The methodology seems strange. Questionnaires were compared based on whether yoga students had just done yoga, vs not in 24 hours (but presumably do yoga regularly). So essentially, there's a statistically significant correlation with time from last session, but it doesn't say anything about correlation with frequency of sessions.

And as usual, correlation is not causation. Was there a control group with another physical activity? Reading? Solving Sudoku puzzles? One could also postulate a perfectly reasonable-sounding theory that people's egos get bigger with any activity where there's a surge in endorphins, regardless of whether that activity is yoga or not.