2019MNRAS.489.2505M

We demonstrate that planet occurrence does not depend on stellar galactocentric velocity in the Solar neighbourhood. Using Gaia DR2 astrometry and radial velocity data, we calculate 3D galactocentric velocities for 197 090 Kepler field stars, 1647 of which are confirmed planet hosts. When we compare the galactocentric velocities of planet hosts to those of the entire field star sample, we observe a statistically significant (KS p-value = 10^–70) distinction, with planet hosts being apparently slower than field stars by ∼40 km s^–1. We explore some potential explanations for this difference and conclude that it is not a consequence of the planet-metallicity relation or distinctions in the samples' thin/thick disc membership, but rather an artefact of Kepler's selection function. Non Kepler-host stars that have nearly identical distances, temperatures, surface gravities, and Kepler magnitudes to the confirmed planet hosts also have nearly identical velocity distributions. Using one of these identical non-host samples, we consider that the probability of a star with velocity v_tot hosting a planet can be described by an exponential function proportional to e^(-v_tot/v0)_. Using a Markov Chain Monte Carlo sampler, we determine that v₀ >976 km s^–1 to 99 per cent confidence, which implies that planets in the Solar neighbourhood are just as likely to form around high-velocity stars as they are around low-velocity stars. Our work highlights the subtle ways in which selection biases can create strong correlations without physical underpinnings.