460-Question Benchmark Overview
How Lewsearch synthetic panels are evaluated against independently fielded survey instruments from Texas, California, and national tri-metro sources.
Lewsearch is evaluated against 460 independently fielded survey questions drawn from three source families: the University of Texas / Texas Politics Project (97 items), the Public Policy Institute of California (148 items), and a pooled Pew / Gallup / official canvass tri-metropolitan set (151 items). A separate pre-registered held-out batch (14 scored items after automated exclusions) was sourced after training and calibration were frozen.
Every published error rate is out-of-sample. We fit calibration on held-out folds and score on data the calibrator never saw during fitting. Pooled calibrated mean absolute error across the full 460-question pool is 7.47 percentage points on marginal response proportions.
Panel-level results (ex-electoral)
- Texas (UT/TPP): 4.88% calibrated MAE on 97 questions
- California (PPIC): 7.77% on 148 questions
- Tri-metro legacy set: 7.86% on 151 questions
- Best subdomain: Texas political approval at 3.43% (n=18)
What this is and is not
These numbers describe marginal distribution accuracy on real survey items administered as of each instrument's field date. They do not claim joint-distribution fidelity, individual-level prediction, or replacement of probability-sample polling for high-stakes inference.
Full item-level results and proprietary system internals are not published. See our transparency boundary note for what we disclose publicly versus under NDA.
Product validation
Customer-facing benchmark tables and live predictions are on lewsearch.com/methodology. Due diligence materials available on request.