The standard cosmology strongly relies upon the Cosmological Principle, which consists on the hypotheses of large scale isotropy and homogeneity of the Universe. Testing these assumptions is, therefore, crucial to determining if there are deviations from the standard cosmological paradigm. In this paper, we use the latest type Ia supernova compilations, namely JLA and Union2.1 to test the cosmological isotropy at low redshift ranges ($z<0.1$). This is performed through a Bayesian selection analysis, in which we compare the standard, isotropic model, with another one including a dipole correction due to peculiar velocities. We find that the Union2.1 sample favors the dipole-corrected model, but the opposite happens for the JLA. Nonetheless, the velocity dipole results are in good agreement with previous analyses carried out with both datasets. We conclude that there are no significant indications for large anisotropic signals from nearby supernova compilations, albeit this test should be greatly improved with the upcoming cosmological surveys.