Too small or too low? New evidence of the 4-factor model