Depends on what you're trying to normalize the data to, in the case of the 0.75 exponent this was done for aerobic capacity to body mass. The question you might ask yourself is whether the power output on an ergometer should be expressed simply as Watts or as watts/kg muscle mass (muscle mass recruited to develop power). On a cycle ergometer expressing power output based on body mass is meaningless because the body is supported by the bike.

The usefulness of scaled/normalized power output is dependent upon how the data are expected to be used. Depending upon the purpose of the scaling, such as to compare individuals without the confounding effect of body mass, it may be appropriate to explore other options for scaling such as relative to lean (as opposed to total) body mass. I disagree with Derek in that I believe expressing scaled power output for cycle ergometry may be of use depending on the reason for scaling. It would allow you to more accurately compare individuals in certain ways, but out of the laboratory, unless the cycling is being performed on an incline, obviously the athlete with the highest absolute power output is going to perform best regardless of their body mass. In a time trial situation, clearly scaling power output is of less value.

The method of allometric scaling (e.g. lean vs. whole body mass) and the specific metric to be scaled will determine the appropriate allometric exponent. Allometric exponents have been empirically derived for a variety of populations (such as for men, for example see http://www.ncbi.nlm.nih.gov/pubmed/23364295 and women, see http://www.ncbi.nlm.nih.gov/pubmed/21462687 , http://www.ncbi.nlm.nih.gov/pubmed/23900902), as well as for different exercise modalities (for example see http://www.ncbi.nlm.nih.gov/pubmed/21701284 , http://www.ncbi.nlm.nih.gov/pubmed/22327083 ).

Thank you for your answers. I want to compare persons with different body composition and/or overweight vs. normal.