338 PA over many years is not a large sample size by any means. That is about half a season worth of PA spread out over 6+ seasons. That really isn't relevant because that could easily be explained by random variation.
If 6+ seasons of data collection isn't enough, then why bother at all? You've naged and moaned about Roger Bernadina since before the All-Star break and he didn't have any more PA's. If 300+ PA aren't enough to spot a trend, then maybe the data isn't being collected or analyzed properly. But to say 6 years isn't enough time is ridiculous.