Experimental design and data evaluation in orthopaedic research.

It has been said that we can be of assistance to one person at a time in the operating room, several individuals at a time in the classroom, and multitudes through our work in the laboratory. This final statement is true only if the conclusions we draw from our studies are valid. Failure to adhere to the scientific method can invalidate the conclusions we report. This can occur from failure to define properly the question to be studied, failure to review the available literature adequately, improper randomization of the experimental subjects, introduction of investigator bias into the data collection process, and failure to include appropriate controls. Use of improper statistics can also invalidate the results of any study. Frequently, this error comes from multiple use of the Student's t test. When this occurs, the probability of reporting a difference when none exists increases proportionally to the number of comparisons performed. Random statistical comparisons among groups are inappropriate and can lead to a substantial increase in the risk of concluding that two sets of data differ when in fact no difference exists. Investigators who use statistics inappropriately and then restate their original hypothesis are guilty of scientific dishonesty. This inappropriate use of statistics represents a flagrant disregard of the scientific method of problem solving.