Why Poor Predictions are a Justification for More (not less) Research

Stevens’ criticisms of political science in her NY Times OP-Ed were not criticisms of political science per se; rather, they were criticisms of the scientific process in general. It is a frustrating process. It would be nice if we could identify all the causes of a particular outcome after doing a limited amount of research. But it actually takes time to identify causes of any natural or social phenomenon. We often develop models that only partially reflect the real world- these models will inevitably give us inaccurate predictions. But those inaccurate predictions give us an opportunity to re-evaluate what we think we know about the world. They allow us to encourage us to (1) find better ways to quantify “qualitative” concepts, (2) develop and use better statistical models that meet the realities of the outcomes we are examining, (3) determine whether we left something out of our analyses, and (4) determine whether we included things in our model we should not have. In other words, poor predictions tell us we need to do more research.

Just because a research program does not lead to all the answers we were looking for does not mean the program is a waste of money. I cannot imagine applying this same logic to other sciences. For example, between 1878 and 1880, Thomas Edison worked on over 3,000 different theories trying to develop an efficient light bulb. He learned 2999 wrong answers until he discovered 1 workable answer. Then other researchers continued to do work on the light bulb and made improvements. And ultimately that is what the goal is- to continue to do research until we find workable answers. The same logic applies to medical research. Doctors still do not know all the causes/cures for several deadly diseases/viruses, and several cancers. Does this mean we should stop funding their research? Quite the contrary- we continue to fund their research in hopes that one day we do find the cures. In short, science is a collaborative and cumulative process.  Bad results and poor predictions are stepping stones to better theories, refined hypotheses, fuller data, and better predictions. 

 

Julie VanDusky-Allen

About Julie VanDusky-Allen

Julie VanDusky-Allen is at Boise State University and received her PhD in Political Science from Binghamton University in 2011. Her research focuses on institutional choice and development, political parties, the legislative process, and Latin American politics.

Leave a Reply