Conference findings withdrawn pending revision
A post-conference audit found material problems in the analytical pipeline behind the version presented at AERA 2026. I have withdrawn the earlier findings and slides from this site rather than leave superseded statistics in circulation.
The analysis is being rebuilt from the official federal release with corrected record linkage, field classification, missingness handling, and first-digit diagnostics. A corrected public version will replace this notice only after the revised work has completed its review and publication process.
The broader research question remains important: which fields in the federal ESSER expenditure release are sufficiently observable and well-structured for a particular analysis, and which require explicit sensitivity checks? The answer needs to be based on the corrected workflow, not the conference pipeline.