It is important to understand how persuasive the most-cited climate change messaging strategies are. In five replication studies, we found limited evidence of persuasive effects of three highly cited strategies (N=3,216). We then conducted a registered report megastudy (N=13,544) testing the effects of the 10 most-cited climate change messaging strategies on Americans’ pro-environmental attitudes and behavior. Six messages significantly affected multiple preregistered attitudes, with effects ranging from one to four percentage points. Persuasiveness varied little across party lines, inconsistent with theories predicting heterogeneous effects for targeted messages. No message increased pro-environmental donations, suggesting costly behaviors are difficult to influence with messaging alone. Inference of mechanisms driving effects was limited as the most impactful messages influenced multiple mediating variables. Taken together, these results identify several persuasive strategies, while also highlighting the limits of short-form messages for increasing Americans’ support for action to address climate change.

Affective polarization – and especially dislike of opposing partisans – is a source of major concern. Consequently, there is great interest in identifying effective approaches for depolarization. Here, we ask whether recent advances in artificial intelligence (AI), namely the development of conversational large language models, can be leveraged to reduce affective polarization. Prior work suggests that such AI-powered chatbots can be powerful tools for changing factual beliefs. However, it remains unclear whether dialogues with an AI chatbot can reduce feelings of animosity towards members of one’s political outgroup. To answer this question, we assign N = 3,874 US-based participants to one of three treatments in which an AI chatbot tries to reduce affective polarization, to a control where they have an irrelevant dialogue with the AI, or to a benchmark condition consisting of the top-performing non-AI intervention identified in a recent mega-study. We find that all of the human-AI dialogue treatments significantly reduce animosity towards opposing partisans and discomfort with having opposing partisans as neighbors or close friends, with the most effective AI treatment achieving treatment effects roughly 70% the size of the effect of the top-performing non-AI intervention benchmark. In addition, we find that the human-AI dialogues increase the likelihood that individuals reciprocate social ties with hypothetical social media profiles associated with their political outgroup by approximately five percentage points, performing at least as well as the non-AI intervention benchmark. We find that supplying the AI chatbot with specific facts about the outgroup increases effectiveness. These findings indicate that providing factual information is an important route to depolarization, and that AI chatbots offer an effective approach for reducing affective polarization that has the potential to be deployed at scale online.

Digital, self-guided, single-session interventions (SSIs) offer a structured psychological intervention within one interaction. We crowdsourced 66 diverse 10-minute SSIs for depression and selected 11 for testing in a pre-registered experiment (ClinicalTrials.gov ID: NCT06856668). American adults (N = 7,505) experiencing elevated depressive symptoms were recruited online and randomly assigned to one of the 11 crowdsourced SSIs, a previously-validated active comparison SSI, or a control without intervention content. Nearly all SSIs boosted agency and hope for improvement immediately after completion (ds ≤ 0.37). However, only two SSIs significantly reduced depression at four-week follow-up (ds = 0.14 and 0.15). Unexpectedly, some SSIs may have decreased readiness to change at four weeks (ds ≤ 0.14). The most successful SSIs provided memorable, engaging, and actionable guidance on a skill that directly addressed users’ struggles. Future work should aim to leverage SSIs’ short-term gains to promote sustained behavior change or service engagement.

Changing behavior is a central challenge in domains ranging from health and finance to voting and education. Drawing on decades of research from psychology and related fields, this review synthesizes high-quality evidence-particularly randomized controlled trials with objective behavioral outcomes-to identify effective strategies for changing behavior. We propose a unifying three-phase framework that conceptualizes behavior change as a process involving: (a) building motivation, (b) following through, and (c) forming durable habits. For each phase, we identify key psychological challenges and review interventions designed to overcome them. By organizing a large and fragmented literature within a coherent framework, this review aims to clarify when and why behavior change interventions succeed and to inform more effective, behaviorally informed policies and practices.

The emergence of large language models (LLMs) has made it possible for generative artificial intelligence (AI) to tackle many higher-order cognitive tasks, with critical implications for industry, government, and labor markets. Here, we investigate whether existing, openly-available LLMs can be used to create messages capable of influencing humans' political attitudes. Across three pre-registered experiments (total N = 4829), participants who read persuasive messages generated by LLMs showed significantly more attitude change across a range of policies - including polarized policies, like an assault weapons ban, a carbon tax, and a paid parental-leave program - relative to control condition participants who read a neutral message. Overall, LLM-generated messages were similarly effective in influencing policy attitudes as messages crafted by lay humans. Participants' reported perceptions of the authors of the persuasive messages suggest these effects occurred through somewhat distinct causal pathways. While the persuasiveness of LLM-generated messages was associated with perceptions that the author used more facts, evidence, logical reasoning, and a dispassionate voice, the persuasiveness of human-generated messages was associated with perceptions of the author as unique and original. These results demonstrate that recent developments in AI make it possible to create politically persuasive messages quickly, cheaply, and at massive scale.

In the United States, in nearly all cases, one must register in order to vote—yet, a substantial portion of the eligible electorate remains unregistered. Despite this, relatively little is known about how to increase the likelihood that a voter registers. Here, we tested the impact of 10 expert-crowdsourced, theoretically-based psychological interventions on a sample of eligible, yet unregistered, U.S. voters ahead of the 2024 presidential election (N = 12,896). Eight of the interventions increased intentions to vote, and five led individuals to click on the voter registration website. Escalating Commitment, which sequentially employed several social pressure strategies, was the strongest intervention across these outcomes. However, none of the interventions had a significant effect on actual voter registration or voter turnout. The results highlight a substantial disconnect between voters’ intentions and their ultimate behaviors. We discuss potential structural and psychological barriers that undermine the translation of intent into action.

In the United States, in nearly all cases, one must register in order to vote—yet, a substantial portion of the eligible electorate remains unregistered. Despite this, relatively little is known about how to increase the likelihood that a voter registers. Here, we tested the impact of 10 expert-crowdsourced, theoretically-based psychological interventions on a sample of eligible, yet unregistered, U.S. voters ahead of the 2024 presidential election (N = 12,896). Eight of the interventions increased intentions to vote, and five led individuals to click on the voter registration website. Escalating Commitment, which sequentially employed several social pressure strategies, was the strongest intervention across these outcomes. However, none of the interventions had a significant effect on actual voter registration or voter turnout. The results highlight a substantial disconnect between voters’ intentions and their ultimate behaviors. We discuss potential structural and psychological barriers that undermine the translation of intent into action.

Recent research has identified several effective strategies for reducing Americans' animosity toward supporters of opposing political parties. However, whether these strategies can durably reduce partisan animosity in a scalable manner and in everyday life remains unclear. We bridge the gap between prior research and useful application by assessing whether exposure to multiple, scalable treatments that portray inparty and outparty members interacting positively, receiving accurate information about one another's views, and learning about cross-party similarities can a) durably shift partisans' sentiments and b) influence partisans' sentiments toward specific, personally known others encountered in everyday life-not only general, hypothetical, or one-off rival partisans. In a longitudinal survey experiment, we find that exposure to three brief, scalable treatments over a week reduces partisan animosity, with effects persisting for at least a month. Moreover, the treatments durably ameliorate animosity toward both general outparty members and a personally known outparty member, specified prior to the treatments. These findings suggest promising avenues for redressing social divisions in real-world contexts.

The emergence of large language models (LLMs) has made it possible for generative artificial intelligence (AI) to tackle many higher-order cognitive tasks, with critical implications for industry, government, and labor markets in the U.S. and globally. Here, we investigate whether existing, openly-available LLMs can be used to create messages capable of influencing humans’ political attitudes. Across three pre-registered experiments (total N = 4,829), we find consistent evidence that assigning participants to read persuasive messages generated by LLMs can lead to attitude change across a range of policies, including highly polarized policies, such as an assault weapons ban, a carbon tax, and a paid parental-leave program. Overall, we found LLM-generated messages were similarly effective in influencing policy attitudes as were messages crafted by lay humans. These results demonstrate that recent developments in AI make it possible to create politically persuasive messages cheaply and at massive scale.

Discrimination in the evaluation of others is a key cause of social inequality around the world. However, relatively little is known about psychological interventions that can be used to prevent biased evaluations. The limited evidence that exists on these strategies is spread across many methods and populations, making it difficult to generate reliable best practices that can be effective across contexts. In the present work, we held a research contest to solicit interventions with the goal of reducing discrimination based on physical attractiveness using a hypothetical admissions task. Thirty interventions were tested across four rounds of data collection (total N > 20,000). Using a signal detection theory approach to evaluate interventions, we identified two interventions that reduced discrimination by lessening both decision noise and decision bias, while two other interventions reduced overall discrimination by only lessening noise or bias. The most effective interventions largely provided concrete strategies that directed participants' attention toward decision-relevant criteria and away from socially biasing information, though the fact that very similar interventions produced differing effects on discrimination suggests certain key characteristics that are needed for manipulations to reliably impact judgment. The effects of these four interventions on decision bias, noise, or both also replicated in a different discrimination domain, political affiliation, and generalized to populations with self-reported hiring experience. Results of the contest for decreasing attractiveness-based favoritism suggest that identifying effective routes for changing discriminatory behavior is a challenge and that greater investment is needed to develop impactful, flexible, and scalable strategies for reducing discrimination. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

Undemocratic practices, such as voter suppression and election interference, threaten democracies worldwide. Across four studies (N = 4,350), we find that informational and motivational factors drive Americans' support for such practices. Partisans drastically overestimate how much opponents support undemocratic practices, which decreases people's willingness to defend democracy themselves (S1-S2). One remedy for this dynamic is to inform people about the extent to which their rivals actually support democracy, but in polarized contexts, people are incurious about the true beliefs of outpartisans. To address this, we test a new method for improving democratic attitudes-changing beliefs about cross-party empathy. Empathizing across disagreements can improve connections and boost persuasion. When people learn about these valued consequences of empathic engagement, their curiosity about outpartisans increases (S3), and they choose to learn about opponents' support for democracy, which reduces their own support for undemocratic practices and politicians (S4). Our findings suggest that fostering support for democracy requires not just informational strategies but also motivational ones. The power of our combined approach comes in that-instead of presenting people with information about outpartisans-it induces them to seek out that information themselves. Together, these results highlight how cross-party empathy beliefs can increase people's curiosity about those they disagree with and disrupt processes of political escalation.

Megastudies are experiments that test many treatments simultaneously using the same outcomes, control condition and sample, and are a promising tool that can provide unique insights relative to other research designs. We identify five critical decisions in designing megastudies and suggest potential solutions for each.

Concern over democratic erosion has led to a proliferation of proposed interventions to strengthen democratic attitudes in the United States. Resource constraints, however, prevent implementing all proposed interventions. One approach to identify promising interventions entails leveraging domain experts, who have knowledge regarding a given field, to forecast the effectiveness of candidate interventions. We recruit experts who develop general knowledge about a social problem (academics), experts who directly intervene on the problem (practitioners), and nonexperts from the public to forecast the effectiveness of interventions to reduce partisan animosity, support for undemocratic practices, and support for partisan violence. Comparing 14,076 forecasts submitted by 1,181 forecasters against the results of a megaexperiment (n = 32,059) that tested 75 hypothesized effects of interventions, we find that both types of experts outperformed members of the public, though experts differed in how they were accurate. While academics' predictions were more specific (i.e., they identified a larger proportion of ineffective interventions and had fewer false-positive forecasts), practitioners' predictions were more sensitive (i.e., they identified a larger proportion of effective interventions and had fewer false-negative forecasts). Consistent with this, practitioners were better at predicting best-performing interventions, while academics were superior in predicting which interventions performed worst. Our paper highlights the importance of differentiating types of experts and types of accuracy. We conclude by discussing factors that affect whether sensitive or specific forecasters are preferable, such as the relative cost of false positives and negatives and the expected rate of intervention success.

Scholars warn that partisan divisions in the mass public threaten the health of American democracy. We conducted a megastudy ( n = 32,059 participants) testing 25 treatments designed by academics and practitioners to reduce Americans’ partisan animosity and antidemocratic attitudes. We find that many treatments reduced partisan animosity, most strongly by highlighting relatable sympathetic individuals with different political beliefs or by emphasizing common identities shared by rival partisans. We also identify several treatments that reduced support for undemocratic practices—most strongly by correcting misperceptions of rival partisans’ views or highlighting the threat of democratic collapse—which shows that antidemocratic attitudes are not intractable. Taken together, the study’s findings identify promising general strategies for reducing partisan division and improving democratic attitudes, shedding theoretical light on challenges facing American democracy.

Effectively addressing public health crises like the COVID-19 pandemic requires persuading the mass public to change their behavior in significant ways. Many efforts to encourage behavior change-such as public service announcements, social media posts, and billboards-involve short, persuasive appeals, yet the effectiveness of these messages is unclear. Early in the COVID-19 pandemic, we tested whether short messages could increase intentions to comply with public health guidelines. To identify promising messages, we conducted two pretests (n = 1,596) in which participants rated the persuasiveness of 56 unique messages: 31 based on the persuasion and social influence literatures and 25 drawn from a pool of crowdsourced messages generated by online respondents. The four top-rated messages emphasized: (1) civic responsibility to reciprocate the sacrifices of health care workers, (2) caring for the elderly and vulnerable, (3) a specific, sympathetic victim, and (4) limited health care system capacity. We then conducted three well-powered, pre-registered experiments (total n = 3,719) testing whether these four top-rated messages, and a standard public health message based on language from the CDC, increased intentions to comply with public health guidelines, such as masking in public spaces. In Study 1, we found the four messages and the standard public health message significantly outperformed a null control. In Studies 2 and 3, we compared the effects of persuasive messages to the standard public health message, finding that none consistently out-performed the standard message. This is in line with other research showing minimal persuasive effects of short messages after the very early stages of the pandemic. Across our studies, we found that (1) short messages can increase intentions to comply with public health guidelines, but (2) short messages featuring persuasive techniques from the social science literature did not substantially outperform standard public health messages.

There is substantial concern about democratic backsliding in the United States. Evidence includes notably high levels of animosity toward out-partisans and support for undemocratic practices (SUP) among the general public. Much less is known, however, about the views of elected officials-even though they influence democratic outcomes more directly. In a survey experiment conducted with state legislators (N = 534), we show that these officials exhibit less animosity toward the other party, less SUP, and less support for partisan violence (SPV) than the general pub-lic. However, legislators vastly overestimate the levels of animosity, SUP, and SPV among voters from the other party (though not among voters from their own party). Further, those legislators randomly assigned to receive accurate information about the views of voters from the other party reported significantly lower SUP and marginally significantly lower partisan animosity toward the other party. This suggests that leg-islators' democratic attitudes are causally linked to their perceptions of other-party voters' democratic attitudes. Our findings highlight the importance of ensuring that office holders have access to reliable information about voters from both parties.

The emergence of transformer models that leverage deep learning and web-scale corpora has made it possible for artificial intelligence (AI) to tackle many higher-order cognitive tasks, with critical implications for industry, government, and labor markets in the US and globally. Here, we investigate whether the currently most powerful, openly-available AI model – GPT-3 – is capable of influencing the beliefs of humans, a social behavior recently seen as a unique purview of other humans. Across three preregistered experiments featuring diverse samples of Americans (total N=4,836), we find consistent evidence that messages generated by AI are persuasive across a number of policy issues, including an assault weapon ban, a carbon tax, and a paid parental-leave program. Further, AI-generated messages were as persuasive as messages crafted by lay humans. Compared to the human authors, participants rated the author of AI messages as being more factual and logical, but less angry, unique, and less likely to use story-telling. Our results show the current generation of large language models can persuade humans, even on polarized policy issues. This work raises important implications for regulating AI applications in political contexts, to counter its potential use in misinformation campaigns and other deceptive political activities.

The past decade has witnessed a proliferation of big team science (BTS), endeavours where a comparatively large number of researchers pool their intellectual and/or material resources in pursuit of a common goal. Despite this burgeoning interest, there exists little guidance on how to create, manage and participate in these collaborations. In this paper, we integrate insights from a multi-disciplinary set of BTS initiatives to provide a how-to guide for BTS. We first discuss initial considerations for launching a BTS project, such as building the team, identifying leadership, governance, tools and open science approaches. We then turn to issues related to running and completing a BTS project, such as study design, ethical approvals and issues related to data collection, management and analysis. Finally, we address topics that present special challenges for BTS, including authorship decisions, collaborative writing and team decision-making.

While most Americans perceive current levels of inequality in the US as too high, economically progressive candidates who champion redistributive policies designed to reduce inequality rarely win elections. Here we propose that progressive candidates achieve greater support by framing their policy platforms in terms of values that resonate beyond their progressive base. In two experiments (total N = 4,138), including one pre-registered experiment conducted on a national probability sample, we found that a presidential candidate who framed his progressive economic platform using values consistent with the “binding” moral foundations – e.g., patriotism, family, and respect for tradition – as opposed to values consistent with the “individualizing” foundations – e.g., equality and social justice – received significantly stronger support. This effect was driven by increased support among conservatives and, unexpectedly, among moderates as well. By comparison, a manipulation of how progressive the candidate’s platform was had small and inconsistent effects. Despite the potential gains associated with binding framing, analyses using (a) text-based analysis of every word spoken in a presidential debate since 2000, and (b) machine learning-based measures of recent presidential candidates’ economic views, reveal that appeals to binding values are least common among economically progressive candidates. These findings show, however, that the alignment between values and economic policies is malleable, suggesting economically progressive candidates can build broader coalitions by reframing the values they associate with their platforms.