Researchers Mikko Rönkkö and Yvette Baurne from JSBE have been awarded with Best Paper Award at the AOM annual meeting

Mikko Rönkkö's and Yvette Baurne’s research addresses one of the most misunderstood topics in quantitative research: the normality assumption in regression analysis.

Regression is one of the most used statistical analysis methods in business administration and, more broadly, in social sciences. All statistical methods make assumptions about the data, and understanding these assumptions is a prerequisite for reliable research. According to Rönkkö, however, the normality assumption is often described incorrectly in leading scientific journals and in educational materials used, for example, at the University of Jyväskylä.

"In our research, we review statistical theory and test our hypotheses using an extensive Monte Carlo simulation, in which we simulate tens of thousands of artificial datasets under a wide variety of conditions,” says Rönkkö. “In a way, we aren’t saying anything that isn’t already covered in good econometrics books; rather, we’re translating this into language that every researcher can understand and showing just how many and what kinds of misconceptions are associated with this analytical method.”

Deviations from normality is not usually a problem in regression analysis

The findings of Rönkkö's and Baurne's research are clear: deviations from normality is generally not a problem in regression analysis. Normality assumption is needed only in small samples to ensure accurate statistical inference; in larger samples, the same result follows automatically from the theory applicable to large samples without any assumptions about the distribution.

“Our simulations confirmed this: regression analysis produced unbiased estimates and correctly calibrated statistical tests in virtually all of the conditions studied, including in heavily skewed datasets,” says Rönkkö.

"Furthermore, we demonstrated that commonly recommended “correction methods,” such as winsorization or median regression, have serious problems that have not been widely recognized. Instead, researchers should focus on assumptions that actually affect the reliability of results, such as endogeneity and heteroscedasticity."

Benefits for various fields of research

Regression analysis is used everywhere: in management and leadership research, economics, medicine, and political science. Misconceptions about normality assumption will lead to poor analytical choices—that is, data is transformed unnecessarily, less effective methods are used, or, at worst, misleading results are published.

“Our research helps researchers, teachers, and students make better methodological choices and focus their attention on the right questions,” says Rönkkö.

At the forefront of applied quantitative research

According to Rönkkö, it is wonderful that this research received recognition, for many reasons. The Academy of Management is the world’s largest and most prestigious academic association in the field of management science, and its annual conference brings together thousands of researchers from around the world.

The association is divided into divisions representing different specializations, with the Research Methods Division focusing specifically on research methods and their application. Organizational Research Methods is the premier research methods journal in the organizational sciences.

"The Division Best Paper Award is presented to the paper judged to be the best among those accepted for the conference. The competition is global and covers methodological expertise across the entire field of management and leadership research. This kind of recognition is particularly meaningful because work on methodological clarification doesn’t always receive attention,” Rönkkö says happily.

The Research Methods Division has six awards: Best Student Paper, Research Methods Division Best Paper, Organizational Research Methods Best Paper, Early Career Award, Distinguished Career Award, and Advancement of Methodology Award.

"Personally, this is a very meaningful award and highlights my profile internationally. I have now received three of these six awards. In addition, my article was voted as the best quantitative paper published by Organizational Research Methods in 2025. These show that worldwide, I'm at the very top of the field in applied quantitative research,” says Rönkkö.

"This is also a great achievement for our project funded by the Academy of Finland. Furthermore, the award will be a significant boost to Yvette’s career, as she defended her dissertation just over half a year ago and is already receiving recognition at our leading conference."

Regression analysis and the normal distribution have long been of interest

Rönkkö has long been interested in research on regression analysis methods.

"One of the cornerstones of my teaching has been that when I teach a particular analytical method, I often also show examples of what happens if the method is used incorrectly or if its assumptions do not hold true for the data being analyzed. About 10 years ago, I was creating teaching materials on the use of regression and wanted to generate datasets in which regression would yield incorrect results. No matter how hard I tried, I couldn’t “break regression” by using non-normally distributed datasets. I began to wonder if, like the authors of many of the books and articles I had read, I had completely misunderstood this issue, says Rönkkö. “It turns out that I, like many other people teaching doctoral level research methods, had understood this issue incorrectly.”

According to Rönkkö, researchers examine the normal distribution and write about it as if it would be a serious problem requiring corrective measures.

"This contradiction is what motivated our research. We wanted to use both theory and simulations to determine whether this concern is justified. The answer—that it generally isn’t—was actually already known before we wrote this article, but now our project funded by the Academy of Finland made it possible to carry out this research."

"I’ve also been teaching this topic for years through videos and courses, so it’s rewarding that the scientific community is now confirming what I’ve long been trying to convey: the normal distribution is generally not the problem researchers should be devoting their energy to," Rönkkö notes.

See also:

• Award-winning Rönkkö’s and Baurne’s research
• Video: The normality assumption (Rönkkö's YouTube channel)
• Academy of Management