Valid conclusions about user behavior should only be made within the research method used by each study. For example, if a researcher interviews academic faculty to determine if they prefer print or electronic sources, the conclusions should only state what faculty prefer, not what faculty actually use.
