Equivalence of structured and original prompt formats

Establish whether the structured answer-field rendering is equivalent to the original candidate-aggregation prompt for purposes of measuring accuracy and causal answer-field effects.

Background

The answer-field intervention changes the candidate presentation by adding an explicit reported-final-answer field and manipulating that field while retaining identical reasoning prefixes. The intervention finds that substituted answer values steer outputs, but masking does not measurably recover accuracy.

Because the structured rendering differs from the original prompt, the authors conduct a bridge comparison. Its confidence interval is not contained within the pre-fixed equivalence margin, so the paper cannot transfer the structured intervention's causal conclusions directly to the original-format experiment.

References

On the original aid materials the same pipeline reproduces the direction of both published effects while neither interval clears zero, which points at the materials without saying what in them carries the phenomenon. We offer that as a conjecture, not a finding.

— The Audit Decides the Verdict: Instrument Effects Rival Demographic Bias in LLM Decision Audits  (2609.09048 - Vohra et al., 8 Sep 2026) in Discussion and Conclusion

Equivalence with the original prompt was not established.

— Selection, Recombination, or a Fresh Solve? A Candidate-Free Control for Single-Pass Test-Time Aggregation  (2608.18379 - Farmanfarmaian, 18 Aug 2026) in Section 3.3, “A structured answer-field intervention”; Appendix F, “The structured answer-field intervention: details”

A full-cohort rescoring under a single symmetric extraction rule has not been carried out and remains the outstanding measurement; the convention correction, by contrast, has been applied to all 42 deployments.

— HEPToolBench 1.2: Testing How Reliably Language Models Can Drive Particle Physics Software  (2608.28232 - Singh et al., 28 Aug 2026) in Section 2.2, paragraph “Convention and extraction audit” (Sec. scoring-policy)