H012 — Vocabulary Bias

Status Before Review

Ready for limited review.

Why This Hypothesis Is In Scope

Comparator v2 and v3 directly bear on wording-driven agreement inflation.

Evidence Available

• v2 flat semantic and v3 dimensional comparison

Supporting Observations

• dimensional normalization reduced over-merging compared with v2

Challenging Observations

• provided schema still shapes outputs
• unresolved dimensional matches remain high

Threats To Validity

• vocabulary effects may be confounded with schema pressure
• unresolved items may hide residual vocabulary bias

Instrument Issues

• unresolved dimensional count remains very high

Reviewer Assessment

Choose one:

• insufficient_evidence
• tentatively_supported
• tentatively_challenged
• mixed_evidence
• blocked_by_instrument_issue
• defer_until_next_ecr

Rationale

Next Evidence Needed

• unresolved-item sample
• comparison of schema-shaped versus freer outputs

Confidence Update Recommendation

Do not update confidence here.

Options:

• no_update
• prepare_increase_for_review_board
• prepare_decrease_for_review_board
• prepare_split_hypothesis
• prepare_suspend_hypothesis
• prepare_retire_hypothesis

Notes