The study of studies

Until recently, Brian Wansink was a rock star in the field of nutrition science.

Wansink’s research about shopping and eating habits drove numerous industry marketing strategies, and his work was cited more than 20,000 times, according to the online-media outlet Vox.

But some of his findings were called into question recently, and a total of 13 of his studies have been retracted. Though he stands by his findings, Vox reports Wansink is retiring June 30 from Cornell University, where he has been director of the school’s Food and Brand Lab.

The development didn’t surprise University of Idaho associate professor of marketing Berna Devezer.

Devezer is part of a multi-disciplinary team at the school looking for ways to improve traditional research methods by examining an emerging pattern that shows results from many scientific studies can’t be reproduced.

The work, still in its early stages, has the potential to change how studies are conducted for everything from pharmaceuticals to self-help books.

“When you hear about these ideas that are too easy, very likely they may not be true or they may only be true under very narrow circumstances and for a few people,” Devezer said. “So take those things with a grain of salt. Reality tends to be much more complicated. Solutions are usually not that easy.”

Her present research is the latest project in a career that began in her native country of Turkey. She helped usher companies into the internet age before moving to Pullman to earn graduate degrees at Washington State University.

Business Profile talked with Devezer about her examination of scientific methods, how companies can polish their images after bad publicity and her choice to make the Palouse her home.

Business Profile: What can you share about your current research?

Berna Devezer: Besides me, we have a philosopher, a statistician (my husband Erkan Buzbas) and a computer scientist. We want to understand basically what is going on. Why can’t we reproduce results? Under what conditions should we expect results to be reproducible? What leads to these problems? We also want to understand how does science progress? How do we accumulate knowledge if things are nonreproducible this often?

There are a lot of reasons this has occurred. A lot of this was attributed to research misconduct. Part of it was attributed to misuse of statistical methods. Statistical and analytical training was not sufficient, so scientists were implementing things the wrong way.

It’s challenging because science is not that linear. It progresses in a very chaotic manner. You cannot really trace where knowledge came from. It goes through so many different processes.

It’s really very exciting. But it’s also something a lot of scientists wouldn’t want to hear about. (Our early findings are) that even if you do things right, you may not end up finding reproducible results. Or even if your results are reproducible, it doesn’t mean you have made a discovery.

BP: You are clear that you have a lot more work to do and want to explore many additional questions. But is there anything your initial look at the problem indicates would help?

BD: One of the quality indicators that should be taken into account is the transparency of reporting and the availability of raw data.

A lot of industry and academic researchers are promoting open science practices. It’s not only open data. We find open data is not even as important as open materials, procedures and processes. Reporting everything that you did before you designed the study, everything you did after you got to your data and started analyzing it. A lot of the problems come from people running hundreds of analyses and reporting only one of them. That is is a lack of transparency. That result may be one of the positive results among 100 negative results.

BP: How does the fast pace that information travels in the 24-hour-a-day news cycle and on social media make this issue worse?

BD: Businesses actually consume scientific outputs. A lot of times studies get misrepresented ... on social media and media in general. If it gets publicized too quickly, it may have serious consequences for businesses and for the customers as well. They may end up overhyping some results that have not been corroborated by external evidence, and they may base new products or new services or the idea that they sell on wrong conclusions. Companies should look at the data, (not just press releases and news accounts).

Science does not necessarily progress by finding a lot of counterintuitive, interesting stuff. It’s usually very incremental.

But these kinds of (novel) results are the things that industry typically finds interesting. There’s going to be a biased selection there toward taking wrong results seriously. The boring results can be true, but they don’t necessarily interest a lot of people.

BP: Some of your earlier research looked at successful ways for companies to rebound after bad publicity. What did you learn?

BD: You might have seen there are videos on YouTube. (A man believes) United Airlines broke his guitar, and he made a series of videos about them. (United offered $3,000 in compensation for the guitar, according to news accounts.) That type of retaliation might happen, and it might actually have a huge affect on businesses. ... We study what can brands do to regain that confidence and to recreate loyalty with customers who feel like they have been failed by the company.

The company definitely needs to make an apology. But an apology only works when it’s perceived as authentic. A boilerplate response that doesn’t address the actual damage that has been done by the company, that doesn’t address anything personal about the customer, that doesn’t try to understand or center the customer in the middle of this experience — that doesn’t work. The customer needs to feel it has a potential of leading to behavioral change from the perspective of the company. That part is compensation. It doesn’t have to be monetary. But something that will offset the unfairness that they have experienced. A combination of the two actually seems to work the best.

BP: You had a successful career in Turkey before arriving in the United States. What made you decide to move?

BD: My then-boyfriend (Buzbas) came to the states for his graduate studies. We got married after he started his program, and then I decided to move here. I always wanted to go back to school. I didn’t feel like I was fulfilling my potential in industry. It felt like I was working for people rather than pursuing my own ideas. You have tasks to fulfill. You do those, but there wasn’t much room for creativity or my own original work. I like thinking a lot about theoretical problems. That’s why I thought I need to change careers and find a way to make it my job to read about things, to think about things and to pursue my own ideas.

I feel like the United States provides a lot of opportunities for me. The academia in Turkey right now, especially because of the current government, is suffering. I wouldn’t want to be a professor there. A lot of my colleagues back in Turkey have had political issues. The environment there is a little more risky.

Williams may be contacted at ewilliam@lmtribune.com or (208) 848-2261.