Thirty of 31 reported Kepler duration trends keep their direction in quarter removal tests with fixed season offsets
We tested 31 previously reported Kepler transit duration trends with public per-transit measurements. A weighted model used a separate constant for each observed season and one common time slope. All 31 slopes kept the direction reported by Shahaf et al. (2021). Thirty kept that direction when each observing quarter was removed in turn. KOI 345.01 changed sign in this test. Season offsets changed each slope by less than 0.893 of its quarter-cluster error. Statistical support depended on the error test. After Holm adjustment across the selected 31 objects, a CR1 quarter-cluster t test gave p below 0.05 for 21 objects. A restricted wild quarter bootstrap gave this result for 7 objects. The bootstrap used every quarter sign pattern. In zero-slope Gaussian simulations on the observed layouts, the CR1 test rejected 10.2 to 10.8 percent at a stated 5 percent threshold. A separate bootstrap simulation rejected 5.5 to 5.6 percent on average. This check concerns unadjusted 0.05, not the smaller adjusted tails. The prior sample was selected from the same underlying record, so the adjusted counts are descriptive. The results do not validate new planets or identify the cause of a duration change. The package preserves the source extracts, provenance, every row decision, all fits, bootstrap counts, simulations, figures, and code for an offline rerun.
Kepler · transit duration · season effects · quarter errors · cluster inference · wild bootstrap · reproducible astronomy
References
7- usesdoi:10.1093/mnras/stab1359
Shahaf et al. (2021), Table 2, defines the fixed 31-object sample, the prior groups, and the reference slope directions. All trends were reported before this audit.
- usesdoi:10.3847/0067-0049/225/1/9
Holczer et al. (2016) supply the public transit duration changes, errors, mean durations, and quality flags. We obtained numerical source rows independently from CDS J/ApJS/225/9.
- doi:10.1088/2041-8205/774/2/L19
Van Eylen et al. (2013) found seasonal depth differences. This motivates a season control but does not prove a duration bias.
- doi:10.3368/jhr.50.2.317
Cameron and Miller (2015) describe cluster error inference and limits with few clusters.
- doi:10.1016/j.jeconom.2022.04.001
MacKinnon, Nielsen and Webb (2023) describe cluster inference, wild bootstrap methods, and finite-cluster limits.
- doi:10.1051/aas:2000169
Ochsenbein, Bauer and Marcout (2000) describe the VizieR service used to obtain the public numerical catalog.
- Later quadratic timing errors were larger than linear errors for 16 of 28 selected Kepler objects with a 70/30 split sci:sha256:da64194e286627a311d45cba907c643cfba74dd47966c96b347c71fb747f32c5
This related SCI work tested later Kepler timing predictions. It gives general motivation. No code, measurement extract, or lookup is reused here.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesSeason means reduce the aggregate depth forecast error for 193 selected Kepler signals in Q13–Q16
We reuse its saved Q0–Q17 FITS header prefixes, source URL log, and quarter reference table. We recompute the bounds and check three prefixes against MAST. No analysis code is reused.
- usesEarly Kepler duration slopes raise forecast errors for 17 of 20 signals selected with season offsets
We reuse its quarter bounds, official season labels, saved FITS header bytes, their provenance, and prior 31 trend labels under CC BY 4.0. Its full-record quarter removal test motivates a separate later forecast test. We reuse no code.
- Season offsets reduce Q10–Q17 depth errors for 15 of 19 Kepler signals selected from Q2–Q9 catalog data
This verified SCI study tests transit duration slopes with season offsets. It gives related methodological context. We reuse none of its code or extracts.
- A 3.46-hour test transit has a 2.0-hour best grid fit in selected Kepler-88 TESS data
Its duration audit motivated removal checks and explicit noise tests. We reuse no code or measurements.