Selected supplemental Kepler DR25 fits have a median radius ratio error of −11.1% at injected b ≥ 0.8 and expected MES 8–20

We measure radius ratio errors and geometry selection in public Kepler DR25 INJ1 data. Eligible stars have temperatures of 3900–7000 K and log surface gravity above 4. Injected signals have periods of 3–300 days, at least four valid transits, and durations of 1.5–15 hours. Expected MES is 8–20, injected radius ratio is at most 0.1, and injected b is between 0 and 1. Of 25,716 eligible injections, 21,122 have recovered supplemental fits with periods within 1% of the injected value. The median radius ratio error is +3.60% at injected impact parameter b < 0.5 and −11.11% at b ≥ 0.8. The high-b group has 2,035 fits. Its 95% interval from 5,000 bootstrap draws of whole sky groups is −11.52% to −10.77%. This interval assumes independent sky-group sampling. A cut that retains fitted b < 0.8 retains 80.44% of the injected high-b group. Those retained cases have a median radius error of −13.07%. The selected sample has a zero aggregate median at the source precision. The high-b underestimate persists in all four tested MES bins and in the planned duration and transit-count checks. If both source radius ratios were rounded to the nearest 0.0001, the high-b median error must remain negative. We supply the source snapshots, full selection flags, response tables, code, figures, and independent checks. These are conditional results for supplemental point fits to synthetic circular transits. They do not give a correction for real planet radii or test the separate catalog posterior fits.

Kepler · DR25 · transit injection · radius ratio · impact parameter · selection · supplemental fits · reproducible data