Why compare 3 years of young developing goalies. We all know that Reimer made great developments last season, for him only last season should be counted. We also should assume that Bernier was little prepared for last season, for Bernier I would only count 2013/14 for comparisons. Andersen might be better than last season because he was injured in November.
That same argument gets said about any youngish player who has a career year, and it's often just that. It's easy to accept narratives as established truth, and it can lead you astray. Couple of things to consider:
1) Reimer and Bernier are not
that young anymore.
2) The circumstances of their deployment hasn't changed drastically.
3) The reason you expand sample sizes is to eliminate the effect of streaks. What you are doing in your argument is try to show each goalie at their best, when that doesn't capture the whole of their game.
4) For example, you want to exclude Bernier's last season. But the mental weakness and such that ruined last season for Bernier is part of his make. By ignoring it, you ignore a crucial part of him as a goaltender.
5) In fact, the times that it actually helps accuracy of an evaluation to limit sample sizes are precious few. Usually when you do, you just help yourself get the conclusions you want to get, subconsciously or not.