Reproducibility is not construct validity: LLM measurement of institutionally situated communication
The paper argues that reproducible LLM-based measurements of institutionally situated communication may still lack construct validity, highlighting a methodological risk in AI evaluation.
The paper argues that reproducible LLM-based measurements of institutionally situated communication may still lack construct validity, highlighting a methodological risk in AI evaluation.
Auto-scored: A methodological critique that raises awareness of validity limits in LLM measurement, a minor framing risk.