Free unknown vote choices allowed either US or China to lead in 80 of 1,872 UN country and year records from 2013–2022

We tested whether absent and missing votes can change the order of agreement with the United States and China. We used Harvard Dataverse release 30.0 of UN General Assembly voting data. The main test kept 886 selected roll calls dated 2013–2022 where both reference countries cast known votes. Known country votes stayed fixed. Each absent or missing country vote could be yes, abstain, or no. Abstention scored halfway between yes and no. We derived exact attained score limits. Of 1,872 country and year records with at least 30 known votes, 1,788 kept one strict order under every allowed choice. China kept higher agreement in 1,603 records. The US kept higher agreement in 185. Both orders were possible in 80 records. Three other records allowed a tie and one order. One record stayed tied. Of the 80 records with both orders, 77 had an absolute observed margin of at least five percentage points. A check limited to explicit country absences gave 78 records with both orders. A check that also freed absent reference votes gave 94. These are hypothetical vote completions, not confidence intervals or estimates of how absent countries would vote. The source has coverage gaps. The result does not establish allegiance, a preference trend, or a causal effect. The package contains source data, provenance, code, all bounds, endpoint choices, figures, a local plan, and repeat instructions. A separate calculation checked every score and witness. A clean source repeat matched all 19 prepared and result files.

United Nations · voting agreement · absence · missing data · partial identification · sensitivity analysis · reproducibility