Deploying Surveys with AI Responsibly Tutorial
EvalCommunity Academy · Survey Deployment Series
Deploying Surveys with AI Responsibly
Higher response rates are not automatically better data. Apply the ETHICS framework before any AI-optimised deployment tactic reaches respondents.
📌 “Higher response rates are not automatically better data.” — ETHICS first. Always.
🛡️ The ETHICS Framework
Does the strategy create unequal response burdens? Do algorithmic decisions favour or disadvantage specific groups?
Would respondents understand why they’re contacted this way? Is the logic explainable in plain language?
Could the strategy cause annoyance, anxiety, privacy concerns, or distress? What is the worst-case respondent experience?
Does the approach respect the original consent agreement? Can respondents opt out easily and clearly?
Does any element pressure respondents to participate against their inclination? Urgency, social norms, incentive escalation?
Does this build or erode trust? Are you acting as a responsible guardian of respondents’ trust and data?
1. ETHICS Deployment Analyzer
Paste an AI-generated deployment strategy or describe a specific tactic. The analyzer applies all six ETHICS lenses and identifies red lines. You make every final decision — no tactic is approved or rejected without your review.
⚖️ Human-in-the-loop assessment
Each lens requires your professional judgment. AI flags concerns → you decide.
All decisions are yours — AI cannot decide where invitation ends and pressure begins.
E · Equity
Does this create unequal contact burdens? Which groups may be disadvantaged?
T · Transparency
Is the logic explainable without jargon? Are reasons disclosed?
H · Harms
Could this annoy, stress, or invade privacy? Worst-case scenario?
I · Informed Consent
Does it exceed what respondents agreed to? Is opt-out easy?
C · Coercion Risk
Urgency, social norms, escalating incentives → pressure or invitation?
S · Stewardship
Does this build or erode trust with participants and communities?
🚨 Red Line Detector — Never cross these lines
❌ Deception or hidden tracking · ❌ Contact frequency that feels harassment · ❌ Incentive escalation that mimics bribery · ❌ Ignoring opt-out requests · ❌ Using urgency to manufacture consent.
Human judgment required: If any red line is detected, the tactic must be redesigned before field deployment.
🤖 8 AI Prompts for Ethical Survey Deployment
“Review this survey contact strategy. Identify any group that may receive more reminders or fewer based on connectivity, language, or digital literacy. Suggest equity adjustments.”
“Flag urgency phrases, social proof pressure, or escalating incentives that could pressure participation. Rewrite to an invitation tone.”
“Given original consent: [paste text]. Does this deployment tactic introduce new channels or frequencies not disclosed? Propose opt-out language for every contact.”
Tactic Builder
Design AI-assisted deployment that respects ETHICS: equitable contact intervals, plain-language transparency, no harm triggers, explicit opt-out, no urgency framing, and trust-centered messaging.
Audit Log & Stewardship
Maintain an ETHICS audit trail for every AI deployment: record prompts, red line reviews, consent versions, and final human approval.
📋 ETHICS Quality Assurance Checklist
Practice Lab
Simulate tactics: see red flags before real deployment.
Red Flags catalog
“Last day”, “only 2 spots”, urgency countdowns → coercion.
Reference Card
Printable ETHICS one-pager for field teams.
EvalCommunity Academy · Survey Deployment Series
Deploying Surveys with AI Responsibly Toolkit
Full ETHICS Analyzer · Tactic Builder · Red Line Detector · 8 AI Prompts · Audit Trail · Fully Offline
Get the complete framework, red line detector, and deployment prompts used by responsible M&E teams.
EvalCommunity Tutorial — ETHICS framework · Deploying Surveys with AI Responsibly. Part of the AI in Monitoring & Evaluation certificate program.
Higher response rates are not automatically better data — ETHICS first. Always.
📎 Learn more: evalcommunity.com/toolkits/deploying-surveys-with-ai-responsibly-toolkit
