Opposition votes came later in 551 complete Wikipedia candidate windows from 2005 to 2013

We measured vote timing in a fixed Wikipedia administrator-request archive. We selected 551 candidate windows with complete dates, at most eight elapsed days, unique voters, one recorded outcome, and both vote signs. Each candidate had equal weight. The mean probability that opposition came after support was 0.5771, with half credit for equal-minute ties. The descriptive 95 percent candidate bootstrap interval was 0.5583 to 0.5960. The means were 0.5600 for failed candidates and 0.5972 for successful candidates. Both exceeded a reference that shuffled signs within candidates. Their ranking reversed when we required three votes of each sign. An expanded sample included missing dates. Sharp mean bounds were 0.5045 to 0.6265 overall and 0.4994 to 0.6332 for successful candidates. Thus, missing dates can remove the successful-group timing advantage. These bounds assume each short observed candidate record came from one request. The archive lacks request identifiers. This study repeats an earlier timing finding with a pair score and adds missing-date checks. It does not establish causal social influence. The package preserves source provenance, code, aggregate results, resampling draws, figures, arithmetic checks, and reproduction commands.

Wikipedia · collective decisions · vote timing · missing data · rank statistics · reproducible research