What is Research Ethics?

Research ethics is the set of principles governing how research treats its participants and its findings: the duties owed to the people studied (respect, protection, honesty about what's happening) and to the people who will rely on the results (accuracy, transparency, resistance to convenient distortion). The canonical framework comes from the Belmont Report's three principles: respect for persons (autonomy, expressed chiefly through informed consent), beneficence (maximise benefit, minimise harm), and justice (fair distribution of research's burdens and benefits). Academic research operationalises these through ethics review boards; product research rarely has one, which raises the bar for the team's own judgment rather than lowering it.

The Participant-Facing Duties

Consent, genuinely informed. Covered fully under its own entry: plain-language understanding of what's collected and why, specific permission for recordings and their uses, real freedom to withdraw.

Privacy and data stewardship. Session recordings, voice, and faces are personal data. Collect the minimum the question needs, restrict access, retain only as long as stated, and anonymise in reporting: quotes stripped of identifiers, clips shared only with the permission that covers the sharing. Under regimes like GDPR this is law; before it is law, it is the beneficence principle applied to data.

Fair treatment and compensation. Participants' time has value; pay it fairly and promptly, including for sessions that fail for technical reasons. Screen honestly (people deserve to know what they're applying for), avoid recruiting that exploits desperation, and keep incentives below the level that overrides judgment about sensitive asks.

Protection from harm. Product research rarely risks bodily harm, but distress, embarrassment, and professional exposure are real: the participant whose struggling session becomes a company-wide punchline, the employee whose candid feedback reaches their manager with their name attached. Beneficence means designing so candour cannot hurt the candid.

The Truth-Facing Duties

Ethics extends past the session into the analysis. Reporting findings that the data actually supports; showing the disconfirming cases alongside the flattering quotes; separating what participants said from what the team hopes; disclosing methods and sample honestly (who was studied, who wasn't, what the frame missed); and resisting the quiet pressure to launder a stakeholder's preference through a study designed to confirm it. Research commissioned to validate a decision already made, then presented as inquiry, deceives colleagues rather than participants, and it is the most common ethical failure in product research by a wide margin.

Everyday Practice

1. Build ethics into the tooling defaults.
Consent screens before recording, automatic notice of what's captured, access controls on response libraries: platforms like Ballpark ship these as defaults, and defaults do more ethical work than policies, because they operate when nobody is thinking about them.

2. Run a lightweight pre-study check.
Five questions before fielding: does consent cover everything collected? could participation harm anyone who is honest? is the incentive fair? is any group bearing the burden without the benefit? would we be comfortable if a participant read the study plan? Ten minutes, most studies pass, and the ones that don't needed the conversation.

3. Treat sensitive topics as a higher tier.
Health, money, children, employment: raise the consent clarity, lower the identifiability, and give participants explicit outs mid-study.

4. Keep an audit trail of permissions.
Which participants consented to what uses, recorded where the future clip-sharer will actually check. Consent that can't be looked up will eventually be exceeded.

The Takeaway

Research ethics is two loyalties held at once: to the people who lend themselves to your studies, and to the truth their participation produces. Consent plainly, collect minimally, store faithfully, compensate fairly, report honestly, and let the tooling's defaults carry the standard on the days attention lapses. Teams that practise this don't just avoid scandal; they get better data, because participants can tell when they're safe to be honest.