What actually works in practice Forget the academic models for a minute. Here’s what I’ve found genuinely works when creating content for vulnerable people:

Two studies investigated associations between habits and identity, in particular what people consider as their “true self.” Habit-identity associations were assessed by within-participant correlations between self-reported habit and associated true self ratings of 80 behaviors. The behaviors were instantiations of 10 basic values. In Study 1, significant correlations were observed between individual differences in the strength of habit-identity associations, measures of cognitive self-integration (prioritizing self-relevant information), self-esteem, and an orientation toward an ideal self. Study 2 further tested the assumption that habits are associated with identity if these relate to important goals or values. An experimental manipulation of value affirmation demonstrated that, compared to a control condition, habit-identity associations were stronger if participants explicitly generated the habit and true self ratings while indicating which values the behaviors would serve. Taken together, the results suggest that habits may serve to define who we are, in particular when these are considered in the context of self-related goals or central values. When habits relate to feelings of identity this comes with stronger cognitive self-integration, higher self-esteem, and a striving toward an ideal self. Linking habits to identity may sustain newly formed behaviors and may thus lead to more effective behavior change interventions.

But sometimes it can be hard to find the time (and motivation) to exercise. So, what's the least amount of exercise you can get away with doing while still seeing these benefits? That answer depends on how fit you are to begin with. Here's some good news: the lower your starting point is in terms of fitness, the less you have to do to see a benefit.

wait this graph is crazy BART installed anti-fare-hopping gates and the amount of station maintenance and cleanup they had to do went to basically zero strong evidence that the poor condition of public transit is fairly easy to fix + caused by a very small group of people

The Choctaw Nation drives were first launched in 2020, born from community feedback. Lacey Callahan, grants operation manager for the behavioral health center, explains that their original approach — hour-long formal presentations — wasn’t working. “What we heard from our community is that (the presentations) did not feel safe,” Callahan said. “What felt safer to them was to discreetly come through on their terms, when it was convenient for them, receive a smaller training just on how to use it, and not have law enforcement present.” The tribe now strategically places these drives based on precise data analysis. Mason Emert, an epidemiologist with the Choctaw public-health department, studies statewide overdose information collected in a federally developed program called the Overdose Detection Mapping Application Program (ODMAP), where users enter overdose data into a cross-jurisdictional database.

Nearly five years ago, in April 2020, just a few months into the pandemic, Jay van Bavel – a social psychologist at New York University – published a landmark paper aimed at helping those in power do just that. Bavel and colleagues' paper was influential as it made clear that governments couldn't rely solely on rules and regulations; they needed to motivate the public to follow them. In doing so, they proposed 19 behavioural principles – policies that were rooted in decades of psychological research and designed to help guide the public's response to the pandemic. Many of these principles were adopted by governments around the world, in the hope that psychology might help 'nudge' us, or even shove us, towards safer behaviours. These principles took many forms, from encouraging a shared sense of identity – 'we're all in this together' type thinking – to targeting fake news and misinformation. These questions were especially important to my own research, on adolescent social development and mental health. Could we really get young people to stick to these restrictions? And to what extent were the huge social sacrifices being asked of young people worth it in the long run? These unknown questions were hugely important to understand and to collect data on, and to understand what policies worked and were worth implementing. So, my interest was piqued, and it wasn't too long before I found myself in the unusual position of having first-hand experience of very different approaches to the behavioural principles in action…

Changing health information to match specific cultures can improve health outcomes. However, there are no government rules to make health information fit different cultures. We made a cultural tailoring score to test health materials. Graduate students from the target cultures tested it on COVID-19 vaccine ads. The score showed that cultural elements made the ads work better and helped people understand health messages. We suggest testing the score more to give researchers a simple tool for creating better health materials.

To support intervention developers in selecting BCTs to target specific MoAs, an online resource, called the Theory and Technique Tool (TaTT), was previously developed. This tool provides an evidence-based grid showing which BCTs are likely or unlikely to change certain MoAs. Recently, new tools—the Behaviour Change Technique Ontology (BCTO) and Mechanisms of Action Ontology (MoA Ontology)—were developed to include a wider range of BCTs and MoAs and provide more precise and computer-readable BCT and MoA definitions. By aligning the TaTT with these newer tools, we can support (1) ontology users in hypothesising about likely BCT-MoA links, and (2) TaTT users in identifying more detailed yet relevant BCTs and MoAs from the ontologies and using these in computer applications. This study aimed to map the newer ontologies’ categories to the TaTT’s 74 BCTs and 26 MoAs. Researchers carefully compared and discussed definitions from both tools to create mappings. The study found that 85 BCTs in the newer ontology corresponded to 74 BCTs from the TaTT, and 56 MoAs in the newer ontology corresponded to 26 MoAs from the TaTT. By linking the ontologies to the TaTT, this work makes it easier to use these tools together. This helps design and report behaviour change interventions more clearly and supports advanced uses like automated data analysis.

What can behavioural scientists do differently when working on complex problems? Given the need for differentiated methods for different kinds of systems, and particular caution about existing approaches for complex and chaotic domains, applied behavioural scientists should be considering the appropriateness of the cornerstones of applied work, such as defining target behaviours. early on in strategic development, as shown in the table below. The Cynefin framework is a ‘sense-making’ framework – where sense-making is defined by the author David Snowden as “making sense of the world in order to act in it”. It distinguishes between 3 primary systems: ordered, complex, chaotic, which are defined by the type of constraints (or absence of constraints) in that system. Each type of system is described not just how it is constrained, but also describes how to best take action.

What can behavioural scientists do differently when working on complex problems? Given the need for differentiated methods for different kinds of systems, and particular caution about existing approaches for complex and chaotic domains, applied behavioural scientists should be considering the appropriateness of the cornerstones of applied work, such as defining target behaviours. early on in strategic development, as shown in the table below. The Cynefin framework is a ‘sense-making’ framework – where sense-making is defined by the author David Snowden as “making sense of the world in order to act in it”. It distinguishes between 3 primary systems: ordered, complex, chaotic, which are defined by the type of constraints (or absence of constraints) in that system. Each type of system is described not just how it is constrained, but also describes how to best take action.

Fundamentals of Inclusive Research Presenter(s): Cherish Boxall, Heidi Green, Frances Sherratt, Shaun Treweek decorative image to accompany text This guide provides four approaches to making research more inclusive. Groups of people, such as those from minoritised/racialised ethnicities, impaired capacity, and those experiencing socioeconomic disadvantage, generally experience poorer health outcomes than groups of people with more societal privilege. In parallel, these groups have been historically underserved in health research. This situation means that the findings of health research might not be transferable to the people who stand to benefit most, potentially allowing health inequalities to continue and most definitely not contributing to solving the problem of inequity. There are many historical and current reasons for under-representation in research. The most common reasons include a lack of trust and ease of access (e.g., small visit windows, cover for dependants). Although different groups might face unique barriers, this practical resource will provide a starting place to help research be more inclusive through a broad suite of approaches. The four pillars of the fundamentals of inclusive research are access, relevance, trust and recognition. > Download the workbook PDF with a check guide to get started on inclusive practice and community engagement.

Our Behavioral Determinants Framework overlays the foundational theories from social psychology and behavioral economics onto the core challenge of marketing – identifying the factors that influence what people do. Our interventions are aimed at those behavioral determinants that move audiences beyond awareness and into action. To simplify it all, we categorize these behavioral determinants into three categories: Fun, Easy and Popular. After all, fun, easy and popular is the heart of why we humans do just about anything. The Behavioral Determinants Framework helps us frame target behaviors to address an audience’s own wants and needs rather than trying to convince them to change their values and beliefs.

obs-to-be-done is a great concept for innovators, helping to take the customer perspective and discovering customer insights for innovation and growth strategies. When applying JTBD in practice, however, innovators often get lost. The Job Hierarchy, developed by Vendbridgeand applied in dozens of JTBD projects, can help to maintain orientation and focus, and thereby to exploit the full potential of this powerful concept. Die Job Hierarchie As the word hierarchy implies, we use it to think JTBD in three different levels: The Bigger Why The Deeper Why The Lower How

Intervention Mapping is the most comprehensive approach to systematic behavior change (O'Cathain et al., 2019). For this reason, applying Intervention Mapping can been daunting. In this paper, I discuss the mistakes I made when applying the various iterative steps of Intervention Mapping in the design, implementation, and evaluation of an intervention aiming to reduce HIV stigma in health care settings in the Netherlands.

Personalized nudging (PeN) promises greater intervention effectiveness, especially for heterogenous populations. However, developments in PeN are hindered due to a lack of conceptual clarity and high methodological variability. We present a framework for PeN to tackle these challenges. We argue that personalization is contingent on personal data availability and choice environment malleability. Applying these factors to a nudge’s content, design, and underlying mechanism, we suggest that various levels of PeN exist, from simple name changes to more technologically sophisticated adaptive approaches. These levels highlight various novel methodological considerations, which we split into theory-driven (top down) and data-driven (bottom up) approaches. Finally, we discuss how our framework supports practitioner goals and reveals future research directions.

Find the Perfect Check-In Question for Your Group Stop struggling to think of engaging check-in questions. Get 15 personalized recommendations based on the wisdom of Nadia's 25 years as a community arts facilitator.

Yet the latest research shows something inconvenient for that narrative: a sharp reduction in fentanyl overdoses that started before Trump took office, almost certainly in response to policy under his predecessor Joe Biden. As researchers noted in a paper published in the journal Science this week, fatal overdoses from synthetic opioids like fentanyl plummeted after peaking at 76,000 in 2023 in the US, dropping by over a third by the end of 2024. (Full numbers aren’t in yet for 2025, but provisional data from the CDC suggests another double-digit percentage drop.) The researchers proposed a possible explanation, writing that a “major disruption in the illicit fentanyl trade, possibly tied to Chinese government actions,” may have “translated into sharp reductions in overdose mortality beginning in mid- or late-2023 and continued into 2024 across both the US and Canada.” In other words, as Axiom reports, diplomatic pressure has proven far more effective than efforts to crack down on drug dealers on the street.

Creating a bot persona document. The part teams usually miss: Persona is the consistent personality users infer from your agent’s language choices — documented as constraints that guide every utterance. It prevents tone drift. It gives QA something testable. It makes “fun vs. professional” a decision, not a debate. And when persona isn’t defined, the agent becomes inconsistent in the places users notice most: greetings, errors, and handoffs.

To meet UK Net-Zero emissions targets, meat consumption must decrease. We present results from two studies evaluating interventions to reduce purchasing of meat-containing meals across university cafeterias in Oxford, UK. Study 1 tested whether two dynamic descriptive norm messages changed meal purchasing. Over eight weeks, four cafeterias displayed a norm message incorporating a socially ‘close’ referent group and three cafeterias displayed a message incorporating a socially ‘distant’ referent group. Two cafeterias were assigned a no-message control condition. A generalised linear mixed effect model suggested both messages decreased odds of cafeteria diners purchasing vegetarian meals, in comparison to control, 'Close' Message: Ratio of Odds Ratios (ORs)=0.79, 95% 95% CI [0.72, 0.86]; 'Remote' Message: Ratio of ORs=0.84, 95% CI [0.76,0.92]. Study 2 involved three pre-post experiments testing whether different interventions changed meal purchasing: re-positioning vegetarian products, increasing vegetarian availability, and introducing vegetarian defaults. Generalised linear models suggested each intervention was associated with significant increases in odds of diners purchasing vegetarian meals, Positioning: OR=1.33, 95% CI [1.24,1.44]; Availability: OR=1.60, 95% CI [1.45, 1.75]; Defaults: OR=1.77, 95% CI [1.61, 1.95]. These study results could be due to norm messaging being less effective at promoting vegetarian meals than interventions in availability, defaults, and positioning. But, given the study designs, they could instead be due to self-selection effects, or regression to the mean.

PCS yields three important discoveries in this investigation: First, context variables are more predictive of behavior for some individuals than others. Second, contrary to common wisdom, there is no “magic number” for how long it takes to form a habit. Instead, the speed of habit formation appears to vary significantly between behavioral domain: Gym habits take months to form and handwashing habits take weeks to form. Third, consistent with prior research on nonhuman animals, more habitual gymgoers are reward-insensitive, responding less to a well-designed behavioral intervention

This section contains over 240 case studies of social marketing / behavior change programs from around the globe, making it the largest open-access collection in the world. It includes a broad sampling of programs to offer a wide variety of approaches and tools used, locations, types of organizations and participants, activities being promoted and problems being addressed. Most of these case studies illustrate approaches that have worked. However, examples of potential pitfalls are also included to provide you with a realistic map of the terrain ahead.

Massive snowfalls like the one that hit the US east coast this week usually spell trouble for traffic. But critics of America's car-centric transport network are using the snow - and Twitter - to demonstrate how roads should be redesigned to make them safer for pedestrians.

Before the lights went out, my social enterprise, DevelopMetrics, turned those tools loose on the USAID archive—one last look at what half a century of development really taught us. If you allocate grants, run programs, or shape policy, this is the closest thing we have to a postmortem on how tens of billions of dollars in development aid actually behaved over the course of decades in the wild. It offers a model for future learning on a mass scale, and the results affirm some important guiding principles as the development ecosystem considers how to build going forward.

This playbook gives you a system to map, diagnose, and eliminate those forces, so your change work actually changes works. Inside this playbook, you’ll get: -> A 4-step framework for identifying what’s blocking adoption -> How to score and map behavioral friction -> What most change models miss (and how to fix it) -> Real-world strategies to turn insight into implementation

Because here’s what most marketers miss: triggering emotion isn’t the same as encoding emotion. You can make people feel something in the moment and leave zero trace in memory. You can stimulate without imprinting. You can shock, confuse, or provoke and still be completely forgettable. Real emotional coding doesn’t just activate feelings. It attaches those feelings to fundamental human drives. The deep psychological forces that shape how we see ourselves and move through the world.

The Columbia Digital Storytelling Lab (DSL) proudly presents the 8th edition of its annual “Immersive Things“ list, celebrating groundbreaking projects at the intersection of storytelling, technology, design and code. This year's selections highlight how creators are leveraging emerging technologies  (artificial intelligence, virtual and augmented reality, and generative systems ) to craft experiences that challenge traditional narratives and redefine audience engagement.

See what people are searching everywhere, from Google to ChatGPT

When you set out to convince, you make the conversation into a battle to be won and lost. You bring force and, as Newton’s laws of motion tell us, force creates counter-force. You’re creating resistance before you’ve even finished your argument. Winning means that in the end, we will just have one side (our own!). We want that unity of opinion, but we create opposition in the process.

What is the SHeLL EDITOR? Clear and simple health information is fundamental to high quality, safe, and person-centred healthcare. In practice though, writing health information in plain language is often difficult to get right. The SHeLL Editor uses fine-grained real-time feedback to help you learn and apply evidence-based health literacy strategies to written health information.

This paper offers a refreshed and expanded view of how behavioural science can support sustainable development. It presents a comprehensive, evidence-based resource designed to help countries integrate behavioural insights into their policies and programmes for achieving the Sustainable Development Goals (SDGs). At the heart of the paper is a global database of 201 behavioural and nudge interventions, each aligned with one or more of the 17 SDGs. You can explore the full database here: https://docs.google.com/spreadsheets/d/1tWy0X2Aq08kIUNYG-Cw5FQsvakdKSyKGKen7_hc2F48/edit?gid=1627241714#gid=1627241714

This means I can finally give you advice based on real usage patterns instead of hunches. I annotated OpenAI’s chart with some suggestions about when to use free versus advanced models.