of information across measurement phases, scoring method, and the balance between true and false information. Measurement phases. The measurement phases capture not only the immediate effects of prebunking interventions but also their long-term impact on sustaining improved discrimination ability over time 30 . Topic. The topic of information materials may influence emotional responses and attentional allocation. In particular, political information is often closely tied to individuals’ preexisting ideological beliefs 31 , which can directly affect the effectiveness of prebunking interventions. Images. The presence of images in information materials may lead to the photo truth effect 32 , where individuals are more likely to believe information that includes images, regardless of its actual accuracy. This effect may influence both the discrimination ability and the discrimination criterion after the intervention. Social media indicators. The inclusion of social media indicators (e.g., likes, shares, or comments) in information materials may shift individuals’ attention away from assessing accuracy and toward social validation 33 . This shift could limit the effectiveness of prebunking interventions in improving discrimination ability. Consistency of information across measurement phases. The consistency of information across measurement phases—such as whether the same content is presented in the immediate posttest and the one-week follow-up—may influence the stability of the intervention effects 34 . When information remains consistent across phases, it cannot prevent participants from verifying its accuracy outside the experimental setting, potentially distorting or obscuring the true effects of the intervention. Scoring Method. This study employs signal detection theory to conduct a secondary analysis of raw data from studies using different scoring methods. The choice of scoring method may impact the accuracy of the analysis. Specifically, if an odd-numbered scale (e.g., a 5-point scale) is used 13 , 30 , 35 , the midpoint cannot be clearly classified as 0 or 1 20 , potentially leading to the results of the calculation of discrimination ability and discrimination criterion. Balance Between True and False Information. In some studies, the number of true and false information items presented in the experimental materials is unbalanced 11 , 35 , 36 . Such imbalances may influence individuals’ discrimination criterion 37 , as they might adjust their judgment strategies based on the relative frequency of true versus false information. To address these limitations, this study introduces a dual methodological innovation. First, this study reconstructs individual-level metadata based on signal detection theory, distinguishing between discrimination ability and discrimination criterion. Second, Bayesian Network Meta-Analysis is performed on individual participant data, integrating 30 independent studies from 15 literatures (total N = 30,530 participants). This approach enables three key advancements: evaluating the effectiveness of prebunking interventions using discrimination ability, assessing the negative spillover effects of prebunking interventions through discrimination criterion, and estimating the moderating effects of demographic variables and study-level characteristics to identify optimal intervention strategies tailored to different populations. Results Study and Participant Characteristics As of 21 October 2024, a total of 16,345 literatures had been screened, as shown in the PRISMA flowchart (Fig. 1 ). After full-text screening, 44 literatures met the inclusion criteria. Of these, 15 literatures provided analyzable individual participant data (IPD), while 29 literatures either did not provide IPD or provided incomplete IPD (with no response from the corresponding authors to email queries) and did not report the relevant outcome measures. Following the principles of signal detection theory, different continuous scoring methods were standardized into categorical scoring approaches. In cases where studies used odd continuous scoring methods, median-based scoring results were recoded as missing and participants with missing values were excluded from the analyses. The overall attrition rate was 27.68%. For the preliminary and further analysis, a total of 14 literatures comprising 25 independent studies and 19,260 participants were included. In the extended analysis, 15 literatures with 30 independent studies and a total of 30,530 participants were analyzed. In the sensitivity analyses, 5 literatures with 8 independent studies and a total of 12,331 participants were analyzed. Sample size details is presented in Table S3 of the supplementary material. Risk of Bias Assessment All included studies exhibited no high risk of bias. Specifically, 33.33% were classified as low risk, while 66.67% were categorized as having some concerns, as shown in supplementary material Figure S1 , Figure S2 and Table S5. Preliminary outcomes The preliminary analysis examined the effects of different prebunking interventions on discrimination ability and discrimination criterion at the posttest. Figure 2 a presents the network plot for discrimination ability and discrimination criterion at the posttest. The network connections were complete, with no isolated nodes observed. Moreover, models with discrimination ability and discrimination criterion as dependent variables satisfy the consistency assumption. Detailed results are provided in Supplementary Table S6. For discrimination ability as the outcome variable, the results indicated that, compared to the control group, financial incentives significantly improved discrimination ability (mean difference [MD] = 0.199; 95% CI, 0.089 to 0.314). The results also show that τ was 0.086 (95% CI, 0.020 to 0.169). For discrimination criterion as the outcome variable, the results indicated that, compared to the control group, inoculation strategies (MD = 0.104; 95% CI, 0.008 to 0.212), media literacy (MD = 0.106; 95% CI, 0.049 to 0.157), and writing letters to elders (MD = 0.106; 95% CI, 0.016 to 0.201) all increased the strictness of individuals' discrimination criteria, making them more likely to classify information as false. The results indicate that τ was 0.044 (95% CI, 0.004 to 0.107). See Fig. 3 for the preliminary analysis results. Further outcomes Building on the preliminary analysis, we further conducted an IPD network meta-regression, incorporating individual-level effect modifiers (gender, age, education, analytical thinking, and region) and study-level effect modifiers (topic, images, social media indicators, consistency of information across measurement phases, scoring method, and balance between true and false information). This further analysis examined the influence of these effect modifiers on discrimination ability and discrimination criterion at the posttest. The network plot for the further analyses were consistent with the preliminary analysis (Fig. 2 a), and the proportion of multiple imputations for missing data is provided in Table S4 of the supplementary materials. Further analysis also found no evidence of model inconsistency for discrimination ability and discrimination criterion based on the node-splitting test results. See Supplementary Table S6 for details. For discrimination ability, the results indicated that gender, age, education, and analytical thinking significantly moderated the intervention effects. Specifically, compared to females, males exhibited stronger discrimination ability after the intervention ( β = 0.082, 95% CI: [0.050, 0.113]). Individuals aged 50 and above showed greater improvement in discrimination ability compared to those aged 18–29 ( β = 0.093, 95% CI: [0.046, 0.140]). Regarding education, individuals with some college/university ( β = 0.102, 95% CI: [0.054, 0.148]) and those with higher degree ( β = 0.149, 95% CI: [0.101, 0.196]) demonstrated significantly stronger discrimination ability than those with high school or less. Additionally, individuals with higher analytical thinking exhibited greater improvements in discrimination ability following the intervention ( β = 0.397, 95% CI: [0.345, 0.449]). The random effects estimate was τ = 0.094 (95% CI: [0.005, 0.284]). For discrimination criterion, gender, age, education, analytical thinking, and region significantly moderated the intervention effects. Males demonstrated significantly stricter discrimination criteria compared to females ( β = 0.034, 95% CI: [0.006, 0.062]). Similarly, individuals aged 50 and above had significantly stricter discrimination criteria compared to those aged 18–29 ( β = 0.124, 95% CI: [0.081, 0.166]). In terms of education, individuals with some college/university ( β = 0.096, 95% CI: [0.051, 0.139]) and those with higher degree ( β = 0.102, 95% CI: [0.062, 0.145]) exhibited stricter discrimination criteria than those with high school or less. Furthermore, higher analytical thinking was associated with stricter discrimination criteria after the intervention ( β = 0.314, 95% CI: [0.266, 0.361]). Regarding regional differences, compared to individuals from Africa, those from Asia exhibited significantly more lenient discrimination criteria after the intervention (β = -0.135, 95% CI: [-0.248, -0.023]), while individuals from Europe ( β = 0.199, 95% CI: [0.072, 0.331]) and Oceania ( β = 0.140, 95% CI: [0.006, 0.273]) demonstrated significantly stricter discrimination criteria. The random effects estimate was τ = 0.054 (95% CI: [0.002, 0.201]). The results are presented in Fig. 4 , with the complete findings provided in Table S8 of the supplementary materials. Extended outcomes The extended analysis further investigated the effects of different prebunking interventions on discrimination ability and discrimination criterion across multiple phases, including the pretest, posttest, one-week posttest (posttest 1w), three-week posttest (posttest 3w), four-week posttest (posttest 4w), and posttest at 7–21 days (posttest 7–21d). Figure 2 b presents the network plot for discrimination ability and discrimination criterion in the extended analysis. The network connections were complete, with no isolated nodes observed. In the extended analysis, the node-splitting test results indicated that the models for discrimination ability and discrimination criterion met the consistency assumption. See Supplementary Table S6 for details. For discrimination ability, compared to the posttest, discrimination ability was higher at posttest 3w (β = 0.154, 95% CI: [0.114, 0.193]) and posttest 4w (β = 0.181, 95% CI: [0.126, 0.238]). The random effects estimate was τ = 0.084 (95% CI: [0.024, 0.155]). For discrimination criterion, compared to the posttest, the discrimination criterion was more lenient at the pretest (β = -0.179, 95% CI: [-0.306, -0.051]), whereas it became stricter at posttest 3w (β = 0.109, 95% CI: [0.073, 0.145]) and posttest 4w (β = 0.168, 95% CI: [0.117, 0.217]). The random effects estimate was τ = 0.039 (95% CI: [0.003, 0.094]). The results are presented in Fig. 5 , with the complete findings provided in Table S9 of the supplementary materials. Sensitivity Analysis Results Given that the further analysis involved multiple imputed datasets, a sensitivity analysis was conducted using non-imputed data to ensure the robustness of the findings. The network diagram is presented in Fig. 2 c. The results indicated that the findings from the multiple imputed datasets were relatively stable. However, additional significant effects were observed in the non-multiple interpolated data. Specifically, individuals whose gender was other (e.g., non-binary) demonstrated higher discrimination ability following the intervention compared to females (β = 0.232, 95% CI: [0.018, 0.441]). Additionally, compared to individuals from Africa, those from North America demonstrated significantly stricter discrimination criteria after the intervention (β = 0.149, 95% CI: [0.008, 0.285]). The complete sensitivity analysis results are provided in the Table S10 of the supplementary materials. Publication Bias Assessment To evaluate potential publication bias, we employed funnel plots and Egger’s regression test. The funnel plot visualization (see Figures S3-S6 for supplementary materials) indicated a relatively symmetrical distribution of effect sizes across studies, suggesting a low likelihood of publication bias. Egger’s regression test results further supported this observation. For the preliminary analysis, neither discrimination ability ( t = 0.06, p = 0.950) nor discrimination criterion ( t = -0.46, p = 0.649) showed statistically significant bias. Similarly, for the extended analysis, discrimination ability ( t = -0.28, p = 0.780) and discrimination criterion ( t = -1.14, p = 0.262) did not exhibit significant regression intercepts. These findings indicate that no substantial publication bias was detected in the included effect sizes. Discussion To date, based on signal detection theory, this study is the first to systematically separate individuals' discrimination ability and discrimination criterion from 30 independent experiments across 15 studies (total sample size: N = 30,530). Additionally, we employed an IPD network meta-analysis to explore the effects of prebunking interventions on discrimination ability and discrimination criterion, while further examining the individual-level effect modifiers such as age, gender, and education. This innovative approach not only addresses previous limitations in effect measurement but also provides solid empirical evidence for developing more precise and effective misinformation prevention strategies. Traditional research generally suggests that prebunking interventions such as media literacy 11 , 22 , 26 , inoculation strategies 28 , 34 , 36 , and writing letters to elders 30 , 38 effectively enhance individuals’ judge ability, primarily based on improvements in overall information judgment scores. However, this approach fails to distinguish whether these improvements stem from an actual increase in discrimination ability or merely an adjustment of the discrimination criterion. Using a signal detection theory framework, our findings indicate that media literacy training, inoculation strategies, and writing letters to elders did not enhance discrimination ability. Instead, their primary effect was to make individuals adopt a more conservative discrimination criterion, leading them to classify information as false more often—negative spillover effect 21 . This aligns with prior research that separately analyzed inoculation strategies, which similarly found no significant improvement in discrimination ability but only a shift in the discrimination criterion 15 , 23 . Moreover, interventions such as accuracy prompts, feedback, and bias awareness did not improve discrimination ability or alter the discrimination criterion, suggesting that previous studies may have overestimated their effectiveness 13 , 35 . Notably, our results reveal that financial incentives significantly enhanced discrimination ability without affecting individuals’ discrimination criterion, highlighting a promising intervention strategy compared to traditional approaches. The effectiveness of financial incentives in enhancing discrimination ability may be attributed to their unique mechanisms related to cognitive resource and motivation. From the cognitive resource perspective, financial incentives may encourage individuals to allocate greater attention and cognitive effort during information processing. Prior research has shown that external rewards can activate neural circuits associated with task performance, such as the executive control functions of the prefrontal cortex 39 , thereby improving individuals’ ability to process complex information. From the motivational perspective, financial incentives can significantly enhance individuals’ engagement and focus on information discrimination. This motivational boost may drive individuals to employ deeper cognitive processing strategies, such as carefully analyzing the source of information, assessing content consistency, and evaluating logical coherence 40 , ultimately leading to improved discrimination ability. Most importantly, financial incentives do not lead to a stricter discrimination criterion, thereby avoiding negative spillover effects. Shifts in discrimination criteria are often associated with individuals’ perceptions of the consequences of errors and their risk-avoidance strategies. For instance, media literacy training may lead individuals to believe that misinformation has severe consequences, prompting them to adopt a stricter discrimination criterion 17 , 22 , while inoculation strategies may familiarize individuals with misinformation, making them more cautious in their judgments 15 , 23 . Fortunately, the core function of financial incentives is to motivate individuals to maximize accuracy in their task performance rather than heighten their vigilance against misinformation 13 . Consequently, financial incentives do not lead to a stricter discrimination criterion but instead enable individuals to more effectively distinguish between true and false information within their existing judgment framework. However, the practical application of financial incentives in real-world settings may be limited due to issues such as feasibility, cost, and ethical concerns. As such, future research could explore alternative strategies that focus on mitigating the negative impact of other prebunking interventions on discrimination criteria. For instance, media literacy should not only teach individuals how to identify misinformation but also emphasize recognizing accurate information. Meanwhile, this study utilized IPD to analyze the effects of various effect modifiers on information discrimination ability and discrimination criterion. Compared to traditional group-level mean comparisons, this approach allows for a more precise estimation of the impact of effect modifiers on discrimination ability and discrimination criterion while accounting for individual differences. The results indicate that individual-level effect modifiers—such as gender, age, education, analytical thinking, and region—significantly influenced both discrimination ability and discrimination criterion. In contrast, study-level effect modifiers—including information topic, the presence of images, the inclusion of social media indicators, the consistency of information across measurement phases, scoring method, and the balance between true and false information—did not show significant effects on either discrimination ability or discrimination criterion. The results indicate that males (compared to females), individuals over 50 (compared to those aged 18–29), those with some college/university or higher degree (compared to those with high school or less), and individuals with higher analytical thinking ability exhibited both greater discrimination ability and more stringent discrimination criterion after the intervention. Regarding regional differences, compared to individuals from Africa, those from Asia exhibited a more lenient discrimination criterion, whereas individuals from Europe and Oceania adopted a more stringent discrimination criterion. These findings not only help explain discrepancies in judgment characteristics across different demographic groups observed in previous research but also provide valuable insights for developing targeted prebunking interventions. Furthermore, this study employed a multi-phase measurement design to systematically examine the trajectory of changes in discrimination ability and discrimination criterion following interventions. Compared to studies relying on single-time-point assessments, this approach offers a significant advantage by capturing both immediate intervention effects and the temporal dynamics of judgment ability and criteria. The results indicate that individuals did not exhibit an immediate improvement in discrimination ability post-intervention but instead showed a stricter discrimination criterion. This pattern remained stable for one to three weeks, with no significant enhancement in discrimination ability. However, after three to four weeks, discrimination ability significantly improved, accompanied by a further tightening of the discrimination criterion. This finding suggests that improvements in information discrimination ability may be a gradual process, underscoring the importance of long-term interventions and repeated assessments. Although this study introduces methodological and conceptual innovations, it has several limitations. First, our analysis was based on signal detection theory to distinguish between discrimination ability and discrimination criterion. However, some studies used odd-numbered scoring methods, making it impossible to convert their data into SDT parameters, which led to the exclusion of some participants’ data. While the proportion of data loss was relatively small, it may still impact the stability and representativeness of the results. Future research should account for data analysis compatibility at the experimental design stage and encourage the adoption of SDT-based methods to enhance comparability across studies. Second, this study employed Bayesian IPD network meta-analysis to examine the moderating effects of demographic and study-level factors. However, sensitivity analyses revealed some instability in results for certain moderator categories (e.g., the "Other" gender category or participants from North America). This suggests that demographic factors may have complex influences on intervention effects, and the current data may be insufficient to fully capture these dynamics. Future research should pay greater attention to the role of demographic variables, ensure broader population representation, and explore more tailored intervention strategies for different groups. Finally, although this study used network meta-regression to assess intervention effects over different time points, most existing studies lack multi-phase measurement data, making it difficult to evaluate the long-term stability and sustained impact of interventions. This limitation highlights an important direction for future research—enhancing longitudinal assessments of intervention effects to determine whether their impact persists over time. Materials and Methods This meta-analysis was conducted following the PRISMA-NMA guidelines 41 and PRISMA-IPD guidelines 42 . The meta-analysis protocol was preregistered on the OSF platform before data coding commenced ( https://osf.io/qpuz8 ). Literature Search Based on prior research in the field of misinformation interventions, the search terms were determined as follows: AB, TI (“fake news” OR “misinformation” OR “disinformation” OR “rumor*” OR “rumour*”) AND (“intervent*” OR “correct*” OR “debunk*” OR “fight*” OR “detect*” OR “refut*” OR “resolution” OR “strateg*” OR “against”). The literature search was conducted in Web of Science, ProQuest, PubMed, and EBSCO PsycINFO. The initial search was performed on January 23, 2024, and a secondary search was conducted on October 21, 2024. To ensure comprehensive inclusion of relevant studies, reference tracking was used to identify additional eligible literatures. Eligibility Criteria Inclusion criteria: Articles published in English. Studies focusing on interventions targeting misinformation. Prebunking interventions 7 : These are not aimed at addressing specific pieces of misinformation (e.g., fact-checking), have a generalizable effect, are effective before misinformation is encountered, and the intervention content differs from the measurement content. Experimental materials must include both true and false information items. The dependent variable must be individuals' judgment of the accuracy of information items. Non-single-arm studies. Exclusion criteria: Lack of individual-level data (IPD) or studies not reporting signal detection theory-related outcomes (AgD). Interventions that do not contribute to a closed network of comparisons. The initial screening and inclusion of studies were conducted independently by the first author, followed by a review of the included studies by two additional researchers to verify their adherence to the inclusion criteria. Random checks were performed on excluded studies to ensure relevant studies were not overlooked. Any disagreements in the inclusion process were resolved through discussion within a team of three researchers. A flowchart of the study inclusion process is provided in Fig. 1 . Dependent variables based on SDT Before reprocessing the dependent variable, we followed the data cleaning procedures outlined in previous studies (e.g., attention checks, manipulation checks, etc.) to organize the data. Subsequently, based on signal detection theory, we recalculated the individual-level raw data from each study to derive participants' discrimination ability and discrimination criterion for judging the accuracy of information. Specifically, following previous studies 1 , 20 , we adjusted different scoring methods for measuring information accuracy (categorical, 4-point scale, 5-point scale, 6-point scale, 7-point scale) to a binary scoring system (0 = false, 1 = true). In cases where studies used odd continuous scoring methods, median-based scoring results were recoded as missing and participants with missing values were excluded from the analyses. The overall attrition rate was 27.68%. To calculate hits, misses, false alarms, and correct rejections, information materials were coded (1 = true, 0 = false), and participants' judgments were also coded (1 = true, 0 = false). Finally, we applied the point estimation method for EVSDT parameters and the parameter correction method 16 to calculate each participant’s discrimination ability and decision criterion. Coding Process After confirming the included studies, two graduate students in psychology independently coded all potential variables, including the intervention measures. For any discrepancies in the coding, consensus was reached through discussion or by re-reading the original texts. See Table 1 for coding characteristics and in Table S2 of the supplementary materials for coding details. Table 1 Coding of Study Characteristics Characteristics Coding Interventions Control group, Accuracy prompts, Bias awareness, Feedback, Financial incentives, Inoculation strategies, Inoculation + feedback, Media literacy, Writing letters to elders Age Under 18, 18–29, 30–49, Over 50 Gender Male, Female, Other, Prefer not to say Education High school or less, Some college/university, Higher degree Analytical Thinking (CRT) Score = (Number of correct answers) / (Total number of questions) Region Africa, Asia, Antarctica, Europe, North America, Oceania, South America, Other Measurement Phases Pretest, Posttest, Posttest 1w, Posttest 2w, Posttest 3w, Posttest 4w, Posttest 7-21d Topics Climate, Health, Politics, No special theme Images Yes, Partially yes, No Social Media Indicators Yes, No Consistency of Information Across Measurement Phases Yes, Partially consistent, Not involved, No Scoring Method Even, Odd Balance Between True and False Information Yes, No Notes : Age: Studies with direct age reporting were reassigned to corresponding age groups. For studies where participants selected their age range, the median value of the category was used. If studies did not report age but participants were college students, they were classified as 18-29. Education: If studies did not report education level but all participants were high school or college students, they were categorized accordingly. Analytical Thinking (CRT): Different versions (3-item, 5-item, 7-item) were used across studies. The CRT score was computed as the proportion of correct responses. Quality Appraisal We assessed the risk of bias for each study using the Cochrane Risk of Bias Assessment 2.0 to ensure that findings were based on reliable and valid evidence 43 . Specifically, we assessed the risk of bias in the included randomized controlled trials (RCTs), along with their respective protocols and trial registrations, across five domains: (1) randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of the outcome, and (5) selection of the reported result. Each study’s risk of bias in these domains was independently rated as “low risk,” “high risk,” or “uncertain.” Notably, since this study conducted a secondary analysis using IPD, the risk of bias related to the selection of the reported results was consistently rated as low 44 . Data Analysis This study employed Bayesian Network Meta-Analysis (NMA) to analyze individual participant data, aiming to compare the effects of different prebunking interventions on discrimination ability and discrimination criterion. The analysis was conducted in RStudio using the ‘multinma’ package 45 , implementing a one-step Bayesian NMA approach for data modeling and analysis (detailed model parameters are provided in Table S6 of the supplementary materials). The analysis consisted of three stages: preliminary analysis, further analysis, and extended analysis. Model convergence and stability were assessed using effective sample size (ESS) and Rhat indicators, while model fit was evaluated using the deviance information criterion (DIC). To examine consistency between direct and indirect evidence, we constructed unrelated mean effects (UME) model or node-split model. In the preliminary analysis, we focusing on the post-intervention effects of prebunking interventions on discrimination ability and discrimination criterion. Further analysis included effect modifiers to examine their impact on discrimination ability and discrimination criterion at the post-intervention measurement. Specifically, missing data in effect modifiers were imputed using the predictive mean matching (PMM), with detailed missing proportions reported in Table S4 of the supplementary materials. The preliminary analysis model was then extended to include 12 effect modifiers (interventions, age, gender, education, analytical thinking, region, measurement phases, topics, images, social media indicators, consistency of information across measurement phases, scoring method, balance between true and false information) at both individual and study levels to test their moderating effects. In the extended analysis, measurement phases were introduced as an additional effect modifier in order to examine how prebunking interventions influenced discrimination ability and discrimination criterion across different measurement phases. Measurement phases were introduced only in the extended analysis because only a subset of included studies involved follow-up assessments at multiple time points. To ensure the robustness of moderation results, sensitivity analyses were conducted by excluding participants with missing moderator variables. Assessment for Publication Bias Compared to traditional meta-analyses, IPD meta-analyses can partially mitigate the issue of publication bias 46 . However, there remains a potential bias between studies that provided IPD and those that did not. To address this issue as much as possible, we calculated the sample size, mean, and standard deviation of discrimination ability and discrimination criterion for different groups in each study. Additionally, we assessed publication bias using funnel plots and Egger’s test. Declarations Data Availability Statement All data supporting the findings of this study are publicly available via the OSF at: https://osf.io/qz9fj/?view_only=70cdfe5ad3104e6bb441e9cee25b82ee. Code Availability Statement All analysis scripts and code used in this study are openly available at the same OSF repository: https://osf.io/qz9fj/?view_only=70cdfe5ad3104e6bb441e9cee25b82ee. References Sultan, M. et al. Susceptibility to online misinformation: A systematic meta-analysis of demographic and psychological factors. Proc. Natl. Acad. Sci. U.S.A. 121, e2409329121 (2024). Featherstone, J. D. & Zhang, J. Feeling angry: the effects of vaccine misinformation and refutational messages on negative emotions and vaccination attitude. J. Health Commun. 25, 692–702 (2020). Bastick, Z. Would you notice if fake news changed your behavior? An experiment on the unconscious effects of disinformation. Computers in Human Behavior 116, 106633 (2021). Wang, Q., Yang, X. & Xi, W. Effects of group arguments on rumor belief and transmission in online communities: An information cascade and group polarization perspective. Information & Management 55, 441–449 (2018). Rulis, M. The Influences of Misinformation on Incidences of Politically Motivated Violence in Europe. The International Journal of Press/Politics (2024) doi: 10.1177/19401612241257873 . Lewandowsky, S., Stritzke, W. G. K., Freund, A. M., Oberauer, K. & Krueger, J. I. Misinformation, disinformation, and violent conflict: From Iraq and the “War on Terror” to future threats to peace. American Psychologist 68, 487–501 (2013). Tay, L. Q., Hurlstone, M. J., Kurz, T. & Ecker, U. K. H. A comparison of prebunking and debunking interventions for implied versus explicit misinformation. BRITISH JOURNAL OF PSYCHOLOGY 113, 591–607 (2022). Martel, C. & Rand, D. G. Fact-checker warning labels are effective even for those who distrust fact-checkers. Nat. Hum. Behav. 8, 1957–1967 (2024). Chan, M. S., Jones, C. R., Hall Jamieson, K. & Albarracín, D. Debunking: A meta-analysis of the psychological efficacy of messages countering misinformation. Psychol Sci 28, 1531–1546 (2017). Walter, N. & Tukachinsky, R. A Meta-Analytic Examination of the Continued Influence of Misinformation in the Face of Correction: How Powerful Is It, Why Does It Happen, and How to Stop It? Communication Research 47, 155–177 (2020). Guess, A. M. et al. A digital media literacy intervention increases discernment between mainstream and false news in the United States and india. PNAS 117, 15536–15545 (2020). Spampatti, T., Hahnel, U., Trutnevyte, E. & Brosch, T. Psychological inoculation strategies to fight climate disinformation across 12 countries. Nat. Hum. Behav. (2023) doi: 10.1038/s41562-023-01736-0 . Rathje, S., Roozenbeek, J., Van Bavel, J. J. & van der Linden, S. Accuracy and social motivations shape judgements of (mis)information. Nat. Hum. Behav. 7, 892–903 (2023). Huang, G., Jia, W. & Yu, W. Media Literacy Interventions Improve Resilience to Misinformation: A Meta-Analytic Investigation of Overall Effect and Moderating Factors. Communication Research 0, 00936502241288103. Modirrousta-Galian, A. & Higham, P. A. Gamified inoculation interventions do not improve discrimination between true and fake news: Reanalyzing existing research with receiver operating characteristic analysis. Journal of Experimental Psychology: General 152, 2411–2437 (2023). Batailler, C., Brannon, S. M., Teas, P. E. & Gawronski, B. A Signal Detection Approach to Understanding the Identification of Fake News. Perspect Psychol Sci 17, 78–98 (2022). Lyons, B., King, A. J. & Kaphingst, K. A. A Health Media Literacy Intervention Increases Skepticism of Both Inaccurate and Accurate Cancer News Among U.S. Adults. Ann Behav Med kaae054 (2024) doi: 10.1093/abm/kaae054 . Higham, P. A., Modirrousta-Galian, A. & Seabrooke, T. Mean rating difference scores are poor measures of discernment: The role of response criteria. Current Opinion in Psychology 56, 101785 (2024). Gawronski, B., Nahon, L. S. & Ng, N. L. A signal-detection framework for misinformation interventions. Nat Hum Behav 1–3 (2024) doi: 10.1038/s41562-024-02021-4 . Pfänder, J. & Altay, S. Spotting false news and doubting true news: a systematic review and meta-analysis of news judgements. Nat Hum Behav doi: 10.1038/s41562-024-02086-1 . van der Meer, T. G. L. A., Hameleers, M. & Ohme, J. Can Fighting Misinformation Have a Negative Spillover Effect? How Warnings for the Threat of Misinformation Can Decrease General News Credibility. Journalism Studies 24, 803–823 (2023). Hoes, E., Aitken, B., Zhang, J., Gackowski, T. & Wojcieszak, M. Prominent misinformation interventions reduce misperceptions but increase scepticism. Nat Hum Behav 28, 1545–1553 (2024). Graham, M. E. et al. Mixed News about the Bad News Game. JC 6, 58 (2023). De Neys, W. & Pennycook, G. Logic, Fast and Slow: Advances in Dual-Process Theorizing. Curr Dir Psychol Sci 28, 503–509 (2019). Sladek, R. M., Bond, M. J. & Phillips, P. A. Age and gender differences in preferences for rational and experiential thinking. Personality and Individual Differences 49, 907–911 (2010). Fendt, M., Nistor, N., Scheibenzuber, C. & Artmann, B. Sourcing against misinformation: Effects of a scalable lateral reading training based on cognitive apprenticeship. Computers in Human Behavior 146, 107820 (2023). Modirrousta-Galian, A., Higham, P. A. & Seabrooke, T. Effects of inductive learning and gamification on news veracity discernment. Journal of Experimental Psychology: Applied 29, 599–619 (2023). Lees, J., Banas, J. A., Linvill, D., Meirick, P. C. & Warren, P. The spot the troll quiz game increases accuracy in discerning between real and inauthentic social media accounts. Pnas Nexus 2, (2023). Pennycook, G. & Rand, D. G. Lazy, not biased: Susceptibility to partisan fake news is better explained by lack of reasoning than by motivated reasoning. Cognition 188, 39–50 (2019). Orosz, G., Farago, L., Paskuj, B. & Kreko, P. Strategies to combat misinformation: Enduring effects of a 15-minute online intervention on critical-thinking adolescents. COMPUTERS IN HUMAN BEHAVIOR 159, (2024). Pennycook, G. & Rand, D. G. Lack of partisan bias in the identification of fake (versus real) news. Trends in Cognitive Sciences 25, 725–726 (2021). Fenn, E., Ramsay, N., Kantner, J., Pezdek, K. & Abed, E. Nonprobative photos increase truth, like, and share judgments in a simulated social media environment. Journal of Applied Research in Memory and Cognition 8, 131–138 (2019). Epstein, Z., Sirlin, N., Arechar, A., Pennycook, G. & Rand, D. The social media context interferes with truth discernment. Sci. Adv. 9, (2023). Roozenbeek, J., Maertens, R., McClanahan, W. & van der Linden, S. Disentangling Item and Testing Effects in Inoculation Research on Online Misinformation: Solomon Revisited. Educational and Psychological Measurement 81, 340–362 (2021). List, J. A., Ramirez, L. M., Seither, J., Unda, J. & Vallejo, B. H. Critical thinking and misinformation vulnerability: experimental evidence from Colombia. PNAS Nexus 3, pgae361 (2024). Leder, J. et al. Feedback exercises boost discernment of misinformation for gamified inoculation interventions. J Exp Psychol Gen 153, 2068–2087 (2024). Meyer-Grant, C. G. & Jakob, M. Ranking tasks in recognition memory: A direct test of the two-high-threshold contrast model. Journal of Experimental Psychology: General (2025) doi: 10.1037/xge0001700 . Orosz, G., Paskuj, B., Faragó, L. & Krekó, P. A prosocial fake news intervention with durable effects. Sci. Rep. 13, (2023). Otis, J. M. et al. Prefrontal cortex output circuits guide reward seeking through divergent cue encoding. Nature 543, 103–107 (2017). Stone, D. & Ziebart, D. A. A model of financial incentive effects in decision making. Organ. Behav. Hum. Decis. Process. 61, 250–261 (1995). Hutton, B. et al. The PRISMA Extension Statement for Reporting of Systematic Reviews Incorporating Network Meta-analyses of Health Care Interventions: Checklist and Explanations. Ann Intern Med 162, 777–784 (2015). Stewart, L. A. et al. Preferred Reporting Items for a Systematic Review and Meta-analysis of Individual Participant Data. JAMA 313, 1657 (2015). Higgins JPT et al. Cochrane Handbook for Systematic Reviews of Interventions version 6.5 (updated August 2024). (2024). Karyotaki, E. et al. Internet-Based Cognitive Behavioral Therapy for Depression. JAMA Psychiatry 78, 361 (2021). Phillippo, D. M. multinma: Bayesian Network Meta-Analysis of Individual and Aggregate Data. 0.7.2 https://doi.org/10.32614/CRAN.package.multinma (2020). Nevitt, S. J., Sudell, M., Cividini, S., Marson, A. G. & Tudur Smith, C. Antiepileptic drug monotherapy for epilepsy: a network meta-analysis of individual participant data. Cochrane Db. Syst. Rev. 2022, (2022). Additional Declarations There is NO Competing Interest. Supplementary Files SupportingInformation.docx Supporting Information Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6660774","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":459862510,"identity":"a91ca85f-67c3-4309-b3b3-fbdde527cadd","order_by":0,"name":"Xiaojun Sun","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDklEQVRIiWNgGAWjYLCCBAYGZiA6wMBgYAETYyZGCxuQMpAgUgsE8BgACSK0yM/IPSbxoMaG3eB2z9cNPwokGPjbzx6TYKiwTmxgP3sAmxaDG3nJBgnH0pgN7pzddrMH6DCJM3lpEgxn0hMbePISsGqRyDF8kMB2mNngRu62GzxALQYSPGYSjG2HExskwE7F4rAcgwMJ/0Bacp7d/APX8g+3FoYbQFsS28Ba2G4jbGnArcXgzBtjg8S+NGbJG2lmt2WA6iXO5BhbJBxLN27jycHusPYcM8kf32yS+W4kP7v55o+NHH/7GcMbH2qsZfvZz2B3GBQkwxg8YBIUVGz41AOBHQH5UTAKRsEoGMkAAB4/V/sxOKR7AAAAAElFTkSuQmCC","orcid":"","institution":"Central China Normal University","correspondingAuthor":true,"prefix":"","firstName":"Xiaojun","middleName":"","lastName":"Sun","suffix":""},{"id":459862511,"identity":"1440f118-4da8-40ae-ba3e-0241773de968","order_by":1,"name":"Xuqing Bai","email":"","orcid":"https://orcid.org/0000-0003-4339-441X","institution":"Central China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Xuqing","middleName":"","lastName":"Bai","suffix":""},{"id":459862512,"identity":"acc10e56-2d6c-4d96-b362-d492d3d55249","order_by":2,"name":"Bizhong Chen","email":"","orcid":"","institution":"Central China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Bizhong","middleName":"","lastName":"Chen","suffix":""},{"id":459862513,"identity":"92f4e816-cf19-4f33-94f0-c6283d6b77cf","order_by":3,"name":"Gengfeng Niu","email":"","orcid":"","institution":"Central China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Gengfeng","middleName":"","lastName":"Niu","suffix":""},{"id":459862514,"identity":"feafa6f6-b70b-4a0d-b222-e1e327a21f9f","order_by":4,"name":"Peipei Mao","email":"","orcid":"","institution":"Central China Normal University","correspondingAuthor":false,"prefix":"","firstName":"Peipei","middleName":"","lastName":"Mao","suffix":""}],"badges":[],"createdAt":"2025-05-14 06:00:43","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-6660774/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6660774/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":83438387,"identity":"4ae3c048-5034-43ab-9a98-0207c11bd211","added_by":"auto","created_at":"2025-05-26 09:00:19","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":98680,"visible":true,"origin":"","legend":"\u003cp\u003eLiterature inclusion flowchart.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/629680079ea9d504c2aeee8d.png"},{"id":83438910,"identity":"35d587cf-f47f-494d-8ff8-8098fdea22a3","added_by":"auto","created_at":"2025-05-26 09:08:19","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":23370,"visible":true,"origin":"","legend":"\u003cp\u003eNetwork Plot. 2a. Network plot of the preliminary and further analysis. 2b. Network plot of the extended analysis. 2c. Network plot of the sensitivity analysis.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/0475d119f0b3fb2a64d81262.png"},{"id":83438911,"identity":"ea293092-5f29-4b7b-a948-ca2fb04642e4","added_by":"auto","created_at":"2025-05-26 09:08:19","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":14658,"visible":true,"origin":"","legend":"\u003cp\u003ePreliminary analysis results.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/52091f208d91d626505f1239.png"},{"id":83438392,"identity":"6d91c656-f08f-44e9-9452-1acf320db953","added_by":"auto","created_at":"2025-05-26 09:00:19","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":40856,"visible":true,"origin":"","legend":"\u003cp\u003eFurther analysis results.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/dae62024691616d6e20c499c.png"},{"id":83438390,"identity":"4b46f049-f3c0-430f-b81c-77f00f24aa63","added_by":"auto","created_at":"2025-05-26 09:00:19","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":69830,"visible":true,"origin":"","legend":"\u003cp\u003eExtended analysis results.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/2a6234220344110a82e5fc3c.png"},{"id":84565942,"identity":"339ed4df-17e4-4293-881d-717fd64f8304","added_by":"auto","created_at":"2025-06-13 14:09:14","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1189790,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/b3b342fe-7eda-4383-8728-98043cb990fa.pdf"},{"id":83438914,"identity":"b7964629-65fd-40f6-958b-62b72b62d027","added_by":"auto","created_at":"2025-05-26 09:08:20","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":680822,"visible":true,"origin":"","legend":"Supporting Information","description":"","filename":"SupportingInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6660774/v1/0010a61b20f0a08b662f5cac.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"The Impact of Prebunking Interventions Against Misinformation on Discrimination Ability and Criterion: An IPD Network Meta-Analysis","fulltext":[{"header":"Introduction","content":"\u003cp\u003eMisinformation has posed significant threats to critical societal issues, including politics, health, and climate. Such misinformation not only disseminates inaccurate content among individuals \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e but also induces negative emotion \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e and irrational behaviors \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Moreover, misinformation can exacerbate group polarization \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e, incite social unrest \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, and even contribute to geopolitical conflicts \u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. Given these severe consequences, combating misinformation is of paramount importance.\u003c/p\u003e \u003cp\u003eResearchers across various fields have made considerable efforts to combat misinformation, which can generally be categorized into two types: prebunking interventions and debunking interventions \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. Debunking interventions, which correct misinformation after exposure (e.g., fact-checking labels), have been extensively studied, and their overall effectiveness has been systematically evaluated in several meta-analyses \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. However, the continued influence of misinformation\u0026mdash;despite corrections\u0026mdash;raises questions about the limitations of this approach, such as delayed exposure, repetition effects, and the first-mover advantage enjoyed by false content \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. In contrast, prebunking interventions seek to reduce susceptibility to misinformation prior to exposure, using strategies like media literacy \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e, inoculation strategies \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, and financial incentives \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Despite their growing application, the mechanisms underlying their effects remain less well understood, with three key issues yet to be fully addressed.\u003c/p\u003e \u003cp\u003eFirst, the critical question is whether prebunking interventions are actually effective. Even including meta-analytic studies \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, most of the early studies of prebunking interventions found that the interventions were effective. Nevertheless, with further research, some scholars began to recognize and acknowledge that the effects of these preventive interventions are limited \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. This significant shift in perspective can be attributed to a change in how intervention effectiveness is measured. Earlier studies often assessed the effectiveness of prebunking interventions by measuring the accuracy of correctly identifying true and false information \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, or by calculating the difference in accuracy between true and false information \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. The main limitation of these measures is that they cannot determine whether the intervention's effectiveness is due to an improvement in discrimination ability or a stricter discrimination criterion \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Therefore, researchers have started to advocate for the use of signal detection theory \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e to differentiate between individuals' discrimination ability and discrimination criterion as a measure of their ability to judge information \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e or as an indicator of intervention effectiveness \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eSecond, an important accompanying issue is the potential negative spillover effects of prebunking interventions. Negative spillover effects refer to individuals becoming more likely to judge both true and false information as false, as a consequence of adopting a stricter discrimination criterion \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Previous research has shown that interventions, such as media literacy \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e and inoculation strategies \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e, reduce individuals' trust in false information while simultaneously undermining their trust in true information. Such negative spillovers may discourage the discrimination and dissemination of true information and undermine the original intent of the prebunking interventions. Unfortunately, there is a lack of extensive evidence within the signal detection theory framework to evaluate both the effectiveness of prebunking interventions (discrimination ability) and their potential negative spillover effects (discrimination criterion), which are crucial for combating misinformation.\u003c/p\u003e \u003cp\u003eThird, although prebunking interventions have broad applicability, their effectiveness may be influenced by individual characteristics (e.g., age) and study characteristics (e.g., topic). According to dual-process theory \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e, individuals rely on two systems for information processing: System 1, which is fast, intuitive, and automatic, and System 2, which is slow, analytical, and effortful. In prebunking interventions, interventions such as accuracy prompts and feedback are primarily designed to encourage individuals to shift their processing from System 1 to System 2, thereby enhancing the effectiveness of the intervention. However, individual characteristics such as age, gender, education, analytical thinking (Cognitive Reflection Test, CRT), and region may influence the extent to which System 1 and System 2 are engaged in information processing. \u003cb\u003eGender.\u003c/b\u003e Compared to female, male is more likely to engage System 2 in information processing \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, suggesting that interventions may be more effective for male. However, this could also increase the likelihood of overthinking, making their discrimination criterion stricter \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. \u003cb\u003eAge.\u003c/b\u003e Older individuals often possess higher crystallized intelligence, which makes them more likely to correctly discriminate information after intervention. However, their decision-making may be more cautious, leading to a stricter discrimination criterion \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. In contrast, adolescents, whose cognitive control abilities are not fully developed, are more likely to rely on System 1 processing \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, even after interventions, resulting in lower discrimination ability and a looser discrimination criterion. \u003cb\u003eEducation.\u003c/b\u003e Higher education levels are typically associated with stronger critical thinking and information evaluation skills, making individuals more likely to engage System 2 for deeper analysis. As a result, those with higher education are better able to distinguish signals from noise after an intervention, leading to improved discrimination ability \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. \u003cb\u003eAnalytical thinking\u003c/b\u003e. Analytical thinking (CRT) is a key measure of an individual's tendency for reflective thinking \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. Compared to individuals with low analytical thinking, those with high analytical thinking are more likely to engage System 2 processing after an intervention, thereby enhancing their discrimination ability. \u003cb\u003eRegion.\u003c/b\u003e Regional differences may influence individuals' trust mechanisms and discrimination criterion through variations in media environments and social norms, leading to cross-cultural heterogeneity in the effectiveness of prebunking interventions\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn addition to individual differences, study characteristics may also affect the effectiveness of prebunking interventions, including measurement phases, topic, images, social media indicators, consistency of information across measurement phases, scoring method, and the balance between true and false information. \u003cb\u003eMeasurement phases.\u003c/b\u003e The measurement phases capture not only the immediate effects of prebunking interventions but also their long-term impact on sustaining improved discrimination ability over time \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. \u003cb\u003eTopic.\u003c/b\u003e The topic of information materials may influence emotional responses and attentional allocation. In particular, political information is often closely tied to individuals\u0026rsquo; preexisting ideological beliefs \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e, which can directly affect the effectiveness of prebunking interventions. \u003cb\u003eImages.\u003c/b\u003e The presence of images in information materials may lead to the photo truth effect \u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e, where individuals are more likely to believe information that includes images, regardless of its actual accuracy. This effect may influence both the discrimination ability and the discrimination criterion after the intervention. \u003cb\u003eSocial media indicators.\u003c/b\u003e The inclusion of social media indicators (e.g., likes, shares, or comments) in information materials may shift individuals\u0026rsquo; attention away from assessing accuracy and toward social validation \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. This shift could limit the effectiveness of prebunking interventions in improving discrimination ability. \u003cb\u003eConsistency of information across measurement phases.\u003c/b\u003e The consistency of information across measurement phases\u0026mdash;such as whether the same content is presented in the immediate posttest and the one-week follow-up\u0026mdash;may influence the stability of the intervention effects \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. When information remains consistent across phases, it cannot prevent participants from verifying its accuracy outside the experimental setting, potentially distorting or obscuring the true effects of the intervention. \u003cb\u003eScoring Method.\u003c/b\u003e This study employs signal detection theory to conduct a secondary analysis of raw data from studies using different scoring methods. The choice of scoring method may impact the accuracy of the analysis. Specifically, if an odd-numbered scale (e.g., a 5-point scale) is used \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e, the midpoint cannot be clearly classified as 0 or 1 \u003csup\u003e20\u003c/sup\u003e, potentially leading to the results of the calculation of discrimination ability and discrimination criterion. \u003cb\u003eBalance Between True and False Information.\u003c/b\u003e In some studies, the number of true and false information items presented in the experimental materials is unbalanced \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Such imbalances may influence individuals\u0026rsquo; discrimination criterion \u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e, as they might adjust their judgment strategies based on the relative frequency of true versus false information.\u003c/p\u003e \u003cp\u003eTo address these limitations, this study introduces a dual methodological innovation. First, this study reconstructs individual-level metadata based on signal detection theory, distinguishing between discrimination ability and discrimination criterion. Second, Bayesian Network Meta-Analysis is performed on individual participant data, integrating 30 independent studies from 15 literatures (total N\u0026thinsp;=\u0026thinsp;30,530 participants). This approach enables three key advancements: evaluating the effectiveness of prebunking interventions using discrimination ability, assessing the negative spillover effects of prebunking interventions through discrimination criterion, and estimating the moderating effects of demographic variables and study-level characteristics to identify optimal intervention strategies tailored to different populations.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy and Participant Characteristics\u003c/h2\u003e \u003cp\u003eAs of 21 October 2024, a total of 16,345 literatures had been screened, as shown in the PRISMA flowchart (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). After full-text screening, 44 literatures met the inclusion criteria. Of these, 15 literatures provided analyzable individual participant data (IPD), while 29 literatures either did not provide IPD or provided incomplete IPD (with no response from the corresponding authors to email queries) and did not report the relevant outcome measures.\u003c/p\u003e \u003cp\u003eFollowing the principles of signal detection theory, different continuous scoring methods were standardized into categorical scoring approaches. In cases where studies used odd continuous scoring methods, median-based scoring results were recoded as missing and participants with missing values were excluded from the analyses. The overall attrition rate was 27.68%.\u003c/p\u003e \u003cp\u003eFor the preliminary and further analysis, a total of 14 literatures comprising 25 independent studies and 19,260 participants were included. In the extended analysis, 15 literatures with 30 independent studies and a total of 30,530 participants were analyzed. In the sensitivity analyses, 5 literatures with 8 independent studies and a total of 12,331 participants were analyzed. Sample size details is presented in Table S3 of the supplementary material.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eRisk of Bias Assessment\u003c/h3\u003e\n\u003cp\u003eAll included studies exhibited no high risk of bias. Specifically, 33.33% were classified as low risk, while 66.67% were categorized as having some concerns, as shown in supplementary material Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Figure S2 and Table S5.\u003c/p\u003e\n\u003ch3\u003ePreliminary outcomes\u003c/h3\u003e\n\u003cp\u003eThe preliminary analysis examined the effects of different prebunking interventions on discrimination ability and discrimination criterion at the posttest. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea presents the network plot for discrimination ability and discrimination criterion at the posttest. The network connections were complete, with no isolated nodes observed. Moreover, models with discrimination ability and discrimination criterion as dependent variables satisfy the consistency assumption. Detailed results are provided in Supplementary Table S6.\u003c/p\u003e \u003cp\u003eFor discrimination ability as the outcome variable, the results indicated that, compared to the control group, financial incentives significantly improved discrimination ability (mean difference [MD]\u0026thinsp;=\u0026thinsp;0.199; 95% CI, 0.089 to 0.314). The results also show that τ was 0.086 (95% CI, 0.020 to 0.169).\u003c/p\u003e \u003cp\u003eFor discrimination criterion as the outcome variable, the results indicated that, compared to the control group, inoculation strategies (MD\u0026thinsp;=\u0026thinsp;0.104; 95% CI, 0.008 to 0.212), media literacy (MD\u0026thinsp;=\u0026thinsp;0.106; 95% CI, 0.049 to 0.157), and writing letters to elders (MD\u0026thinsp;=\u0026thinsp;0.106; 95% CI, 0.016 to 0.201) all increased the strictness of individuals' discrimination criteria, making them more likely to classify information as false. The results indicate that τ was 0.044 (95% CI, 0.004 to 0.107). See Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e for the preliminary analysis results.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eFurther outcomes\u003c/h3\u003e\n\u003cp\u003eBuilding on the preliminary analysis, we further conducted an IPD network meta-regression, incorporating individual-level effect modifiers (gender, age, education, analytical thinking, and region) and study-level effect modifiers (topic, images, social media indicators, consistency of information across measurement phases, scoring method, and balance between true and false information). This further analysis examined the influence of these effect modifiers on discrimination ability and discrimination criterion at the posttest.\u003c/p\u003e \u003cp\u003eThe network plot for the further analyses were consistent with the preliminary analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea), and the proportion of multiple imputations for missing data is provided in Table S4 of the supplementary materials. Further analysis also found no evidence of model inconsistency for discrimination ability and discrimination criterion based on the node-splitting test results. See Supplementary Table S6 for details.\u003c/p\u003e \u003cp\u003eFor discrimination ability, the results indicated that gender, age, education, and analytical thinking significantly moderated the intervention effects. Specifically, compared to females, males exhibited stronger discrimination ability after the intervention (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.082, 95% CI: [0.050, 0.113]). Individuals aged 50 and above showed greater improvement in discrimination ability compared to those aged 18\u0026ndash;29 (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.093, 95% CI: [0.046, 0.140]). Regarding education, individuals with some college/university (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.102, 95% CI: [0.054, 0.148]) and those with higher degree (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.149, 95% CI: [0.101, 0.196]) demonstrated significantly stronger discrimination ability than those with high school or less. Additionally, individuals with higher analytical thinking exhibited greater improvements in discrimination ability following the intervention (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.397, 95% CI: [0.345, 0.449]). The random effects estimate was τ\u0026thinsp;=\u0026thinsp;0.094 (95% CI: [0.005, 0.284]).\u003c/p\u003e \u003cp\u003eFor discrimination criterion, gender, age, education, analytical thinking, and region significantly moderated the intervention effects. Males demonstrated significantly stricter discrimination criteria compared to females (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.034, 95% CI: [0.006, 0.062]). Similarly, individuals aged 50 and above had significantly stricter discrimination criteria compared to those aged 18\u0026ndash;29 (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.124, 95% CI: [0.081, 0.166]). In terms of education, individuals with some college/university (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.096, 95% CI: [0.051, 0.139]) and those with higher degree (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.102, 95% CI: [0.062, 0.145]) exhibited stricter discrimination criteria than those with high school or less. Furthermore, higher analytical thinking was associated with stricter discrimination criteria after the intervention (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.314, 95% CI: [0.266, 0.361]). Regarding regional differences, compared to individuals from Africa, those from Asia exhibited significantly more lenient discrimination criteria after the intervention (β = -0.135, 95% CI: [-0.248, -0.023]), while individuals from Europe (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.199, 95% CI: [0.072, 0.331]) and Oceania (\u003cem\u003eβ\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.140, 95% CI: [0.006, 0.273]) demonstrated significantly stricter discrimination criteria. The random effects estimate was τ\u0026thinsp;=\u0026thinsp;0.054 (95% CI: [0.002, 0.201]).\u003c/p\u003e \u003cp\u003eThe results are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, with the complete findings provided in Table S8 of the supplementary materials.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eExtended outcomes\u003c/h3\u003e\n\u003cp\u003eThe extended analysis further investigated the effects of different prebunking interventions on discrimination ability and discrimination criterion across multiple phases, including the pretest, posttest, one-week posttest (posttest 1w), three-week posttest (posttest 3w), four-week posttest (posttest 4w), and posttest at 7\u0026ndash;21 days (posttest 7\u0026ndash;21d).\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb presents the network plot for discrimination ability and discrimination criterion in the extended analysis. The network connections were complete, with no isolated nodes observed. In the extended analysis, the node-splitting test results indicated that the models for discrimination ability and discrimination criterion met the consistency assumption. See Supplementary Table S6 for details.\u003c/p\u003e \u003cp\u003eFor discrimination ability, compared to the posttest, discrimination ability was higher at posttest 3w (β\u0026thinsp;=\u0026thinsp;0.154, 95% CI: [0.114, 0.193]) and posttest 4w (β\u0026thinsp;=\u0026thinsp;0.181, 95% CI: [0.126, 0.238]). The random effects estimate was τ\u0026thinsp;=\u0026thinsp;0.084 (95% CI: [0.024, 0.155]).\u003c/p\u003e \u003cp\u003eFor discrimination criterion, compared to the posttest, the discrimination criterion was more lenient at the pretest (β = -0.179, 95% CI: [-0.306, -0.051]), whereas it became stricter at posttest 3w (β\u0026thinsp;=\u0026thinsp;0.109, 95% CI: [0.073, 0.145]) and posttest 4w (β\u0026thinsp;=\u0026thinsp;0.168, 95% CI: [0.117, 0.217]). The random effects estimate was τ\u0026thinsp;=\u0026thinsp;0.039 (95% CI: [0.003, 0.094]). The results are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, with the complete findings provided in Table S9 of the supplementary materials.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity Analysis Results\u003c/h2\u003e \u003cp\u003eGiven that the further analysis involved multiple imputed datasets, a sensitivity analysis was conducted using non-imputed data to ensure the robustness of the findings. The network diagram is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec. The results indicated that the findings from the multiple imputed datasets were relatively stable. However, additional significant effects were observed in the non-multiple interpolated data. Specifically, individuals whose gender was other (e.g., non-binary) demonstrated higher discrimination ability following the intervention compared to females (β\u0026thinsp;=\u0026thinsp;0.232, 95% CI: [0.018, 0.441]). Additionally, compared to individuals from Africa, those from North America demonstrated significantly stricter discrimination criteria after the intervention (β\u0026thinsp;=\u0026thinsp;0.149, 95% CI: [0.008, 0.285]). The complete sensitivity analysis results are provided in the Table S10 of the supplementary materials.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePublication Bias Assessment\u003c/h3\u003e\n\u003cp\u003eTo evaluate potential publication bias, we employed funnel plots and Egger\u0026rsquo;s regression test. The funnel plot visualization (see Figures S3-S6 for supplementary materials) indicated a relatively symmetrical distribution of effect sizes across studies, suggesting a low likelihood of publication bias. Egger\u0026rsquo;s regression test results further supported this observation. For the preliminary analysis, neither discrimination ability (\u003cem\u003et\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.06, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.950) nor discrimination criterion (\u003cem\u003et\u003c/em\u003e = -0.46, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.649) showed statistically significant bias. Similarly, for the extended analysis, discrimination ability (\u003cem\u003et\u003c/em\u003e = -0.28, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.780) and discrimination criterion (\u003cem\u003et\u003c/em\u003e = -1.14, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.262) did not exhibit significant regression intercepts. These findings indicate that no substantial publication bias was detected in the included effect sizes.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eTo date, based on signal detection theory, this study is the first to systematically separate individuals' discrimination ability and discrimination criterion from 30 independent experiments across 15 studies (total sample size: N\u0026thinsp;=\u0026thinsp;30,530). Additionally, we employed an IPD network meta-analysis to explore the effects of prebunking interventions on discrimination ability and discrimination criterion, while further examining the individual-level effect modifiers such as age, gender, and education. This innovative approach not only addresses previous limitations in effect measurement but also provides solid empirical evidence for developing more precise and effective misinformation prevention strategies.\u003c/p\u003e \u003cp\u003eTraditional research generally suggests that prebunking interventions such as media literacy \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e,\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, inoculation strategies \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e, and writing letters to elders \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e effectively enhance individuals\u0026rsquo; judge ability, primarily based on improvements in overall information judgment scores. However, this approach fails to distinguish whether these improvements stem from an actual increase in discrimination ability or merely an adjustment of the discrimination criterion. Using a signal detection theory framework, our findings indicate that media literacy training, inoculation strategies, and writing letters to elders did not enhance discrimination ability. Instead, their primary effect was to make individuals adopt a more conservative discrimination criterion, leading them to classify information as false more often\u0026mdash;negative spillover effect \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. This aligns with prior research that separately analyzed inoculation strategies, which similarly found no significant improvement in discrimination ability but only a shift in the discrimination criterion \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Moreover, interventions such as accuracy prompts, feedback, and bias awareness did not improve discrimination ability or alter the discrimination criterion, suggesting that previous studies may have overestimated their effectiveness \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Notably, our results reveal that financial incentives significantly enhanced discrimination ability without affecting individuals\u0026rsquo; discrimination criterion, highlighting a promising intervention strategy compared to traditional approaches.\u003c/p\u003e \u003cp\u003eThe effectiveness of financial incentives in enhancing discrimination ability may be attributed to their unique mechanisms related to cognitive resource and motivation. From the cognitive resource perspective, financial incentives may encourage individuals to allocate greater attention and cognitive effort during information processing. Prior research has shown that external rewards can activate neural circuits associated with task performance, such as the executive control functions of the prefrontal cortex \u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e, thereby improving individuals\u0026rsquo; ability to process complex information. From the motivational perspective, financial incentives can significantly enhance individuals\u0026rsquo; engagement and focus on information discrimination. This motivational boost may drive individuals to employ deeper cognitive processing strategies, such as carefully analyzing the source of information, assessing content consistency, and evaluating logical coherence \u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e, ultimately leading to improved discrimination ability. Most importantly, financial incentives do not lead to a stricter discrimination criterion, thereby avoiding negative spillover effects. Shifts in discrimination criteria are often associated with individuals\u0026rsquo; perceptions of the consequences of errors and their risk-avoidance strategies. For instance, media literacy training may lead individuals to believe that misinformation has severe consequences, prompting them to adopt a stricter discrimination criterion \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e, while inoculation strategies may familiarize individuals with misinformation, making them more cautious in their judgments \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Fortunately, the core function of financial incentives is to motivate individuals to maximize accuracy in their task performance rather than heighten their vigilance against misinformation \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Consequently, financial incentives do not lead to a stricter discrimination criterion but instead enable individuals to more effectively distinguish between true and false information within their existing judgment framework. However, the practical application of financial incentives in real-world settings may be limited due to issues such as feasibility, cost, and ethical concerns. As such, future research could explore alternative strategies that focus on mitigating the negative impact of other prebunking interventions on discrimination criteria. For instance, media literacy should not only teach individuals how to identify misinformation but also emphasize recognizing accurate information.\u003c/p\u003e \u003cp\u003eMeanwhile, this study utilized IPD to analyze the effects of various effect modifiers on information discrimination ability and discrimination criterion. Compared to traditional group-level mean comparisons, this approach allows for a more precise estimation of the impact of effect modifiers on discrimination ability and discrimination criterion while accounting for individual differences. The results indicate that individual-level effect modifiers\u0026mdash;such as gender, age, education, analytical thinking, and region\u0026mdash;significantly influenced both discrimination ability and discrimination criterion. In contrast, study-level effect modifiers\u0026mdash;including information topic, the presence of images, the inclusion of social media indicators, the consistency of information across measurement phases, scoring method, and the balance between true and false information\u0026mdash;did not show significant effects on either discrimination ability or discrimination criterion. The results indicate that males (compared to females), individuals over 50 (compared to those aged 18\u0026ndash;29), those with some college/university or higher degree (compared to those with high school or less), and individuals with higher analytical thinking ability exhibited both greater discrimination ability and more stringent discrimination criterion after the intervention. Regarding regional differences, compared to individuals from Africa, those from Asia exhibited a more lenient discrimination criterion, whereas individuals from Europe and Oceania adopted a more stringent discrimination criterion. These findings not only help explain discrepancies in judgment characteristics across different demographic groups observed in previous research but also provide valuable insights for developing targeted prebunking interventions.\u003c/p\u003e \u003cp\u003eFurthermore, this study employed a multi-phase measurement design to systematically examine the trajectory of changes in discrimination ability and discrimination criterion following interventions. Compared to studies relying on single-time-point assessments, this approach offers a significant advantage by capturing both immediate intervention effects and the temporal dynamics of judgment ability and criteria. The results indicate that individuals did not exhibit an immediate improvement in discrimination ability post-intervention but instead showed a stricter discrimination criterion. This pattern remained stable for one to three weeks, with no significant enhancement in discrimination ability. However, after three to four weeks, discrimination ability significantly improved, accompanied by a further tightening of the discrimination criterion. This finding suggests that improvements in information discrimination ability may be a gradual process, underscoring the importance of long-term interventions and repeated assessments.\u003c/p\u003e \u003cp\u003eAlthough this study introduces methodological and conceptual innovations, it has several limitations. First, our analysis was based on signal detection theory to distinguish between discrimination ability and discrimination criterion. However, some studies used odd-numbered scoring methods, making it impossible to convert their data into SDT parameters, which led to the exclusion of some participants\u0026rsquo; data. While the proportion of data loss was relatively small, it may still impact the stability and representativeness of the results. Future research should account for data analysis compatibility at the experimental design stage and encourage the adoption of SDT-based methods to enhance comparability across studies. Second, this study employed Bayesian IPD network meta-analysis to examine the moderating effects of demographic and study-level factors. However, sensitivity analyses revealed some instability in results for certain moderator categories (e.g., the \"Other\" gender category or participants from North America). This suggests that demographic factors may have complex influences on intervention effects, and the current data may be insufficient to fully capture these dynamics. Future research should pay greater attention to the role of demographic variables, ensure broader population representation, and explore more tailored intervention strategies for different groups. Finally, although this study used network meta-regression to assess intervention effects over different time points, most existing studies lack multi-phase measurement data, making it difficult to evaluate the long-term stability and sustained impact of interventions. This limitation highlights an important direction for future research\u0026mdash;enhancing longitudinal assessments of intervention effects to determine whether their impact persists over time.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003eThis meta-analysis was conducted following the PRISMA-NMA guidelines \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e and PRISMA-IPD guidelines \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. The meta-analysis protocol was preregistered on the OSF platform before data coding commenced (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://osf.io/qpuz8\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n\u003ch2\u003eLiterature Search\u003c/h2\u003e\n\u003cp\u003eBased on prior research in the field of misinformation interventions, the search terms were determined as follows: AB, TI (\u0026ldquo;fake news\u0026rdquo; OR \u0026ldquo;misinformation\u0026rdquo; OR \u0026ldquo;disinformation\u0026rdquo; OR \u0026ldquo;rumor*\u0026rdquo; OR \u0026ldquo;rumour*\u0026rdquo;) AND (\u0026ldquo;intervent*\u0026rdquo; OR \u0026ldquo;correct*\u0026rdquo; OR \u0026ldquo;debunk*\u0026rdquo; OR \u0026ldquo;fight*\u0026rdquo; OR \u0026ldquo;detect*\u0026rdquo; OR \u0026ldquo;refut*\u0026rdquo; OR \u0026ldquo;resolution\u0026rdquo; OR \u0026ldquo;strateg*\u0026rdquo; OR \u0026ldquo;against\u0026rdquo;). The literature search was conducted in Web of Science, ProQuest, PubMed, and EBSCO PsycINFO. The initial search was performed on January 23, 2024, and a secondary search was conducted on October 21, 2024. To ensure comprehensive inclusion of relevant studies, reference tracking was used to identify additional eligible literatures.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n\u003ch2\u003eEligibility Criteria\u003c/h2\u003e\n\u003cp\u003eInclusion criteria:\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eArticles published in English.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eStudies focusing on interventions targeting misinformation.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003ePrebunking interventions \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e: These are not aimed at addressing specific pieces of misinformation (e.g., fact-checking), have a generalizable effect, are effective before misinformation is encountered, and the intervention content differs from the measurement content.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eExperimental materials must include both true and false information items.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eThe dependent variable must be individuals' judgment of the accuracy of information items.\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eNon-single-arm studies.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eExclusion criteria:\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\n\u003cp\u003eLack of individual-level data (IPD) or studies not reporting signal detection theory-related outcomes (AgD).\u003c/p\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003eInterventions that do not contribute to a closed network of comparisons.\u003c/p\u003e\n\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eThe initial screening and inclusion of studies were conducted independently by the first author, followed by a review of the included studies by two additional researchers to verify their adherence to the inclusion criteria. Random checks were performed on excluded studies to ensure relevant studies were not overlooked. Any disagreements in the inclusion process were resolved through discussion within a team of three researchers. A flowchart of the study inclusion process is provided in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\n\u003ch2\u003eDependent variables based on SDT\u003c/h2\u003e\n\u003cp\u003eBefore reprocessing the dependent variable, we followed the data cleaning procedures outlined in previous studies (e.g., attention checks, manipulation checks, etc.) to organize the data. Subsequently, based on signal detection theory, we recalculated the individual-level raw data from each study to derive participants' discrimination ability and discrimination criterion for judging the accuracy of information. Specifically, following previous studies \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e, we adjusted different scoring methods for measuring information accuracy (categorical, 4-point scale, 5-point scale, 6-point scale, 7-point scale) to a binary scoring system (0\u0026thinsp;=\u0026thinsp;false, 1\u0026thinsp;=\u0026thinsp;true). In cases where studies used odd continuous scoring methods, median-based scoring results were recoded as missing and participants with missing values were excluded from the analyses. The overall attrition rate was 27.68%. To calculate hits, misses, false alarms, and correct rejections, information materials were coded (1\u0026thinsp;=\u0026thinsp;true, 0\u0026thinsp;=\u0026thinsp;false), and participants' judgments were also coded (1\u0026thinsp;=\u0026thinsp;true, 0\u0026thinsp;=\u0026thinsp;false). Finally, we applied the point estimation method for EVSDT parameters and the parameter correction method \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e to calculate each participant\u0026rsquo;s discrimination ability and decision criterion.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n\u003ch2\u003eCoding Process\u003c/h2\u003e\n\u003cp\u003eAfter confirming the included studies, two graduate students in psychology independently coded all potential variables, including the intervention measures. For any discrepancies in the coding, consensus was reached through discussion or by re-reading the original texts. See Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e for coding characteristics and in Table S2 of the supplementary materials for coding details.\u003c/p\u003e\n\u003cdiv class=\"gridtable\"\u003e\n\u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n\u003cdiv class=\"colspec\" align=\"left\"\u003e\u0026nbsp;\u003c/div\u003e\n\u003ctable id=\"Tab1\" border=\"1\"\u003e\u003ccaption\u003e\n\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n\u003cdiv class=\"CaptionContent\"\u003e\n\u003cp\u003eCoding of Study Characteristics\u003c/p\u003e\n\u003c/div\u003e\n\u003c/caption\u003e\n\u003cthead\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003cth style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eCharacteristics\u003c/p\u003e\n\u003c/th\u003e\n\u003cth style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eCoding\u003c/p\u003e\n\u003c/th\u003e\n\u003c/tr\u003e\n\u003c/thead\u003e\n\u003ctbody\u003e\n\u003ctr style=\"height: 48px;\"\u003e\n\u003ctd style=\"height: 48px;\" align=\"left\"\u003e\n\u003cp\u003eInterventions\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 48px;\" align=\"left\"\u003e\n\u003cp\u003eControl group, Accuracy prompts, Bias awareness, Feedback, Financial incentives, Inoculation strategies, Inoculation\u0026thinsp;+\u0026thinsp;feedback, Media literacy, Writing letters to elders\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eAge\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eUnder 18, 18\u0026ndash;29, 30\u0026ndash;49, Over 50\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eGender\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eMale, Female, Other, Prefer not to say\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eEducation\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eHigh school or less, Some college/university, Higher degree\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eAnalytical Thinking (CRT)\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eScore = (Number of correct answers) / (Total number of questions)\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eRegion\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eAfrica, Asia, Antarctica, Europe, North America, Oceania, South America, Other\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eMeasurement Phases\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003ePretest, Posttest, Posttest 1w, Posttest 2w, Posttest 3w, Posttest 4w, Posttest 7-21d\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eTopics\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eClimate, Health, Politics, No special theme\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eImages\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eYes, Partially yes, No\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eSocial Media Indicators\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eYes, No\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 48px;\"\u003e\n\u003ctd style=\"height: 48px;\" align=\"left\"\u003e\n\u003cp\u003eConsistency of Information Across Measurement Phases\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 48px;\" align=\"left\"\u003e\n\u003cp\u003eYes, Partially consistent, Not involved, No\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eScoring Method\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eEven, Odd\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003ctr style=\"height: 35px;\"\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eBalance Between True and False Information\u003c/p\u003e\n\u003c/td\u003e\n\u003ctd style=\"height: 35px;\" align=\"left\"\u003e\n\u003cp\u003eYes, No\u003c/p\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tbody\u003e\n\u003ctfoot\u003e\n\u003ctr style=\"height: 13.2871px;\"\u003e\n\u003ctd style=\"height: 13.2871px;\" colspan=\"2\"\u003e\n\u003cp\u003e\u003cstrong\u003eNotes\u003c/strong\u003e:\u003c/p\u003e\n\u003col\u003e\n\u003cli\u003e\u003cstrong\u003eAge:\u003c/strong\u003e Studies with direct age reporting were reassigned to corresponding age groups. For studies where participants selected their age range, the median value of the category was used. If studies did not report age but participants were college students, they were classified as 18-29.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eEducation:\u003c/strong\u003e If studies did not report education level but all participants were high school or college students, they were categorized accordingly.\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eAnalytical Thinking (CRT):\u003c/strong\u003e Different versions (3-item, 5-item, 7-item) were used across studies. The CRT score was computed as the proportion of correct responses.\u003c/li\u003e\n\u003c/ol\u003e\n\u003c/td\u003e\n\u003c/tr\u003e\n\u003c/tfoot\u003e\n\u003c/table\u003e\n\u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\n\u003ch2\u003eQuality Appraisal\u003c/h2\u003e\n\u003cp\u003eWe assessed the risk of bias for each study using the Cochrane Risk of Bias Assessment 2.0 to ensure that findings were based on reliable and valid evidence \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. Specifically, we assessed the risk of bias in the included randomized controlled trials (RCTs), along with their respective protocols and trial registrations, across five domains: (1) randomization process, (2) deviations from intended interventions, (3) missing outcome data, (4) measurement of the outcome, and (5) selection of the reported result. Each study\u0026rsquo;s risk of bias in these domains was independently rated as \u0026ldquo;low risk,\u0026rdquo; \u0026ldquo;high risk,\u0026rdquo; or \u0026ldquo;uncertain.\u0026rdquo; Notably, since this study conducted a secondary analysis using IPD, the risk of bias related to the selection of the reported results was consistently rated as low \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec17\" class=\"Section2\"\u003e\n\u003ch2\u003eData Analysis\u003c/h2\u003e\n\u003cp\u003eThis study employed Bayesian Network Meta-Analysis (NMA) to analyze individual participant data, aiming to compare the effects of different prebunking interventions on discrimination ability and discrimination criterion. The analysis was conducted in RStudio using the \u0026lsquo;multinma\u0026rsquo; package \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e, implementing a one-step Bayesian NMA approach for data modeling and analysis (detailed model parameters are provided in Table S6 of the supplementary materials). The analysis consisted of three stages: preliminary analysis, further analysis, and extended analysis. Model convergence and stability were assessed using effective sample size (ESS) and Rhat indicators, while model fit was evaluated using the deviance information criterion (DIC). To examine consistency between direct and indirect evidence, we constructed unrelated mean effects (UME) model or node-split model.\u003c/p\u003e\n\u003cp\u003eIn the preliminary analysis, we focusing on the post-intervention effects of prebunking interventions on discrimination ability and discrimination criterion. Further analysis included effect modifiers to examine their impact on discrimination ability and discrimination criterion at the post-intervention measurement. Specifically, missing data in effect modifiers were imputed using the predictive mean matching (PMM), with detailed missing proportions reported in Table S4 of the supplementary materials. The preliminary analysis model was then extended to include 12 effect modifiers (interventions, age, gender, education, analytical thinking, region, measurement phases, topics, images, social media indicators, consistency of information across measurement phases, scoring method, balance between true and false information) at both individual and study levels to test their moderating effects. In the extended analysis, measurement phases were introduced as an additional effect modifier in order to examine how prebunking interventions influenced discrimination ability and discrimination criterion across different measurement phases. Measurement phases were introduced only in the extended analysis because only a subset of included studies involved follow-up assessments at multiple time points.\u003c/p\u003e\n\u003cp\u003eTo ensure the robustness of moderation results, sensitivity analyses were conducted by excluding participants with missing moderator variables.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec18\" class=\"Section2\"\u003e\n\u003ch2\u003eAssessment for Publication Bias\u003c/h2\u003e\n\u003cp\u003eCompared to traditional meta-analyses, IPD meta-analyses can partially mitigate the issue of publication bias \u003csup\u003e\u003cspan class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e. However, there remains a potential bias between studies that provided IPD and those that did not. To address this issue as much as possible, we calculated the sample size, mean, and standard deviation of discrimination ability and discrimination criterion for different groups in each study. Additionally, we assessed publication bias using funnel plots and Egger\u0026rsquo;s test.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData Availability Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data supporting the findings of this study are publicly available via the OSF at: https://osf.io/qz9fj/?view_only=70cdfe5ad3104e6bb441e9cee25b82ee.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode Availability Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll analysis scripts and code used in this study are openly available at the same OSF repository: https://osf.io/qz9fj/?view_only=70cdfe5ad3104e6bb441e9cee25b82ee.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSultan, M. \u003cem\u003eet al.\u003c/em\u003e Susceptibility to online misinformation: A systematic meta-analysis of demographic and psychological factors. \u003cem\u003eProc. Natl. Acad. Sci. U.S.A.\u003c/em\u003e 121, e2409329121 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeatherstone, J. D. \u0026amp; Zhang, J. Feeling angry: the effects of vaccine misinformation and refutational messages on negative emotions and vaccination attitude. \u003cem\u003eJ. Health Commun.\u003c/em\u003e 25, 692\u0026ndash;702 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBastick, Z. Would you notice if fake news changed your behavior? An experiment on the unconscious effects of disinformation. \u003cem\u003eComputers in Human Behavior\u003c/em\u003e 116, 106633 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, Q., Yang, X. \u0026amp; Xi, W. Effects of group arguments on rumor belief and transmission in online communities: An information cascade and group polarization perspective. \u003cem\u003eInformation \u0026amp; Management\u003c/em\u003e 55, 441\u0026ndash;449 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRulis, M. The Influences of Misinformation on Incidences of Politically Motivated Violence in Europe. \u003cem\u003eThe International Journal of Press/Politics\u003c/em\u003e (2024) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/19401612241257873\u003c/span\u003e\u003cspan address=\"10.1177/19401612241257873\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLewandowsky, S., Stritzke, W. G. K., Freund, A. M., Oberauer, K. \u0026amp; Krueger, J. I. Misinformation, disinformation, and violent conflict: From Iraq and the \u0026ldquo;War on Terror\u0026rdquo; to future threats to peace. \u003cem\u003eAmerican Psychologist\u003c/em\u003e 68, 487\u0026ndash;501 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTay, L. Q., Hurlstone, M. J., Kurz, T. \u0026amp; Ecker, U. K. H. A comparison of prebunking and debunking interventions for implied versus explicit misinformation. \u003cem\u003eBRITISH JOURNAL OF PSYCHOLOGY\u003c/em\u003e 113, 591\u0026ndash;607 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartel, C. \u0026amp; Rand, D. G. Fact-checker warning labels are effective even for those who distrust fact-checkers. \u003cem\u003eNat. Hum. Behav.\u003c/em\u003e 8, 1957\u0026ndash;1967 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChan, M. S., Jones, C. R., Hall Jamieson, K. \u0026amp; Albarrac\u0026iacute;n, D. Debunking: A meta-analysis of the psychological efficacy of messages countering misinformation. \u003cem\u003ePsychol Sci\u003c/em\u003e 28, 1531\u0026ndash;1546 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWalter, N. \u0026amp; Tukachinsky, R. A Meta-Analytic Examination of the Continued Influence of Misinformation in the Face of Correction: How Powerful Is It, Why Does It Happen, and How to Stop It? \u003cem\u003eCommunication Research\u003c/em\u003e 47, 155\u0026ndash;177 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuess, A. M. \u003cem\u003eet al.\u003c/em\u003e A digital media literacy intervention increases discernment between mainstream and false news in the United States and india. \u003cem\u003ePNAS\u003c/em\u003e 117, 15536\u0026ndash;15545 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpampatti, T., Hahnel, U., Trutnevyte, E. \u0026amp; Brosch, T. Psychological inoculation strategies to fight climate disinformation across 12 countries. \u003cem\u003eNat. Hum. Behav.\u003c/em\u003e (2023) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41562-023-01736-0\u003c/span\u003e\u003cspan address=\"10.1038/s41562-023-01736-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRathje, S., Roozenbeek, J., Van Bavel, J. J. \u0026amp; van der Linden, S. Accuracy and social motivations shape judgements of (mis)information. \u003cem\u003eNat. Hum. Behav.\u003c/em\u003e 7, 892\u0026ndash;903 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, G., Jia, W. \u0026amp; Yu, W. Media Literacy Interventions Improve Resilience to Misinformation: A Meta-Analytic Investigation of Overall Effect and Moderating Factors. \u003cem\u003eCommunication Research\u003c/em\u003e 0, 00936502241288103.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eModirrousta-Galian, A. \u0026amp; Higham, P. A. Gamified inoculation interventions do not improve discrimination between true and fake news: Reanalyzing existing research with receiver operating characteristic analysis. \u003cem\u003eJournal of Experimental Psychology: General\u003c/em\u003e 152, 2411\u0026ndash;2437 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBatailler, C., Brannon, S. M., Teas, P. E. \u0026amp; Gawronski, B. A Signal Detection Approach to Understanding the Identification of Fake News. \u003cem\u003ePerspect Psychol Sci\u003c/em\u003e 17, 78\u0026ndash;98 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLyons, B., King, A. J. \u0026amp; Kaphingst, K. A. A Health Media Literacy Intervention Increases Skepticism of Both Inaccurate and Accurate Cancer News Among U.S. Adults. \u003cem\u003eAnn Behav Med\u003c/em\u003e kaae054 (2024) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/abm/kaae054\u003c/span\u003e\u003cspan address=\"10.1093/abm/kaae054\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHigham, P. A., Modirrousta-Galian, A. \u0026amp; Seabrooke, T. Mean rating difference scores are poor measures of discernment: The role of response criteria. \u003cem\u003eCurrent Opinion in Psychology\u003c/em\u003e 56, 101785 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGawronski, B., Nahon, L. S. \u0026amp; Ng, N. L. A signal-detection framework for misinformation interventions. \u003cem\u003eNat Hum Behav\u003c/em\u003e 1\u0026ndash;3 (2024) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41562-024-02021-4\u003c/span\u003e\u003cspan address=\"10.1038/s41562-024-02021-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePf\u0026auml;nder, J. \u0026amp; Altay, S. Spotting false news and doubting true news: a systematic review and meta-analysis of news judgements. \u003cem\u003eNat Hum Behav\u003c/em\u003e doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41562-024-02086-1\u003c/span\u003e\u003cspan address=\"10.1038/s41562-024-02086-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan der Meer, T. G. L. A., Hameleers, M. \u0026amp; Ohme, J. Can Fighting Misinformation Have a Negative Spillover Effect? How Warnings for the Threat of Misinformation Can Decrease General News Credibility. \u003cem\u003eJournalism Studies\u003c/em\u003e 24, 803\u0026ndash;823 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoes, E., Aitken, B., Zhang, J., Gackowski, T. \u0026amp; Wojcieszak, M. Prominent misinformation interventions reduce misperceptions but increase scepticism. \u003cem\u003eNat Hum Behav\u003c/em\u003e 28, 1545\u0026ndash;1553 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGraham, M. E. \u003cem\u003eet al.\u003c/em\u003e Mixed News about the Bad News Game. \u003cem\u003eJC\u003c/em\u003e 6, 58 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Neys, W. \u0026amp; Pennycook, G. Logic, Fast and Slow: Advances in Dual-Process Theorizing. \u003cem\u003eCurr Dir Psychol Sci\u003c/em\u003e 28, 503\u0026ndash;509 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSladek, R. M., Bond, M. J. \u0026amp; Phillips, P. A. Age and gender differences in preferences for rational and experiential thinking. \u003cem\u003ePersonality and Individual Differences\u003c/em\u003e 49, 907\u0026ndash;911 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFendt, M., Nistor, N., Scheibenzuber, C. \u0026amp; Artmann, B. Sourcing against misinformation: Effects of a scalable lateral reading training based on cognitive apprenticeship. \u003cem\u003eComputers in Human Behavior\u003c/em\u003e 146, 107820 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eModirrousta-Galian, A., Higham, P. A. \u0026amp; Seabrooke, T. Effects of inductive learning and gamification on news veracity discernment. \u003cem\u003eJournal of Experimental Psychology: Applied\u003c/em\u003e 29, 599\u0026ndash;619 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLees, J., Banas, J. A., Linvill, D., Meirick, P. C. \u0026amp; Warren, P. The spot the troll quiz game increases accuracy in discerning between real and inauthentic social media accounts. \u003cem\u003ePnas Nexus\u003c/em\u003e 2, (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePennycook, G. \u0026amp; Rand, D. G. Lazy, not biased: Susceptibility to partisan fake news is better explained by lack of reasoning than by motivated reasoning. \u003cem\u003eCognition\u003c/em\u003e 188, 39\u0026ndash;50 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOrosz, G., Farago, L., Paskuj, B. \u0026amp; Kreko, P. Strategies to combat misinformation: Enduring effects of a 15-minute online intervention on critical-thinking adolescents. \u003cem\u003eCOMPUTERS IN HUMAN BEHAVIOR\u003c/em\u003e 159, (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePennycook, G. \u0026amp; Rand, D. G. Lack of partisan bias in the identification of fake (versus real) news. \u003cem\u003eTrends in Cognitive Sciences\u003c/em\u003e 25, 725\u0026ndash;726 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFenn, E., Ramsay, N., Kantner, J., Pezdek, K. \u0026amp; Abed, E. Nonprobative photos increase truth, like, and share judgments in a simulated social media environment. \u003cem\u003eJournal of Applied Research in Memory and Cognition\u003c/em\u003e 8, 131\u0026ndash;138 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEpstein, Z., Sirlin, N., Arechar, A., Pennycook, G. \u0026amp; Rand, D. The social media context interferes with truth discernment. \u003cem\u003eSci. Adv.\u003c/em\u003e 9, (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoozenbeek, J., Maertens, R., McClanahan, W. \u0026amp; van der Linden, S. Disentangling Item and Testing Effects in Inoculation Research on Online Misinformation: Solomon Revisited. \u003cem\u003eEducational and Psychological Measurement\u003c/em\u003e 81, 340\u0026ndash;362 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eList, J. A., Ramirez, L. M., Seither, J., Unda, J. \u0026amp; Vallejo, B. H. Critical thinking and misinformation vulnerability: experimental evidence from Colombia. \u003cem\u003ePNAS Nexus\u003c/em\u003e 3, pgae361 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeder, J. \u003cem\u003eet al.\u003c/em\u003e Feedback exercises boost discernment of misinformation for gamified inoculation interventions. \u003cem\u003eJ Exp Psychol Gen\u003c/em\u003e 153, 2068\u0026ndash;2087 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeyer-Grant, C. G. \u0026amp; Jakob, M. Ranking tasks in recognition memory: A direct test of the two-high-threshold contrast model. \u003cem\u003eJournal of Experimental Psychology: General\u003c/em\u003e (2025) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1037/xge0001700\u003c/span\u003e\u003cspan address=\"10.1037/xge0001700\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOrosz, G., Paskuj, B., Farag\u0026oacute;, L. \u0026amp; Krek\u0026oacute;, P. A prosocial fake news intervention with durable effects. \u003cem\u003eSci. Rep.\u003c/em\u003e 13, (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOtis, J. M. \u003cem\u003eet al.\u003c/em\u003e Prefrontal cortex output circuits guide reward seeking through divergent cue encoding. \u003cem\u003eNature\u003c/em\u003e 543, 103\u0026ndash;107 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStone, D. \u0026amp; Ziebart, D. A. A model of financial incentive effects in decision making. \u003cem\u003eOrgan. Behav. Hum. Decis. Process.\u003c/em\u003e 61, 250\u0026ndash;261 (1995).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHutton, B. \u003cem\u003eet al.\u003c/em\u003e The PRISMA Extension Statement for Reporting of Systematic Reviews Incorporating Network Meta-analyses of Health Care Interventions: Checklist and Explanations. \u003cem\u003eAnn Intern Med\u003c/em\u003e 162, 777\u0026ndash;784 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStewart, L. A. \u003cem\u003eet al.\u003c/em\u003e Preferred Reporting Items for a Systematic Review and Meta-analysis of Individual Participant Data. \u003cem\u003eJAMA\u003c/em\u003e 313, 1657 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHiggins JPT \u003cem\u003eet al.\u003c/em\u003e Cochrane Handbook for Systematic Reviews of Interventions version 6.5 (updated August 2024). (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaryotaki, E. \u003cem\u003eet al.\u003c/em\u003e Internet-Based Cognitive Behavioral Therapy for Depression. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e 78, 361 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePhillippo, D. M. multinma: Bayesian Network Meta-Analysis of Individual and Aggregate Data. 0.7.2 \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.32614/CRAN.package.multinma\u003c/span\u003e\u003cspan address=\"10.32614/CRAN.package.multinma\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNevitt, S. J., Sudell, M., Cividini, S., Marson, A. G. \u0026amp; Tudur Smith, C. Antiepileptic drug monotherapy for epilepsy: a network meta-analysis of individual participant data. \u003cem\u003eCochrane Db. Syst. Rev.\u003c/em\u003e 2022, (2022).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6660774/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6660774/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePrebunking interventions against misinformation have been widely studied, yet few have effectively distinguished between discrimination ability and discrimination criterion using signal detection theory. This study is the first to systematically analyze the effects of prebunking interventions on discrimination ability and discrimination criterion using network meta-analysis based on individual participant data from 30 independent experiments (N\u0026thinsp;=\u0026thinsp;30,530). Results indicate that prebunking interventions such as media literacy training, inoculation strategies, and writing letters to elders did not enhance discrimination ability but instead led to stricter discrimination criterion, making individuals more likely to judge information as false. Accuracy prompts, feedback, and bias awareness interventions had no significant impact on either discrimination ability or criterion. In contrast, financial incentives significantly improved discrimination ability without altering the discrimination criterion, thereby avoiding negative spillover effects. Further analysis revealed that after the intervention, males, older individuals, those with higher education, and those with greater analytical thinking showed improved discrimination ability but adopted a stricter criterion. Meanwhile, individuals in Asia applied more lenient criterion, whereas those in Europe and Oceania were more stringent. Extended analysis showed that improvements in discrimination ability became evident after three to four weeks but were accompanied by a stricter discrimination criterion. We emphasize the need for future research to employ data analysis approaches grounded in signal detection theory, consider targeted interventions informed by demographic factors, and conduct long-term follow-ups to evaluate the sustained effectiveness of prebunking interventions.\u003c/p\u003e","manuscriptTitle":"The Impact of Prebunking Interventions Against Misinformation on Discrimination Ability and Criterion: An IPD Network Meta-Analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-26 09:00:15","doi":"10.21203/rs.3.rs-6660774/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4d5777f3-35d4-459e-9bc8-076f3de48b5d","owner":[],"postedDate":"May 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":48850297,"name":"Social science/Psychology/Human behaviour"},{"id":48850298,"name":"Humanities/Cultural and media studies"}],"tags":[],"updatedAt":"2025-06-13T14:01:07+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-26 09:00:15","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6660774","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6660774","identity":"rs-6660774","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}