Is there evolution in the infrared Tully-Fisher relation? Comparing two linear regressions

In a recent paper, Puech and co-workers compared K-band Tully–Fisher relations derived for nearby and distant galaxies, respectively. They concluded that the two relations differ, and deduced that there is evolution in the Tully–Fisher relations. The statistical comparison between the two regression lines is re-examined, and it is shown that the statistical test used gives non-significant results. It is argued that better results can be obtained by comparing the ‘inverse’ Tully–Fisher relations, and it is demonstrated by two different methods that the nearby- and distant-sample relations do indeed differ at a very high significance level. One of the statistical methods described is non-parametric, and can be applied very generally to compare linear regressions from two different samples.