In article , "Robert L.
Oldershaw" writes:

There is zero scientific justification for throwing out the pre-2012
data.
---------------------------------------

In the 3 recent responses there seems to be an acknowledgement that
the accuracy of stellar mass estimates does tend to improve with time.

Again, no-one has ever doubted this.

So I am at a loss to understand the very vigorous effort to discredit
my choice of a start date (data posted to arXiv.org or published
during or after 2012) for accepting estimated masses in this specific
test.

There are two reasons. Because, without further justification, it is
arbitrary. If accuracy increases with time, unless there is a very
dramatic and sudden increase in accuracy in 2012, then there is no
objective reason to choose 2012. Also, you said yourself that you chose
this date because you noticed that it seemed to work better for you.
This cherry-picking is unscientific. Consider: If the date is a free
parameter, then just by chance some date will support some hypothesis
better. Choosing that date BECAUSE it supports your own hypothesis is
wrong. Similarly, choosing one statistical test BECAUSE it supports
your hypothesis is wrong.