Separate efficient from empty deference in vote rationales

Determine whether low-information, deferential rationales in sequential Wikipedia Requests for Adminship elections represent efficient deference to evidence already documented by earlier voters or empty deference that substitutes the crowd’s judgment for independent scrutiny, using human-coded validation of the vote-rationale measure.

Background

The paper introduces vote–text divergence to measure cases in which a decisive vote is paired with little evidentiary substance. However, a short rationale such as “Support, per above” may reflect either an absence of independent scrutiny or a rational decision not to repeat evidence already entered into the discussion. Text alone cannot distinguish these interpretations, although both reduce the extent to which the raw vote margin represents distinct evidentiary contributions.

The authors identify human coding of the released stratified comment instrument as the means of adjudicating this distinction. Resolving it would clarify the construct validity of vote–text divergence and determine whether the observed late-vote pattern reflects empty conformity, efficient non-repetition, or both.

References

We cannot fully separate efficient from empty deference with text alone. We do note that even efficient deference does not restore the margin as a measure of independent scrutiny, since a count of eighty supports in which most defer still rests on a handful of independent assessments. That separation is exactly what the released human-coding instrument is for, and it is the paper's main remaining validity question.

— Endorsement Without New Evidence: How Sequential Voting Inflates Mandates in Online Community Governance  (2609.09321 - Chen et al., 8 Sep 2026) in Section 5.5, “Alternative explanations and construct validity”