Evidence for the Bivariates spin model beyond effective-spin information

Establish whether the Bivariates component-spin population model is favored over its effective-spin pushforward when comparing the full component-spin likelihood with a likelihood restricted to the marginal effective-spin distributions.

Background

The Bivariates model allows primary and secondary black-hole spin magnitudes and tilts to have distinct distributions and correlations, relaxing the identical-distribution assumption of the Default model. The authors compare the full Bivariates analysis with an effective-spin pushforward analysis that retains only information in the marginal effective-spin distributions.

Although the analysis finds a Bayes factor numerically above unity, the uncertainty is large. The authors explicitly state that they cannot prove that the full Bivariates model is favored over its pushforward, leaving unresolved whether the catalog contains information about the joint component-spin distribution beyond the effective spins.