Reproducibility Materials for "<b>Predicted Clarity, Coherence, and Monetary Relief in 94,335 Consumer Financial Protection Bureau Complaints"</b>

<p dir="ltr">This repository contains reproducibility materials for the study <b>“Predicted Clarity, Coherence, and Monetary Relief in 94,335 Consumer Financial Protection Bureau Complaints.”</b> The study examines associations between predicted complaint clarity and coherence and company-reported monetary relief in public U.S. Consumer Financial Protection Bureau complaints from 2023–2024.</p><p dir="ltr">Materials include analysis and revision code, non-narrative derived data, sanitized presentation-rating and factual-validation files, source-file hashes, fixed random seeds, model outputs, and machine-readable results for the supplementary and reviewer-requested sensitivity analyses.

These include sample-selection checks, duplicate-narrative analyses, interaction models, scorer-refitting bootstrap analyses, readability benchmarks, semantic-component sensitivity analyses, and richer narrative-length/finer-category specifications.</p><p dir="ltr">Complaint narratives are not redistributed because they may describe sensitive financial experiences. The analyses can be reconstructed from the publicly available CFPB Consumer Complaint Database using complaint identifiers, archived source hashes, and the supplied code and documentation.</p>