Too many use transforms routinely then back-transform without consideration of transformation bias. Transforms (log, for example) can sometimes make the mean of Treatment B larger than the mean of Treatment A, when just the opposite is true in the original scale. For a graph of the reversal of means and examples of papers that use the bias backtransforms, see Parkhurst, D. F. (1998). "Arithmetic versus geometric means for environmental concentration data." Environmental Science and Technology 32(3): 92A-98A.