This paper studies linear regression when a latent regressor is observed through multiple noisy, smooth, and potentially nonlinear measurements. It fixes the latent scale by requiring a consensus measurement function to be linear, then bounds curvature heterogeneity across sources relative to slope. The resulting structural coefficient lies in a closed-form, source-loading-invariant interval centered at a symmetric cross-source estimator. With at least four measurements, the bound is estimable using a split-instrument auxiliary regression. An application combines six AI occupational-exposure measures with 8.88 million American Community Survey person-year observations from 2015–2024. Five retained sources produce a consensus coefficient of -0.239, while the authors caution that this is measurement reconciliation, not a causal estimate of AI displacement.
No heat snapshots are available in the last 24 hours.