2016MNRAS.457.4384S

Motivated by recent discussions, both in private and in the literature, we use a Monte Carlo simulation of planetary systems to investigate sources of bias in determining the mass-radius distribution of exoplanets for the two primary techniques used to measure planetary masses - radial velocities (RVs) and transit timing variations (TTVs). We assert that mass measurements derived from these two methods are comparably reliable - as the physics underlying their respective signals is well understood. Nevertheless, their sensitivity to planet mass varies with the properties of the planets themselves. We find that for a given planet size, the RV method tends to find planets with higher mass while the sensitivity of TTVs is more uniform. This 'sensitivity bias' implies that a complete census of TTV systems is likely to yield a more robust estimate of the mass-radius distribution provided there are not important physical differences between planets near and far from mean-motion resonance. We discuss differences in the sensitivity of the two methods with orbital period and system architecture, which may compound the discrepancies between them (e.g. short-period planets detectable by RVs may be more dense due to atmospheric loss). We advocate for continued mass measurements using both approaches as a means both to measure the masses of more planets and to identify potential differences in planet structure that may result from their dynamical and environmental histories.