New AMV method cuts survey questions by 50% using AI interviews
University of Washington researcher proposes Adaptive Matrix Validation to make surveys conversational
Tyler H. McCormick introduces Adaptive Matrix Validation (AMV) for AI-assisted interviews. Respondents describe experiences naturally while AI maps answers into structured data; a small set of randomized validation questions then corrects measurement errors statistically. The paper develops estimators for item means, subgroup estimates, and regression coefficients. Simulations—including a design-calibration simulation, an American Time Use Survey emulation, and a CHAMPS verbal-autopsy narrative study—show when sparse validation can improve precision and when it cannot.
- AMV reduces structured survey questions by 50–75% by leveraging AI to map natural conversations into tabular data.
- The estimator uses a two-step correction: cross-respondent calibration followed by individual-level error correction from sparse validation questions.
- Simulations on the American Time Use Survey and CHAMPS verbal-autopsy data show maintained or improved precision for means, subgroups, and regression coefficients.
Why It Matters
AMV could slash survey costs and fatigue while keeping data quality high, making large-scale research faster and more inclusive.