Quick Take: Fourteen Recommendations
Content Approach and Tone:
1. Respect Your Audience’s Intelligence
2. Back Assertions with Evidence and Examples
Messenger and Credibility:
3. Keep Your Messengers Somewhat Neutral
4. Introduce Your Messenger
Cultural Relevance and Sensitivity:
5. Harness Entertainment, Pop Culture, and Humor
6. Harness Shared Values and “Latino” Cultural Cues
7. Partner with Latino Voices
Language and Communication:
8. Produce Content in English AND Spanish
9. Shy Away from the Use of Spanglish
10. Don’t Exaggerate Accents
Visual and Media Elements:
11. Tailor Content to the Medium
12. Mix Visual Effects
Continuous Improvement:
13. Gather and Incorporate Audience Feedback into Your Content
Accessibility:
14. Employ Visual, Hearing, and Language Accessibility
Executive Summary
Inoculation, or “prebunking,” is a promising technique in combating misinformation. Inoculation builds resistance against falsehoods by exposing people to weakened forms of false or misleading narratives, or by exposing and explaining to audiences the techniques bad actors might use to manipulate them online.
To understand how prebunking can be best harnessed for Latino communities in the U.S., the Digital Democracy Institute of the Americas (DDIA) conducted the first-of-its kind randomized controlled trial (RCT) with U.S. Latino communities. We tested four approaches to inoculation against the manipulative techniques of emotional language, ad-hominem attacks, and false dichotomies.
The study measured the prebunking videos’ impact on:
Participants’ ability to identify manipulative techniques used in misleading content
Participants’ willingness to share manipulative content
Participants’ ability to detect whether manipulative content was trustworthy or not
We also measured whether participants found the inoculation interventions (videos) themselves to be 1) accurate, 2) entertaining, and 3) shareable.
DDIA tested inoculation interventions crafted by Roozenbeek and colleagues at the University of Cambridge’s Social Decision-Making Lab, along with three adaptations of the interventions designed by Latino content producers and creators with which we partnered: MediaPlus, Cabezahueca Films, and the fact-checking non-profit Factchequeado.
The adapted interventions varied in content, approach, language and design. Each producer was tasked with adapting the same specific parts of the Cambridge video scripts, and were asked to insert references, visual approaches, and aspects of storytelling that, based on their expertise, they best felt would resonate with the Latino audiences with whom they’ve worked.
The three “culturally competent” video variations included:
Approach 1 (MediaPlus):
Fully animated videos: emotional language (English, Spanish), ad-hominem attacks (English, Spanish), false dichotomies (English, Spanish)
Animated Latino-identifying characters as the messengers
Disney references to illustrate the tactics being used
The use of English- and Spanish-language communication, merged (Spanglish) - note: final videos were adapted per what we learned, and no longer use Spanglish. They are available separately in English and Spanish.
Approach 2 (Cabezahueca Films):
Partially animated videos: emotional language, ad-hominem attacks, false dichotomies
A Latino actor as the messenger
Well-known sports, movies, and series scenes (soccer matches, Star Wars, Simpsons) to illustrate the tactics being used
The use of only English communication
Approach 3 (Factchequeado):
Vertical videos, produced for social media, with dissemination on TikTok and Instagram in mind: emotional language, ad-hominem attacks, false dichotomies
Direct-to-camera, real-life Latino man (Venezuelan journalist) as the messenger
Political and issue-specific references (like immigration) and real-world examples to illustrate the tactics being used
Videos fully produced in English and Spanish, separately
The original approach included:
Original Approach (University of Cambridge Social Decision-Making Lab):
Original interventions, partially animated videos: emotional language, ad-hominem attacks, false dichotomies
Non-Latino character as the messenger
Well-known movies and series references (Family Guy, Simpsons, Star Wars)
The use of only English communication
DDIA also worked with Lunario, a YouTube content creation media company, to produce two prebunking “narrative explainers” in Spanish that focused on topics of election security and non-citizens voting.
The two videos and their YouTube shorts featured a Latino content creator speaking directly to camera in a news-style approach. The videos were tested in focus groups, but were not included in the survey. Findings from the focus group conversations are integrated into this analysis. (See the links to all of the full videos, plus versions with English subtitles, in the Appendix.)
Top-Line:
This study revealed that inoculation interventions are overall effective among U.S. Latinos. DDIA found that while all inoculation interventions had some positive effect, culturally competent content (particularly from MediaPlus and Cabezahueca Films) demonstrated enhanced effectiveness across multiple measures (see below). This underscores the value of tailoring inoculation strategies to specific cultural contexts when addressing misinformation among diverse audiences.
The Impact of Prebunking on U.S. Latinos
Participants exposed to any inoculation video, whether the original Cambridge videos or the “culturally competent” variations, demonstrated an improvement in their ability to identify manipulative social media posts (manipulation discernment) and a reduction in their intent to share manipulative content compared to neutral content (sharing discernment).
On average, however, effects on their ability to judge the trustworthiness of social media content (trustworthiness discernment) were weaker and not statistically significant. This may be because the videos we produced and tested in the survey focused on tactics of manipulation rather than on false or misleading narratives and claims (the idea being that spotting manipulation could translate to spotting misinformation).
The strength of interventions varied across approaches and outcomes.
Manipulation Discernment (ability to identify manipulation tactics in social media posts)
The MediaPlus videos (Approach 1) emerged as the most effective at moving respondents toward being better able to identify manipulation tactics, with Cabezahueca (Approach 2), Cambridge (Original Approach), and Factchequeado (Approach 3) following closely.
Reminder: MediaPlus produced fully animated videos and leveraged cross-generational entertainment content such as the Simpsons and Disney.
Sharing Discernment (intent to share manipulative vs. neutral content)
The Cambridge (Original Approach) and Cabezahueca (Approach 2) interventions slightly outperformed MediaPlus (Approach 1) and Factchequeado (Approach 3) in moving respondents toward being less likely to share manipulative content relative to non-manipulative content (see Figure 1: Discernment Across Content Providers).
Reminder: Cambridge used fully animated videos with cross-generational cultural references based on Disney movies and Star Wars.
Reminder: Cabezahueca used a hybrid approach involving a live actor and animation, relying on nearly the same cultural references as Cambridge (e.g., Futurama, Star Wars).
Trustworthiness Discernment (ability to judge whether posts are trustworthy or not)
Most content providers' videos had limited impact on respondents’ abilities to judge whether social media posts they were presented with were trustworthy or not, with MediaPlus being the notable exception showing some effectiveness.
The Importance of Accurate, Entertaining, and Shareable Prebunks
Although all of the prebunking interventions we tested were in some way successful in inoculating Latinos against social media content that employed manipulative tactics, there were some notable differences in how participants felt about the videos themselves, insofar as their accuracy, entertainment value, and shareability.
Specifically, videos that harnessed strong real-world political references and seemed to be produced for specific dissemination on social media were rated as slightly less accurate, entertaining, and shareable. This shows the difficulty of producing prebunks that address politicized topics in polarized environments. It shows the difficulty of communicating on social media, where many Latinos spend time consuming news and information but where they are also highly distrusting of content. And, it shows the fine line we navigate in a context wherein most Latinos are familiar with misinformation narratives and claims but are uncertain (or skeptical) about whether content they see online is true or false, per DDIA polling research released in 2022 (as the Equis Counter-Disinformation Department) and in 2024.
This also highlights a crucial aspect of inoculation: prebunking is most effective when using weakened forms of misinformation, more politically neutral content, and more cultural, “fun” references. Why? Directly referencing misinformation that gets spread along partisan lines or overtly political content, it seems, can cause audiences to “tune out” if they perceive videos as being an attempt to persuade them directly, or – as some focus group participants put it – to “tell them what to think.”
Remember: When it comes to inoculation, the goal is to maximize learning new concepts and building resilience against manipulation (e.g., recognizing ad-hominem attacks online or narratives that may come up) rather than undoing the existing effects of misinformation (fact-checking) or pushing people to adopt certain views (persuasion).
In the next section, we dive into best practices for content creators looking to produce effective prebunking messages for Latinos in the United States.
Note: Best practices take into account the results of a 3,000+ survey of Latino adults conducted between June 14 to July 22, 2024 (see full survey methodology in the Appendix). Recommendations are also informed by the findings from six focus groups of Latino adults conducted in April and May 2024 (two groups of 25-35-year-olds, two groups of 35-45-year-olds, and two groups of 45-55-year-olds). Focus group respondents were self-identified Latinos, screened to represent the states of Texas, Arizona, Nevada, Wisconsin, and Pennsylvania, and from countries of origin including Mexico, Venezuela, Cuba, Colombia, El Salvador, and the Dominican Republic.
Best Practices for Tailoring Prebunking and Inoculation Interventions to U.S. Latinos
The following 14 recommendations each address a crucial aspect of creating more effective prebunking content for Latino audiences – recommendations are divided into the following categories:
Content Approach and Tone: Best practices for message framing and delivery.
Messenger and Credibility: Strategies for establishing audience trust and credibility with the messenger.
Cultural Relevance and Sensitivity: Insights on incorporating cultural nuances and avoiding stereotypes.
Language and Communication: Best practices for language use, including considerations for bilingual content.
Visual and Media Elements: Recommendations on effective use of visuals and mixed media in prebunks.
Continuous Improvement: Recommendations for incorporating feedback to continue improving content over time.
Accessibility: Recommendations for incorporating features to make the content more accessible for a variety of users.
Content Approach and Tone:
Respect Your Audience’s Intelligence
Most Latinos are unsure whether misinformation they encounter is true or false. DDIA research has found that Latinos are reacting skeptically to new, strange or contradictory information presented as fact (our communities are not gullible). Prebunking content and tone should respect the audience's intelligence, and take into consideration the mistrust that may exist with elites or institutions.Focus on “leading people to water, not on forcing them to drink.”
Not everything should be prebunked, and not every audience is at high risk for engagement with misinformation. Focus on inoculating uncertain Latinos against narratives and techniques that are predictable and may get recycled. DDIA is constantly monitoring narratives and techniques being used in Latino spaces online. Learn more here.
Employ the prebunking recipe of “forewarning,” “microdosing” and “refuting” without telling people directly what they should or should not think.
Employ subtle warnings and calls to action, not direct call-outs, when cautioning against manipulation tactics or false or misleading content.
Encourage critical thinking but avoid phrases like “don’t let yourself be fooled,” or “don’t be an idiot,” as such direct call-outs frame the viewer as a victim with no agency of their own. This type of language can also elicit feelings of shame or embarrassment that trigger defensiveness.
Instead, rely on subtle questions in the spirit of:
“Before sharing something online, ask yourself, ‘does this seem true?’”
“I wonder, what is the intent behind this message?”
Rely on subtle calls to action in the spirit of:
“Just because they get personal, doesn’t mean you have to.”
“When things are too black and white, dare to be gray.”
“Keep in mind that if you are being presented with only two options, someone may be trying to manipulate you.”
Introduce scenarios and follow them with questions that guide audiences toward specific conclusions. This can help viewers reflect and analyze without feeling judged.
It’s important to introduce scenarios that viewers find relatable. MediaPlus and Cabezahueca Films utilized references to soccer, to family, and to Shakira to draw parallels between techniques and real-world examples. Disney, The Simpsons, and Star Wars, all popular and widely watched movies and shows among Latinos, also helped guide audiences toward understanding how techniques may get used online. Factchequeado used jokes about pandas, headlines, and emojis to illustrate their points. All of these scenarios played well with audiences in focus group settings.
Back Assertions with Evidence and Examples
Prebunking content tells a story, but it should also include some concise facts and relatable examples that show what you produced is more than your own “opinion.”
Content we tested that backed assertions up with evidence was shown to be trusted more than other types of content, according to focus group participants. Show AND tell, don’t JUST tell. And be careful not to “preach” or “pander.”Incorporate relevant numbers or research findings to support key points.
From time to time, feature or quote recognized experts in the field.
Ensure sources and statistics are clearly cited and easily verifiable by the audience.
Good example: Sylvia Salazar notes statistics about the number of cases brought up about election security and how they were disproven in this video about election security.
Another good example: Sylvia Salazar notes who is responsible for elections in the states, and offers examples of how narratives get used to exploit real-world concerns in this video that counters the narrative about non-citizens voting.
Limit the number of concepts introduced in each video to avoid overwhelming viewers. We recommend one to three concepts per video.
Messenger and Credibility:
Keep Your Messengers Somewhat Neutral
Prebunking and political persuasion are not one and the same. Per Google’s Jigsaw and the University of Cambridge, “prebunking messages are designed to help people identify and resist manipulative content. By forewarning people and equipping them to spot and refute misleading arguments, these messages help viewers gain resilience to being misled in the future.” This goal has implications for who you use as the messenger.
Messengers that are seen as more neutral or well-balanced in their approach can make your inoculation content more effective - be cognizant that if your messenger signals strong partisanship, you may instantly be seen as “having an agenda” by viewers, and thus lose ground and credibility with some communities.
To shed light on the efficacy of three messenger personas in shaping reactions to false narratives (journalists, activists, and a friend next door) the Equis Counter-Disinformation Department (DDIA before we spun off) conducted a panel test of 5,516 Latino adults in the United States in partnership with Civis Analytics in 2022. In defining the most successful messenger, Equis considered the messenger’s effectiveness in achieving three outcomes:
Reducing uncertainty about the trustworthiness of the original false claim
Increasing belief that the original false claim was false
Lowering belief that the original false claim was true
All three messengers — the Journalist, the Activist, and your friend next door — were effective, at a 95% confidence interval, in significantly reducing uncertainty about the original false claim. The English-language journalist, however, was the most effective across the board. This signals that neutral parties may be the best messengers for inoculation content. Check out the full findings about messengers here.
Note: Your choice of messenger should consider the subgroup of Latino you are targeting, the medium you are using, and how far down the “rabbit hole” your audience is in their consumption of false and misleading content (see more on Targeting in the section below).
Introduce Your Messenger
Building audience trust with messengers matters, especially on social media. Per our focus groups, Latinos exhibit some distrust and a heavy dose of skepticism about messengers that signal they are “nameless content creators,” even as these communities indicate spending a significant amount of time consuming news and information produced by influencers on social media. This is especially salient for content produced for Instagram or TikTok.
A simple label with the name of the messenger, a reference to the messenger’s organization, or to other indicators that demystify the messenger, can assuage some of that skepticism. Latinos appreciate knowing who the person or organization communicating to them is.
When relevant, provide credentials or background information to establish your messenger’s “authority” on a topic.
Highlight any relevant shared experience with the viewers, such as a shared lived experience or common concern.
Cultural Relevance and Contextualization:
Harness Entertainment, Pop Culture, and Humor
Culturally competent interventions, particularly those using pop culture, humor, and more apolitical content, slightly outperformed the Cambridge University inoculation content, highlighting the importance of tailoring strategies to specific audiences.
Entertainment value and shareability are crucial factors alongside effectiveness, as they determine the potential reach and impact of inoculation interventions. Overly political references in inoculation videos may hurt the broader reach of these interventions.
Pop Culture vs. Politics: People found references to music, movies, and comedy that illustrated the tactic being used to be both entertaining and engaging. When pop cultural content depicted political settings, that still seemed subtle enough for audiences to find effective (example, a The Simpsons political debate). Real-world political or overt references to election fraud or immigration policies elicited mixed feelings, however, with some folks citing the references as having inserted “bias” into the explanations.
Examples of pop cultural references included: The Simpsons, Disney movie clips, Star Wars, Shakira, soccer games, family references, references to pets or animals (pandas), newspaper clippings and animated stories about a family’s migration story.
Examples of real-world references included: videos of Donald Trump, immigration news clippings, and scenes of the border wall.
Intergenerational References: Respondents of our six focus groups expressed liking the use of examples that multiple generations would understand and connect to, such that a family unit could together sit down and engage with the content.
Incorporate everyday scenarios that the audience can easily connect with. There is no one-size-fits-all recipe for choosing references to use in your content. Content should address a diversity of experiences and consider factors such as age, gender, socioeconomic status, and the cultural significance.
Humor: Humor may be one tool within the toolkit to communicate certain ideas and engage viewers. Funny content is more likely to be watched and shared by viewers. It can also make a subject that is complex more accessible, increasing the likelihood that the audience will engage with the message.
Harness Shared Values and “Latino” Cultural References
Shared Values and Empathy: Stories that evoke shared values and empathy can be particularly compelling for viewers.
Latino communities, per the recently released Equis Latino Playbook, often embrace values of family, personal responsibility, and hard work. When asked an open-ended question about values, a plurality of Latinos (41%) reported family as the value that is most important to them. Harness messaging that explicitly acknowledges the importance of family and centers people’s personal agency in creating a healthier internet (and not as victims of misinformation). For example, using real-life examples where misinformation can lead to negative consequences for family may evoke a sense of responsibility for others.
Latino Cultural Cues: Recent Equis research also found that “using visual cultural ‘cues’ in English-language videos can be an effective way to signal that content was made specifically for Latino audiences, without any backlash among non-Latinos.” Subtly harness visuals of Latino workers, foods, household items, family gatherings, or even Spanish writing (see examples on page 23 of the Equis Latino Playbook).
A note from Equis: “The cultural cues must be credible and authentic in order to succeed. This is not about “Latino coating,” defined as “a superficial marketing strategy that attempts to sprinkle Latino elements onto products, campaigns, or entertainment without a genuine understanding or respect for Latino culture.” Choosing the right cultural cues in your content requires deep cultural competency and should come from a place of respect for Latino cultures.”
Partner with Latino Voices:
Involving Latino creators, journalists, and subject-matter experts in creating prebunking messaging content can enhance its credibility and relevance.
Collaborate with recognized Latino figures or influencers in various fields to create or endorse prebunking messaging content.
Keep in mind your messengers should lean “neutral” (see recommendations above). Local journalists may be a good place to start. Feature diverse Latino voices to represent the community’s varied perspectives and experiences.
Language and Communication:
Produce Content in English AND Spanish
Thoughtful use of language is important for connecting with Latino communities both instrumentally (for comprehension) and symbolically (to signal cultural identity).
Many Latinos report consuming content online in both languages seamlessly, with some even suggesting, per our focus groups, that when it comes to short-form content, they prefer that videos be in Spanish. Content should thus be produced in both English AND Spanish to cater to the multilingual nature of Latino communities.
Invest in Spanish
Monolingual Spanish speakers and Spanish-dominants need to be engaged in Spanish if we want them to hear and understand our messaging
Spanish can communicate values, customs, and habits distinctive to certain groups and can prime Latino identity for English speakers.
Tailor language use to the preferences of the target audience based on demographic factors.
Ensure translations maintain the nuances of the message.
Shy Away from the Use of Spanglish
Bilingual content that employs both languages in one single output doesn’t perform better than other language approaches.
Per Equis Research, Spanglish in particular appears to be the most volatile choice for bilingual content - with potential for big effects in either direction. If you intend to use Spanglish in your interventions, test your content.
Don’t Exaggerate Accents
When creating prebunking messages for Latino audiences, evaluate the use of messengers with accents both from a perspective of authenticity and from a perspective of audience engagement.
Accents can be used by your messenger, but in choosing your messenger, know that very strong accents can signal that you are trying too hard to convey “Latinidad,” sometimes to the point of coming off as inauthentic.
Most English-as-a-second-language speakers and Spanish-dominant Latinos in our focus groups empathized with a detectable accent, but very strong accents were thought to be both hard to understand and over-accommodating/pandering.
Be mindful of accents from different countries. Each Spanish-speaking country has unique accents, as do the states within each of these countries.
Know with which audiences you are speaking, and where they or their families are from.
If you are attempting to engage a large swatch of Latino communities, pick a neutral accent and avoid the use of colloquialisms in the messenger’s speech.
Visual and Media Elements:
Tailor Content to the Medium
Adapt content for different social media platforms, but keep in mind production quality and audience expectations.
Consider factors like video length, aspect ratio, and caption requirements, as well as the audience’s expectations (entertainment, learning) on that platform.
In focus groups, Latino participants noted the differences in how they engage with various platforms - YouTube was noted as being used more for information and learning, while Instagram was used more for entertainment.
But, note that viewers may expect higher quality videos on platforms like YouTube where such videos are the norm (relative to other platforms such as TikTok and Instagram).
Mix Visual Effects
Effective visuals should support and enhance the content rather than distract from it. While visually appealing content is important, it shouldn't come at the expense of the message or credibility of the prebunk.
Incorporate visuals that help break down complex ideas or illustrate key points that reinforce the message.
Use a mix of animation and real-world imagery to maintain a balance between engaging visuals and more serious content.
This comes with a caution - among groups of Latinos between the ages of 25 and 55, videos with 100% animations often convey “juvenile” or “childish” even in cases where the content was engaging.
Continuous Improvement:
Gather and Incorporate Audience Feedback into Your Content
To ensure prebunks remain effective and relevant, it’s crucial to continuously gather and act on feedback from Latino audiences.
Carefully measure whether interventions move outcomes of interest (e.g., misinformation sharing) or closely related proxies (e.g., sharing intention) using rigorous methodologies like randomized controlled trials (RCTs) that allow us to estimate causal effects.
Identification of manipulation and reduced sharing of manipulative content are a good sign that the intervention is working as intended, but carefully monitor how it’s affecting beliefs in genuine content. Though it is rare, inoculation interventions can sometimes produce a “general skepticism” that prompts people to be skeptical about everything they see online, not just misinformation.
Regularly review engagement metrics and comments on prebunk content.
Test different approaches and formats with Latino audiences. Organizations like the Brennan Center for Justice offer testing through Swayable, as do many civil society organizations working in this space.
Be willing to adapt content based on what resonates most effectively, always remembering that Latinos are not a monolith and targeting should be done with that in mind.
Accessibility:
Employ Visual, Hearing, and Language Accessibility
Include closed captions for video content, and also ensure that sufficient color contrast and readable fonts exist in visual materials.
Use language that is clear and straightforward, avoiding the use of jargon, to make content accessible to many audiences.
Targeting Interventions
U.S. Latinos are a very heterogeneous group with diverse backgrounds, beliefs, and lifestyles. Though the overall findings of our experiment point to the effectiveness of inoculation among Latinos, a crucial question is how well these interventions work among different subgroups.
Despite finding no discernible evidence of backlash (as in, people DID NOT become less capable of sifting between manipulative and non-manipulative content) we did observe general variations in the effectiveness of our content – inoculation helped all Latinos on average, but certain groups of Latinos were more responsive to these interventions.
As we highlight in our 2024 report, Latinos who are uncertain about the content they see online versus those who are already highly exposed and believing in misinformation, are better targets for inoculation. Latina women, Facebook users, and those with a closer connection to the immigration experience are more likely to fit into the uncertain category.
Focusing on manipulation discernment, inoculation appears to be particularly effective for women, millennial and Gen X Latinos, English speakers, immigrant citizens, those with less than college education, those with Puerto Rican ancestry, and those with the lowest political interest.
However, our analysis also highlights areas where the intervention may need refinement, such as for men, Spanish speakers, adults over 60, and those under 30 where effects were muted.
In sum, for subgroups that are more likely to fall into the category of “uncertain,” inoculation interventions tend to nudge them toward greater accuracy. However, for some of the subgroups, effects may be weaker due to resistance to correction or a lack of cultural resonance. Parsing out this variation could be an important avenue for future research.
Looking to the Future
Despite the effectiveness of the prebunking interventions created and tested as part of this experiment, it is important to note that effects were generally small. Most interventions moved outcomes by 2 or 3 percentage points, and the largest observed subgroup effects were 6 percentage points among immigrant citizens. This suggests that inoculation interventions “nudge” people toward greater discernment, but may fall short of producing significant changes in behavior.
Overall, our findings highlight the promise of inoculation as a tool for combating misinformation and suggest that culturally competent adaptations of interventions are successful among Latinos. They also suggest that the approach is generally effective, albeit to varying degrees.
Future research should consider how to boost the effectiveness of inoculation by relying on more intensive treatments (e.g., trainings) and potentially pairing inoculation with other promising misinformation-reducing interventions such as digital literacy tips and accuracy nudges.
Methodology
Best practices included in this guidebook are based on findings from a survey and six focus groups of Latino adults in the United States, conducted by DDIA in partnership with YouGov.
The DDIA survey was administered from June 14 to July 22, 2024.
The median age of the sample was 43. Most respondents (92%) completed the survey in English.
The sample was diverse in terms of education, with the largest groups having some college (28%) or being high school graduates (25%).
The gender breakdown was 55% women, 43% men, and 1.5% non-binary or other.
In terms of national origin, 49% identified as Mexican, 15% as Puerto Rican, 6.4% as Cuban, and 29% as other origins.
Registered voters made up 75% of the sample.
The sample showed diverse political affiliations, with the largest groups being strong Democrats (20%) and pure independents (20%). Ideologically, the sample was mostly moderate (37%) and liberal (30%), with about 22% identifying as conservative.
Roughly 48% of respondents were either immigrants or first-generation (born in the U.S. to immigrant parents), while 51% were second or third generation.
After exposure to either placebo or inoculation videos within each trial, respondents were asked to rate six social media posts on three dimensions: whether they were manipulative, whether they would share them, and how “trustworthy” they were.
Ratings were captured using a seven-point Likert scale, ranging from “strongly disagree” to “strongly agree.”
For each manipulation tactic, 10 headlines were taken from Roozenbeek et al. (2022), with five additional headlines focusing on more politically relevant misinformation topics such as anti-LGBTQ+ content, the US-Mexico border, vaccination, and the Gaza war. (A sample headline is included in the Appendix.)
APPENDIX:
Video Links:
Prebunking Videos (Focused on Tactics)
Approach 1 (MediaPlus):
Emotional language (English, Spanish)
Ad-hominem attacks (English, Spanish)
False dichotomies (English, Spanish)
Approach 2 (Cabezahueca Films):
Approach 3 (Factchequeado)
Original Approach (University of Cambridge Social Decision-Making Lab)
Prebunking Videos (Focused on Narratives)
¿Los que no son ciudadanos pueden votar? (Non-Citizens Voting)
Full Video (Spanish): https://youtu.be/4IYmGnz1hNA?si=DBiiVwquqASwnNvb
YouTube Short 1: https://youtube.com/shorts/MDvWFxUv7Nk
YouTube Short 2: https://www.youtube.com/shorts/QeqDov4ALJ4
¿Son seguras las elecciones en EE.UU.? (Election Security)
Full Video (Spanish): https://youtu.be/MO7qd_E6Siw?si=fF4aL6GcMKilSrLf
YouTube Short 1: https://youtube.com/shorts/TJG_CB9c1ps
YouTube Short 2: https://youtube.com/shorts/R5tEcsTZgFo?si=Wgefx7kGFw12h3dS
Example Headline Stimuli
Graphs
Figure A1. Coefficient plot presenting discernment estimates for each content provider and method (points), along with 95% confidence intervals (lines). Statistically significant estimates do not overlap with the 0 line. Higher scores indicate better discernment, such that people are less likely to view manipulative content as trustworthy relative to non-manipulative content (trustworthiness), less likely to share manipulative content vs. non-manipulative content (sharing), and more likely to identify the use of manipulation in manipulative content relative to non-manipulative content (manipulation).
Acknowledgements
The Digital Democracy Institute of the Americas and DDIA Founder and Executive Director Roberta Braga would like to extend a special thank you to the research consultants who helped make this study possible: Yamil Velez who led on polling and quantitative analysis and Eugene Kondratov who led on the qualitative analysis of the focus groups.
DDIA would also like to thank Em Bello-Pardo and Alissa Stollwork at YouGov for their continued partnership and support with polling and surveys, and DDIA Advisory Council Member Enrique Castillo and his team at Enrique Castillo & Associates for their leadership in conducting the focus groups that were so crucial for these learnings.
Thank you to the brilliant producers and content creators with whom we partnered to adapt the interventions for U.S. Latinos: gracias to Beatriz Valdés Mariño and Alejandro Ulloa at MediaPlus, to Mario Ramos at Cabezahueca Films, and to Laura Zommer and Rafael Olavarria at Factchequeado. Thank you to Caitlin Opperman at Equis Research for her insights and feedback on this guide, and to Donald Partyka for designing the final product.
This project would not have been possible without the leadership of the Cambridge University Decision-Making Lab and the researchers who produced the inoculation framework and videos that inspired this research - thank you, Sander van der Linden, Jon Roozenbeek and Mikey Biddlestone for paving the way for future research on prebunking as a solution to information disorder.
The Digital Democracy Institute of the Americas (DDIA) is bringing together insights and actors across the Western Hemisphere to shape a more participatory, inclusive, and resilient digital democracy. We apply research at the intersection of information integrity, belief, and behavior to build trust, connection, and capacity with Latino and Latin American communities and to shape policy interventions that bridge-build and depolarize shared information spaces for healthier democracies in the Americas.
DDIA is a 501(c)(3) organization.
©2024 The Digital Democracy Institute of the Americas (DDIA). All rights reserved. No part of this report may be reproduced, republished, or transmitted in any form or by any means without permission in writing from DDIA, except in the case of quotations in news articles or reviews. Please direct inquiries to [email protected].
A Note on Intellectual Independence:
DDIA is committed to intellectual independence in all work that it performs, including all its projects and published materials. Staff and consultants of DDIA are solely responsible for the research, development, and creation of all materials that bear the DDIA name. In order to live up to the values of intellectual independence, DDIA requires that any donor agree that DDIA shall, at all times, maintain independent control over the work product that DDIA produces, and the conclusions contained therein.
As a 501(c)(3) public charity, DDIA adheres to all financial disclosure requirements mandated by the IRS and other applicable governmental agencies. DDIA maintains a strict commitment to furthering its 501(c)(3) goals of public education via transparent and reliable methods of intellectually independent study.