The Research Behind Strive

The problem we're responding to

Youth wellbeing is declining globally, and it started before the pandemic

David Blanchflower's analysis of Global Minds surveys across 167 UN member countries found that young people aged 18–24 had significantly lower wellbeing than every older age group in the vast majority of countries studied. The decline began around 2011–2013 — well before COVID, though the pandemic made it worse. The pattern is especially pronounced for young women and shows up most clearly in self-administered surveys measuring negative affect: loneliness, anxiety, depression.

What's striking is the historical shift. For decades, the relationship between age and unhappiness followed a U-shape — worst in midlife, better at both ends. That pattern has disappeared. Young people are now the unhappiest age group in most countries.

Blanchflower, D.G. (2025). Declining Youth Well-being in 167 UN Countries. Does Survey Mode, or Question Matter? NBER Working Paper No. 33415. Builds on Blanchflower, Bryson & Xu (2024), NBER WP 32337.

Why the current response isn't enough

Universal school-based mindfulness: the MYRIAD trial

The idea seems obvious: teach young people to pay attention to their inner experience, and surely that helps. The MYRIAD trial tested this at scale.

Willem Kuyken and colleagues randomised 85 UK secondary schools — over 8,000 students aged 11–14 — to either a ten-session mindfulness programme (.b) delivered by classroom teachers, or normal school provision (which typically included some form of social-emotional education). At one-year follow-up, the mindfulness training showed no advantage on any primary outcome: risk for depression, social-emotional functioning, or wellbeing.

The companion paper, by Montero-Marin and colleagues in the same journal issue, found something worse. For students already at higher risk of mental health problems — the students who needed help most — the mindfulness intervention was associated with worse outcomes. Higher risk of depression. Poorer social-emotional functioning. Students in the mindfulness condition also reported lower mindfulness scores after the programme.

The lead investigators, including Mark Greenberg from CASEL, acknowledged the results openly. They pointed to insufficient intervention duration, low student practice outside lessons, and implementation challenges. The finding isn't that mindfulness is harmful in all contexts. It's that this particular delivery model — a short curriculum taught by classroom teachers to universal populations — didn't work. And for vulnerable students, it backfired.

Kuyken, W., Ball, S., Crane, C., et al. (2022). Effectiveness and cost-effectiveness of universal school-based mindfulness training compared with normal school provision. Evidence-Based Mental Health, 25(3), 99–109.

Montero-Marin, J., et al. (2022). School-based mindfulness training in early adolescence: what works, for whom, and how in the MYRIAD trial? Evidence-Based Mental Health, 25(3), 117–124.

Social-emotional learning: small effects that don't persist

Cipriano and colleagues at Yale conducted the most comprehensive recent meta-analysis of universal school-based SEL programmes. Across 424 studies from 53 countries, covering over 575,000 K–12 students, they found statistically significant improvements in skills, attitudes, behaviours, peer relationships, school functioning, and academic achievement compared to controls.

That sounds good until you look at the effect sizes: small, ranging from g = 0.10 to 0.22 depending on the outcome. And at six-month follow-up, the positive effects had largely vanished. Implementation quality varied enormously and moderated outcomes significantly.

Peter Gray's analysis on Substack put it bluntly: even the small transient effects might reflect changes in how participants believed they should respond rather than genuine shifts in functioning. The meta-analysis itself is more measured, but the data supports a concern: programmes that build emotional awareness without building the capacity to act risk leaving young people more articulate about their distress but no more able to change their circumstances.

Cipriano, C., Strambler, M.J., Naples, L.H., et al. (2023). The state of evidence for social and emotional learning: A contemporary meta-analysis of universal school-based SEL interventions. Child Development, 94(5), 1181–1204.

Note: The main global meta-analysis was published in 2023. A 2024 follow-up (Cipriano et al., Social and Emotional Learning: Research, Practice, and Policy, 3, 100029) focused specifically on US studies with similar findings.

Attention-capture dark patterns: the design environment students live in

While schools implement wellbeing programmes, the digital environments students spend hours in daily are engineered to do the opposite. Monge Roffarello and De Russis catalogued what they call “attention-capture dark patterns” — design mechanisms used by platforms to seize and hold user attention: infinite scroll, autoplay, pull-to-refresh (variable rewards modelled on slot machines), and disguised sponsored content.

Their fuller 2023 typology identified eleven distinct patterns, from “Time Fog” (hiding time cues so users lose track of duration) to recommendation algorithms that create endless content loops.

Monge Roffarello, A. & De Russis, L. (2023). Defining, Detecting, and Addressing Attention Capture Deceptive Patterns in Digital Interfaces. CHI '23 Full Paper. Builds on Monge Roffarello & De Russis (2022), CHI '22 Extended Abstract, and Lukoff et al. (2021).

Why agency, not awareness

Passivity is the brain's default. Control is what's learned.

This is Strive's foundational claim, and it comes from one of the most important revisions in the history of psychology.

In 1967, Martin Seligman and Steven Maier discovered learned helplessness: animals exposed to uncontrollable events stopped trying to escape, even when escape became possible. The finding became one of the most cited in psychology, and its application to human depression influenced decades of clinical practice.

In 2016, Maier and Seligman published a revision of their own theory. Using neuroscience evidence unavailable in the 1960s, they showed that the original interpretation was backwards. Passivity isn't learned. It's the brain's default response to prolonged aversive events, mediated by serotonergic activity in the dorsal raphe nucleus, which inhibits escape behaviour.

What has to be learned is control. The medial prefrontal cortex detects when actions actually produce outcomes. That detection — which Maier and Seligman call “learned control” — inhibits the dorsal raphe nucleus, overriding default passivity. But this detection has to be built through repeated experiences where actions produce visible results.

A 2023 follow-up paper by Baratta, Seligman, and Maier extended the implications: early experience with control literally “immunised” animals against later helplessness. The experience of agency in early life created lasting resilience. This is direct support for what Strive is trying to do — not as treatment after the fact, but as prevention through the accumulation of genuine experiences of control.

Maier, S.F. & Seligman, M.E.P. (2016). Learned Helplessness at Fifty: Insights from Neuroscience. Psychological Review, 123(4), 349–367.

Baratta, M.V., Seligman, M.E.P. & Maier, S.F. (2023). From helplessness to controllability: toward a neuroscience of resilience. Frontiers in Psychiatry, 14, 1170417.

How Strive applies the research

Bandura's four properties of human agency

Albert Bandura spent decades studying human agency — what he defined as the capacity to intentionally influence one's own functioning and life circumstances. His 2006 paper in Perspectives on Psychological Science identified four core properties. Strive's entire architecture maps onto them.

Intentionality — forming intentions that include action plans and strategies for realising them. Not “I want to exercise more” but “I will run for 20 minutes after school on Mondays and Wednesdays.” In Strive, this is the habit creation flow: choosing what you'll do, when, and where.

Forethought — the temporal extension of agency. Anticipating consequences, setting goals, planning courses of action. Bandura called this what provides “direction, coherence, and meaning to one's life.” In Strive, students don't just set a goal — they plan for the cue that triggers it, the location where it happens, what might go wrong, and what they'll do instead.

Self-reactiveness — monitoring and regulating the execution of plans. In Strive, this is the daily check-in, the weekly dots, the cumulative count that only goes up. Students see their own data without anyone grading it.

Self-reflectiveness — metacognitive evaluation of one's own thinking and actions. Bandura called this “the most distinctly human core property of agency.” This is where Strive made its biggest design trade-off, which we explain in detail below.

Bandura also described three modes of agency: personal (you act alone), proxy (you influence others who have resources to help), and collective (you pool your efforts with a group). Strive's group mechanic supports all three — personal habit tracking, the teacher as supportive proxy, the group as collective accountability.

Bandura, A. (2006). Toward a psychology of human agency. Perspectives on Psychological Science, 1(2), 164–180.

Bandura, A. (2001). Social cognitive theory: An agentic perspective. Annual Review of Psychology, 52, 1–26.

Bandura, A. (1997). Self-Efficacy: The Exercise of Control. W.H. Freeman.

Self-efficacy: why mastery experiences matter most

Self-efficacy — perceived capability to execute behaviours for specific outcomes — is the operational mechanism that powers agency. Bandura showed it's built through four sources, and they're not equal.

Mastery experiences are the most powerful. Actually doing something and succeeding builds belief in your capacity more than any other source. This is why Strive's streaks and cumulative counts exist — not as gamification, but as evidence of follow-through that the student can see. The count only goes up. We don't use percentages because they punish inconsistency rather than recognising effort.

Vicarious experiences — seeing similar others succeed — build self-efficacy through modelling. This is why Strive puts teachers and students in the same groups, tracking habits alongside each other. When a student sees their teacher also building a reading habit — and sometimes missing days — that changes the dynamic from assignment to shared practice.

Verbal persuasion — credible encouragement from people whose opinion matters. Strive's identity-focused language (“You're becoming someone who follows through”) draws on this, though we're careful not to over-claim.

Physiological and emotional states — managing the feelings that accompany difficult moments. Implementation intentions help here by automating the decision point, reducing the cognitive load at the moment of action.

Self-efficacy is domain-specific. Confidence in one area doesn't automatically transfer to another. This is why Strive lets students choose their own habits rather than assigning them — the mastery experience has to be in something the student actually cares about.

Implementation intentions: the most practically useful finding in behavioural science

Peter Gollwitzer's research on implementation intentions has been replicated extensively and the effect sizes are genuinely large by social science standards.

The core mechanism: instead of forming a vague goal (“I'll eat healthier”), you create a specific if-then plan (“When I sit down for lunch in the cafeteria, I will eat a piece of fruit before anything else”). A 2006 meta-analysis by Gollwitzer and Paschal Sheeran across 94 studies found a medium-to-large effect size (d = 0.65) on goal attainment.

The if-then format works by creating strong mental links between situational cues and planned responses. When the cue appears, the action follows more automatically. You're not relying on willpower at the decision point — you've already made the decision.

Strive's habit creation flow is an implementation intention. Students specify the cue that triggers their habit and the location where it happens. That's not a UX choice — it's the format the research says works. We also ask students to anticipate their most likely obstacle and plan a backup, drawing on Oettingen and Gollwitzer's work on Mental Contrasting with Implementation Intentions (MCII), which strengthens the effect by preparing for failure points.

Gollwitzer, P.M. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist, 54(7), 493–503.

Gollwitzer, P.M. & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119.

Oettingen, G. & Gollwitzer, P.M. (2010). Strategies of setting and implementing goals: Mental contrasting and implementation intentions. In J.E. Maddux & J.P. Tangney (Eds.), Social Psychological Foundations of Clinical Psychology, 114–135.

Self-Determination Theory: why this has to happen in community

Edward Deci and Richard Ryan's Self-Determination Theory identifies three psychological needs that have to be met for genuine motivation — not compliance, but actually wanting to do the thing.

Autonomy — and this is the most commonly misunderstood term in the framework. Autonomy in SDT doesn't mean independence or doing things alone. It means acting willingly, from your own values, with a sense of self-endorsement. You can follow someone else's suggestion autonomously if you genuinely endorse it. You can make your own choice non-autonomously if you're driven by guilt or pressure. In Strive, students choose their own habits. Nobody assigns them. The “I commit to this” moment is voluntary.

Competence — feeling effective in your interactions with the environment. Not being the best, but experiencing the effect of your own actions. Strive builds this through mastery experiences — streaks, cumulative counts, weekly dots. The count only goes up. We don't use percentages because they punish inconsistency rather than recognising effort.

Relatedness — meaningful social connection and belonging. This is the need most habit apps ignore entirely, and it's the one Strive was built around. When an educator tracks habits alongside their students, it stops being an assignment. It becomes something the group is doing together. That's not a social feature. It's a psychological need.

SDT also makes a critical distinction between autonomous motivation (where you act from genuine interest or personal value) and controlled motivation (where you act from external pressure or internalised guilt). Strive's entire design is oriented toward autonomous motivation — habits students choose to build, not habits imposed on them.

Ryan, R.M. & Deci, E.L. (2000). Self-determination theory and the facilitation of intrinsic motivation, social development, and well-being. American Psychologist, 55(1), 68–78.

Ryan, R.M. & Deci, E.L. (2017). Self-Determination Theory: Basic Psychological Needs in Motivation, Development, and Wellness. Guilford Press.

What we chose not to build, and why

The reflection trade-off

This is the design decision with the highest cost.

Research shows that written self-reflection strengthens self-regulation (Schmitz & Wiese, 2006). Bandura identified self-reflectiveness as the most distinctly human property of agency. Sanna Järvelä's work on self-regulated learning shows that data-as-scaffold — showing students their own learning patterns and prompting them to reflect on those patterns — is a powerful mechanism for developing self-regulation. In an ideal world, Strive would include reflective writing tools that let students examine their own habit data and draw conclusions.

We chose not to collect written reflections. Here's why.

Any system where young people write about their inner experiences creates a disclosure risk. A private reflection box that nobody monitors becomes a sealed container — a student could disclose abuse, self-harm, or crisis, and no one would see it. Prompts like “Why are you struggling with this habit?” can drift into territory that requires clinical oversight we can't provide. For students living with anxiety, depression, OCD, or eating disorders, reflection prompts about habit completion can be actively harmful.

When building tools for young people in schools, safeguarding comes first.

So we designed a different system. Strive uses contextual micro-prompts — short, specific, data-anchored messages that appear at behaviourally meaningful moments. “Yesterday was a miss. Today's a fresh start. Your cue is still: after breakfast in the kitchen.” The prompt seeds reflective thinking. There is no text box. No written response is collected. The prompt appears once, doesn't track whether it was read, and can be dismissed with a tap.

This approach draws on several lines of research. Gollwitzer's work shows that mental rehearsal — forming if-then links mentally — produces the behavioural effect without requiring written output. Rittle-Johnson (2006) found that internally generating explanations improves self-regulation even without verbalisation. And Järvelä's data-as-scaffold principle is exactly what Strive's weekly pattern prompts do: showing students their own completion patterns (“This habit happened every Tuesday and Thursday but not on weekends — interesting pattern”) as a reflective prompt that's inherently safer than open-ended questions about inner experience.

The deeper reflective conversations happen where they should — between teachers and students, face to face, where a teacher can respond if something difficult comes up. Strive provides the data and the prompts. Teachers bring the human part.

We think this is the right call. If evidence emerges for lightweight reflective practices that work for young people without the safeguarding risks, we'll adapt.

Schmitz, B. & Wiese, B.S. (2006). New perspectives for the evaluation of training sessions in self-regulated learning. Zeitschrift für Psychologie, 214(2), 93–104.

Rittle-Johnson, B. (2006). Promoting transfer: Effects of self-explanation and direct instruction. Child Development, 77(1), 1–15.

Järvelä, S. & Hadwin, A. (2013). New frontiers: Regulating learning in CSCL. Educational Psychologist, 48(1), 25–39. See also Järvelä, S., Malmberg, J. & Koivuniemi, M. (2016). Recognizing socially shared regulation by using the temporal sequences of online chat and logs in CSCL. Learning and Instruction, 42, 1–11.

What we deliberately excluded

These aren't gaps we haven't noticed. They're decisions we made deliberately, and we want to explain why.

The broader framework: what “high agency” actually means

Why the popular framing is incomplete

The concept of “high agency” has entered mainstream culture through tech discourse — Eric Weinstein coined it, George Mack popularised it, Andrej Karpathy amplified it. Their instinct is right: the capacity to act intentionally, resourcefully, and persistently matters enormously for human flourishing.

But the academic literature on agency is far richer than the popular version, and the gaps matter for education.

The popular framing individualises a relational phenomenon. Agency doesn't emerge from individual willpower alone. It emerges from the transaction between person and context. Bandura's proxy and collective agency, the OECD's co-agency framework, and Anne Edwards' concept of relational agency all show that the social dimension is constitutive. Strive's group mechanic isn't a nice-to-have — it's core to how agency develops.

It flattens a multi-dimensional capacity into a personality type. “Bias for action + disagreeability + resourcefulness” collapses intentionality, forethought, self-regulation, self-reflection, self-efficacy, and contextual affordance into a single label. Each dimension needs to be developed deliberately.

It ignores structure. Bourdieu's habitus, Sen's capability approach, and de Beauvoir's situated freedom all show that agency is enabled and constrained by material and social conditions. Students start from different positions. Strive should create equitable conditions for agency to develop, not a test of who already has it.

Katharine Greenaway's most important observation for educators: “It might be valuable to look at high agency in a different way: not what makes an individual agentic, but what are the conditions that allow agency to thrive.” That's what Strive tries to be. Not a measurement of who has agency. A set of conditions under which it can develop.

Emirbayer and Mische's temporal dimensions

One framework from the academic literature that we find especially useful, even though it doesn't appear in the Science page, comes from Emirbayer and Mische's 1998 paper “What Is Agency?” They argue that agency has three interpenetrating temporal orientations:

The iterational dimension is past-oriented — building reliable practices through selective reactivation of patterns that have worked before. In Strive, this is habit repetition, streak maintenance, the “never miss twice” principle.

The projective dimension is future-oriented — imagining alternatives, envisioning possible selves. In Strive, this is the identity-focused language: “You're becoming someone who follows through.”

The practical-evaluative dimension is present-oriented — making judgments in response to what's actually happening right now. In Strive, this is the implementation intention at the moment of decision, and the contextual prompts that help students navigate obstacles in real time.

Most habit apps only develop the iterational dimension — repeat the thing. Strive tries to develop all three.

Emirbayer, M. & Mische, A. (1998). What is agency? American Journal of Sociology, 103(4), 962–1023.

What we're still working out

Building an app grounded in research doesn't mean we've solved everything. A few things we're genuinely uncertain about.

Strive is built on research, but it's not finished. The pilot is teaching us things the literature can't.

Full reference list

The problem

• Blanchflower, D.G. (2025). Declining Youth Well-being in 167 UN Countries. NBER Working Paper No. 33415.
• Blanchflower, D.G., Bryson, A. & Xu, X. (2024). The Declining Mental Health of the Young. NBER Working Paper No. 32337.

The current response

• Kuyken, W., Ball, S., Crane, C., et al. (2022). Effectiveness and cost-effectiveness of universal school-based mindfulness training. Evidence-Based Mental Health, 25(3), 99–109.
• Montero-Marin, J., et al. (2022). School-based mindfulness training in early adolescence: what works, for whom, and how in the MYRIAD trial? Evidence-Based Mental Health, 25(3), 117–124.
• Cipriano, C., Strambler, M.J., Naples, L.H., et al. (2023). The state of evidence for social and emotional learning. Child Development, 94(5), 1181–1204.
• Monge Roffarello, A. & De Russis, L. (2023). Defining, Detecting, and Addressing Attention Capture Deceptive Patterns. CHI '23.

Why agency

• Maier, S.F. & Seligman, M.E.P. (2016). Learned Helplessness at Fifty: Insights from Neuroscience. Psychological Review, 123(4), 349–367.
• Baratta, M.V., Seligman, M.E.P. & Maier, S.F. (2023). From helplessness to controllability: toward a neuroscience of resilience. Frontiers in Psychiatry, 14, 1170417.
• Bandura, A. (2006). Toward a psychology of human agency. Perspectives on Psychological Science, 1(2), 164–180.
• Bandura, A. (2001). Social cognitive theory: An agentic perspective. Annual Review of Psychology, 52, 1–26.
• Bandura, A. (1997). Self-Efficacy: The Exercise of Control. W.H. Freeman.

How Strive works

• Gollwitzer, P.M. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist, 54(7), 493–503.
• Gollwitzer, P.M. & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis. Advances in Experimental Social Psychology, 38, 69–119.
• Oettingen, G. & Gollwitzer, P.M. (2010). Strategies of setting and implementing goals. In Maddux & Tangney (Eds.), Social Psychological Foundations of Clinical Psychology, 114–135.
• Ryan, R.M. & Deci, E.L. (2017). Self-Determination Theory: Basic Psychological Needs in Motivation, Development, and Wellness. Guilford Press.
• Ryan, R.M. & Deci, E.L. (2000). Self-determination theory and the facilitation of intrinsic motivation. American Psychologist, 55(1), 68–78.

Design trade-offs

• Schmitz, B. & Wiese, B.S. (2006). New perspectives for the evaluation of training sessions in self-regulated learning. Zeitschrift für Psychologie, 214(2), 93–104.
• Rittle-Johnson, B. (2006). Promoting transfer: Effects of self-explanation and direct instruction. Child Development, 77(1), 1–15.
• Järvelä, S. & Hadwin, A. (2013). New frontiers: Regulating learning in CSCL. Educational Psychologist, 48(1), 25–39.
• Järvelä, S., Malmberg, J. & Koivuniemi, M. (2016). Recognizing socially shared regulation by using the temporal sequences of online chat and logs in CSCL. Learning and Instruction, 42, 1–11.
• Armitage, C.J. (2005). Can the theory of planned behavior predict the maintenance of physical activity? Health Psychology, 24(3), 235–245.
• Dai, H., Milkman, K.L. & Riis, J. (2014). The fresh start effect. Management Science, 60(10), 2563–2582.

The broader framework

• Emirbayer, M. & Mische, A. (1998). What is agency? American Journal of Sociology, 103(4), 962–1023.
• Sen, A. (1999). Development as Freedom. Oxford University Press.
• Bourdieu, P. (1977). Outline of a Theory of Practice. Cambridge University Press.
• De Beauvoir, S. (1947/1976). The Ethics of Ambiguity. Citadel Press.
• Priestley, M., Biesta, G. & Robinson, S. (2015). Teacher Agency: An Ecological Approach. Bloomsbury.
• OECD (2019). OECD Learning Compass 2030 Concept Notes.