This is the second article of a two-part interview with Dr. Brian Jones. Dr. Jones has a PhD in exercise science and is a full-time professor at the University of Louisville where he teaches both undergrad and graduate courses. He approaches his classes from a scientific standpoint with an emphasis on critical thinking.
In a nutshell, what is science? Does science really prove anything?
Science is a process. It is a system for evaluating information based on formulating a hypothesis, carefully testing that hypothesis through data collection and analysis, and revising the hypothesis. If the hypothesis withstands the researcher’s attempt to falsify it then it tentatively stands supported by the research. Nothing in science is ever truly “proven” correct. Scientific fact represents the hypothesis and/or theory which currently has the most supporting evidence and allows us to most accurately predict what will occur in the real world.
Under what conditions are quasi-experimental research designs preferred over experimental research?
A quasi-experiment is an experiment is without a control group. From a purely methodological standpoint, control groups are absolutely necessary and thus the quasi-experiment is an inferior research method. If researchers cannot compare the intervention group or condition to a control then it is difficult to attribute causality to the experimental treatment. However this type of design is desirable and necessary for reasons of ethics. New treatments or drugs can never be compared to no treatment, but must always be compared to the normal standard of care. Intentionally withholding treatment simply for the purposes of research, as was done in the Tuskekee syphilis study in 1932, is unethical and inhumane.
Why are anecdotes considered very weak forms of evidence (if evidence at all)? Many people seem to think that anecdotes (some refer to as “real world evidence”) are superior to what is found in the lab.
Anecdotal evidence comes from experience but lacks the scientific rigor of control, careful manipulation of variables, and statistics. Someone shouldn’t dismiss anecdotal data entirely but should carefully evaluate its source. Compare a product testimonial with a case study published in the Journal of the American Medical Association. Neither are scientific studies but the published case study is much more legitimate than the testimonial. It comes from a medical professional that is not attempting to sell you anything. The problem comes with attempting to generalize from such reports due to the lack of statistical analysis. The average untrained scientist probably gives more weight to anecdotal evidence simply because they do not understand statistics and scientific methodology.
How do you teach your students that correlation does not necessarily imply causation?
To teach that correlation does not necessarily imply causation I give them examples of spurious correlations. For example a study demonstrates a significant correlation between sunspots and the number of Republicans in the senate since 1959. Is there any causal link there? Of course not. I also discuss how the cigarette industry was able to use the argument against correlation as causality to deny the link between cigarettes and lung cancer for many years. Despite the high rates of lung cancer among smokers cigarette manufacturers were able to win many lawsuits because they claimed this evidence did not amount to proof. Experimental research in the late 90’s made the link between smoking and cancer undeniable.
Photo by CIAT, available under a Creative Commons attribution license.