I opened the article had a look at the graphs and closed it again. The y axis of the graph is deliberately chosen to make the effect look stronger and this sort

I opened the article had a look at the graphs and closed it again. The y axis of the graph is deliberately chosen to make the effect look stronger and this sort of thing gets an immediate no from me. Yes if you want to highlight a small effect on a large number you can change the yaxis, usually you indicate this by an interruption in the axis. Bit that's not the case here GPAs go from 1 (or 0) to 4 and any significant effect would show on such a scale (it would not look as large however). The other question is, how where the years chosen, is there a lot of fluctuations year to year? Bottom line, there might very well be a significant effect, however the presentation sets of all sort of alarm bells.