Key Principle
Behavior change design is inherently directive — it attempts to influence what people do. This power creates three categories of ethical obligation:
1. Research Ethics: Protecting participants from harm during the diagnostic phase. 2. Strategy Ethics: Ensuring the intervention is evidence-informed and respects participant agency. 3. Design Ethics: Avoiding manipulation and testing with real learners rather than SME proxies.
Three-Phase Ethical Checklist:
Research Phase:
- Have you understood the context fully, including systemic factors that are causing the behavior gap?
- Are you targeting the right people, or are you asking individuals to compensate for system failures?
- Have you protected research participants from disclosure risk, especially for sensitive behaviors (health, legal, social stigma)?
- Have you used multiple data sources to avoid selection bias?
Solution Strategy Phase:
- Is your intervention strategy evidence-informed (not just familiar or conventional)?
- Does your solution address the actual root cause (COM-B-diagnosed), not the most convenient one?
- Does your solution address the system, not just the individual?
- Are you proposing training when a policy, process, or environmental change would be more effective?
Design Phase:
- Have you tested your design with actual target learners (not just SMEs or stakeholders)?
- Have you considered the stress-learning curve — is your design emotionally calibrated for learning?
- Does your design avoid dark patterns (deceptive practices, false urgency, manufactured vulnerability)?
- Is the consent structure honest about what participants are experiencing?
Why This Matters
Context Failure: Designing a training intervention without understanding the organizational system can cause harm. A manager trained in giving critical feedback but working in a culture where giving feedback has historically led to retaliation will not apply the training — and may be harmed by attempting it. Training individuals to behave differently in a hostile system can expose them to consequences the designer never anticipated.
The Stress-Learning Curve: A small amount of emotional activation focuses attention and improves encoding. A large amount blocks learning and triggers defensive responses. Highly immersive, emotionally intense designs (VR trauma simulations, vivid consequence visualization) can exceed the threshold and produce trauma-response rather than learning. The designer's responsibility is to calibrate intensity and provide recovery mechanisms.
Iterative Learner Testing vs. SME Review: Subject matter experts validate content accuracy — they cannot validate whether the design produces the intended behavioral outcome. Only real learners (members of the target audience, experiencing the design under realistic conditions) can reveal whether the design works. SME approval is a content quality check, not a behavioral effectiveness check.
Dark Patterns in Behavior Change Design: Techniques from behavioral economics and UX (defaults, scarcity messaging, social proof manipulation, commitment escalation) can be applied to produce compliance without informed consent. The ethical question: would participants endorse this intervention if they understood fully what was being done and why? If not, it is a dark pattern.
Good Examples
Pilot with real learners before SME review (Chapter 13): A healthcare communication training module was reviewed and approved by clinical subject matter experts. When tested with actual nurses and residents, it produced anxiety rather than confidence — the scenarios were accurate but emotionally miscalibrated for learners who had no prior experience with the depicted situations. SME review could not have detected this; only real learner testing could.
Valor Nigeria research protocol (Chapter 15): Conducting sensitive HIV-related research with Nigerian men required anonymous participation (WhatsApp-based, no real names), careful framing of questions to avoid shame activation, and co-research methods that gave participants genuine agency over how their experience was interpreted. The research design protected participants from the exact harm the intervention was trying to address.
Teachers who pass but can't teach (Chapter 13): Assessment-based training that certifies learners based on written or verbal responses produces a visible "pass" rate with no guarantee of behavioral performance. Teachers who can correctly answer questions about pedagogical techniques may still fail in live classroom practice. The certification is the output; the behavior is the outcome. Designing for certification rather than behavioral outcome is an ethical failure.
Counterpoints
"It's not manipulation if it's for their own good" — this reasoning eliminates the ethical constraint entirely. The question is not whether the designer believes the outcome is beneficial but whether the participant would endorse the method if fully informed. Behavioral design techniques are tools; their ethical status depends on how they're used.
"SMEs signed off on it" — SME sign-off is not learner testing. The obligation to test with real learners cannot be discharged by proxy.
Key Quotes
"Good intentions don't exempt you from the obligation to test whether your design actually works and doesn't cause harm." — Julie Dirksen, Chapter 13: Designing Responsibly
"Dark patterns are a design choice. So is designing without them." — Julie Dirksen, Chapter 13: Designing Responsibly
Rules of Thumb
- Run the three-phase checklist before finalizing any intervention design.
- Test with real learners before any launch, regardless of SME approval. One real-learner session will reveal more than any number of expert reviews.
- Calibrate emotional intensity explicitly. Decide deliberately how much stress the design introduces and provide recovery mechanisms for high-intensity experiences.
- Ask the consent test for every nudge-style intervention: "Would participants endorse this if they understood it fully?" If uncertain, err toward disclosure.
- When targeting individuals requires them to behave differently than the system rewards, identify and address the systemic risk before proceeding.
Related References
- COM-B Diagnosis and Behavior Analysis — Context failure is a COM-B Systems Opportunity problem; ethical design requires its diagnosis
- Implementation Playbook: From Diagnosis to Deployment — Ethical checklists are embedded in the implementation sequence