Dr Andreas Alfons is an Associate Professor at the Econometric Institute at the Erasmus School of Economics. Alfons develops statistical methods that are robust against atypical data. He states that sharing the code for developed methods is a minimum requirement for open science. “If people have your code, they can use your method: that immediately increases the impact of your research.”
What research are you working on right now?
“I mainly focus on statistical methods that are less sensitive to deviations from assumptions. Right now, I’m working on a Vidi-project about rating scales. I try to develop methods that are less sensitive to outliers in the data and are able to detect them. Outliers meaning individuals who respond very differently to questions than the majority. You could have a statement that most people agree on, but some people do not. This could happen when people don’t read the question properly, for example. But it could also be that you have certain individuals who might think very differently than the majority. As a researcher, you would like to know about these different groups in your data because they could be the most interesting data points. If you are unaware of these groups and put all respondents in the same analysis, you might get results that are neither relevant for the majority group nor for the minority group.”
You were nominated for the ERIM Open Science Award, what was the main reason?
“I was nominated for providing open-source software for all the statistical methods that I develop. If I don’t provide the code, if empirical researchers don’t have the code for applying my method, they simply cannot use it. I believe it is very important for people who develop a methodology, to make methods available in the form of code. That is something I strongly believe in, and that was honoured with this nomination.
With my upcoming papers, I will also make the code from all simulation experiments and empirical examples available.”
When did you dive into this open science development?
“I have always made the code for my methods available, ever since I did my PhD. Also because where I did my PhD, at Vienna University of Technology, there was a strong focus on open-source software. Unfortunately, still many people who do methodological research do not share their code.”
" As a researcher you should believe in what you do, so why not be transparent?”
What is the biggest challenge in practising open science?
“Well, the fact that not everyone is joining the movement. Many methodological articles don’t share any code, not for the method itself nor for any of the experiments, so nobody can use them. I think there should be a grassroots movement to push these open science policies further so that journals take notice, and in the end, make it a requirement. The top journals should require that all the results in a paper can be reproduced easily by providing the code. And because the editors of these journals are also researchers themselves, we should create a strong open science movement among researchers.
Also, there should be more awareness of how important reproducibility is. Personally, I have a math and computer science background. In mathematics, you need to have proof for every result that you put into your paper. You need to show all the steps to verify that your result is correct. In my opinion, we should implement the same culture in empirical research, your code shows all the steps to arrive at your empirical result. Open science should be the status quo.”
What is the biggest surprise benefit of working open and transparent?
“A practical point is that working open was stated by the jury as one of the strong points of my Vidi-project. It was a substantial reason for me to receive the grant. So it can help researchers to get funding. Furthermore, it helps with the dissemination of your work. No one is going to use your method if they don’t have the code. If people can use your method, that means more citations for your paper, so that increases the impact. For my work, there are lots of benefits in really having everything open.
It can also help other researchers to get started. If they are interested in using my method, they might first try to experiment and see if the method really works. And it shows that you did everything the correct way. As a researcher you should believe in what you do, so why not be transparent?”
- Related content
