Analysis plan (pre-registration): chance-alignment and measurement-error verification of the Phygas stage-1 screening

Analysis plan, written before any result of the verification runs was inspected, for the Phygas stage-1 screening of Gaia DR3 stars against young open clusters (volunteer computing, BitBOINC). It fixes in advance how 20 chance-alignment control runs and 20 measurement-error repeats of the whole catalogue are used to decide whether the number of slow, non-member star-cluster passages in the real data exceeds the chance level, and for which clusters: the data and runs, the primary outcome, the tests (an aggregate empirical test and per-cluster t tests with Benjamini-Hochberg correction), a robustness criterion against measurement errors, checks that must pass before any result is interpreted, the rules for what will be claimed, and the known limitations. It also discloses what had been seen before the plan was written: one real run and one single-draw pilot control, in which the excess at the lowest relative speeds was found to be largely known cluster members, which is why members are counted separately. Planned venue of the results: Physics and Astronomy Reports (Istanbul University). Every task of the verification runs is computed by two different volunteers and accepted only when both results agree; all contributors are credited by BOINC user name. Changes to this plan after the date of this record will be listed as deviations.