Can the results be explained by poor randomization and nonpertinent comparisons?