Three essays on conflict and ethnic tensions
On Tuesday 22 October 2024 Monica Beeder will hold a trial lecture on a prescribed topic and defend her thesis for the PhD degree at NHH.
Researchers warn of underestimated uncertainty in academic findings
A groundbreaking study with contributions from more than 300 scholars reveals that researchers often underestimate the uncertainty of their findings.
`The variation in choices researchers make along the way may lead to substantial nonstandard errors, ´ notes Darya Yuferova at NHH.
Darya Yuferova is an Associate Professor at the Department of finance at NHH. She is one of more than 300 co-authors on a paper published in the prestigious Journal of Finance in June 2024. This type of research project where many teams from all over the world work together is quite new.
`And the project is the first experiment in finance that systematically studies the variability in researchers' estimates arising from the evidence-generating process´, says Darya Yuferova.
164 research teams
The paper Nonstandard Errors examines the variability in research outcomes due to differences in researchers’ analysis paths, referred to as nonstandard errors (NSEs). This variability arises when different researchers, using the same data set and testing the same hypotheses, choose divergent methodologies or analysis paths.
`Project coordinators ran an experiment with 164 research teams and 34 peer evaluators participating. Each peer evaluator reviewing about 10 papers, ´says the NHH researcher.
Published in Journal of Finance
- The paper Nonstandard Errors examines the variability in research outcomes due to differences in researchers’ analysis paths. It was published in Journal of Finance, June 2024.
- Darya Yuferova´s main research interests include market microstructure, asset pricing, and behavioral finance. Her research has been published in leading academic journals such as Management Science, Journal of Finance, and Journal of Financial Markets.
All the research teams analyzed identical dataset - spanning 17 years - from the Deutsche Börse (of EuroStoxx 50 index futures). These are one of the most actively traded financial instruments in Europe. The teams were tasked with testing the same hypotheses regarding trends in financial market characteristics for these instruments.
Directly compare
`This allowed us to directly compare results of research teams and see how much the findings varied, even though we all were using the same data and testing the same hypotheses´.
When researchers test hypotheses, they must make many decisions, such as choosing measures, dealing with outliers, and selecting statistical models, Yuferova elaborates.
Idea for app handleling cruise tourism won The Lehmkuhl-scholarship
Nikoline Valaker and Adna Alibasic's master's thesis is "innovative and aims to solve a practical problem", the committee's states.
`If researchers don't make the same decisions, their results can vary a lot increasing the uncertainty of the academic findings´.
The researchers found significant NSE, even for simple hypotheses (i.e., requiring less choices from researchers). For example, one hypothesis showed an NSE of 1.2 percent around a median estimate of -3.3 percent. A more complex and abstract hypothesis had an NSE of 6.7 percent around a median estimate of -1.1 percent. We also found that higher quality papers (rated by peer evaluators) had smaller NSEs.
It's worrying that nonstandard errors or NSEs are so large in academic findings.
Darya Yuferova
Found significant NSE
A lot of choices researchers make can really change their results—especially things like data frequency and choice of statistical models.
`It's worrying that nonstandard errors or NSEs are so large in academic findings. But the good news is that getting feedback from peer evaluators can cut these NSEs in half as we show in our paper. This finding re-emphasizes the importance of peer review-based publication process`, concludes Yuferova.