Abstract :
A not unpopular thesis, when it comes to the confirmation of scientific theories, is that
data which were used in the construction of a theory afford poorer support for that
theory than data that played no role. Some compelling thought experiments have been
offered in favour of this view, not as proof but rather to add some intuitive plausibility.
In this paper I consider such thought experiments and argue that they do not support the
thesis; the perceived importance of prediction over accommodation, at least in these
cases, is illusory.
