Does the left side of square taper cranks loosen more than the right side?
Or is it that observations in general include other than square taper cranks? A two piece crank can only get loose on the left side so if that is included in the stats and observation, then that will skew the data if one tries to subset only square tapers.