Rant: Minimally Sufficient Research Can Maximine Insights

by cv harquail on March 15, 2009

I confess a nagging frustration with certain trends in my academic discipline, Organizational Theory.

While good research on organizations and organizational behavior fascinates and delights my nerdy self, boring and poorly executed research makes me lay my head on my desk in despair. Sometimes I even consider resigning from the editorial boards I was so honored to join. The variety of ways to create bad research approaches infinity.

gauss.jpeg

What bothers me most is when researchers take perfectly good research questions and topics, and then grind them to insignificant bits with their statistical methods. It’s like swatting a fly with a baseball bat.

As a doctoral student, I can remember making fun of organizational climate research and person-organization fit research that relied on fancy-schmancy statistics, but came up with no interesting and actionable results. What was true in the application of logit, probit, and structural equation modeling in the 90’s is now true of agent-based complexity models and Gaussian nets– nice stats, but what do they really tell us? That you took a class last summer at ICPSR? The point of using any particular technique is not to show that you can do the math, but to help us understand a phenomenon.

Why do we care what algorithm fits best, if it tells us nothing about how to help people engage with organizations more effectively?

So I was tickled to see an article in the latest issue of Perspectives on Psychological Science , a journal of the Association for Psychological Science. This special issue is devoted to articles about how to improve psychological research and psychological science as a field. In contains a direct and pertinent challenge to those scholars tempted by methodological overkill:

Minimally Sufficient Research
Chris Peterson, The University of Michigan.

Abstract— Psychology would be improved if researchers stopped using complicated designs, procedures, and statistical analyses for the sole reason that they are able to do so. The present article reviews some of the classic studies in psychology, all of which are breathtakingly simple. The notion of minimally sufficient research is suggested as an ideal worth following. More generally, questions should dictate research methods and statistical analyses, not vice versa.

In science, as in fashion, as in life, sometimes "less is more".

I’ll take a simple T-test and an actionable insight any day.

From the Table of Contents, you can access articles in this special issue for free.

ppsbanner.jpg

{ 1 comment }

Joseph Logan March 19, 2009 at 12:22 pm

Today’s post at PublicOrgTheory is about this post, and friends on Facebook appear to dig it.

Comments on this entry are closed.