Effect size needs to be relativized by the error present from the study. There’s a probability distribution over multiplicative increases which probably has a decent amount of mass below 4x