The cosmic distance duality relation ( CDDR ) , D _ { L } ( 1 + z ) ^ { -2 } / D _ { A } = \eta = 1 , with D _ { L } and D _ { A } , being the luminosity and angular diameter distances , respectively , is a crucial premise in cosmological scenarios . Many investigations try to test CDDR through observational approaches , even some of these also consider a deformed CDDR , i.e. , \eta = \eta ( z ) . In this paper , we use type Ia supernovae luminosity distances and galaxy cluster measurements ( their angular diameter distances and gas mass fractions ) in order to perform a Bayesian model comparison between \eta ( z ) functions . We show that the data here used is unable to pinpoint , with a high degree of Bayesian evidence , which \eta ( z ) function best captures the evolution of CDDR .