Learning to question the evidence: three years of innovation in statistics education
Three years ago, assistant professor Lucie Zicha received a Comenius grant to rethink how statistics is taught at Leiden University College. Looking back, she reflects on how that project evolved into ReplicatED and what it has taught her about helping students understand how science really works.
Learning science by doing science
The project began with a challenge. How do you teach students to think like researchers in just seven weeks? For Zicha, the answer was not to teach more statistical techniques, but to change how students learn them. 'To read a statistics table, you really should understand how science is produced,' she says.
'To read a statistics table, you really should understand how science is produced'
Instead of working with artificial datasets, students now analyse published research papers using the original data collected by researchers. Through ReplicatED, the online platform developed as part of the project, lecturers can easily select studies that match the level of their course and the statistical methods students are learning.
In class, students recreate the analysis step by step. One week they examine how researchers selected their sample. Later they build the same regression model described in the article. Every new method is immediately applied to a real research question.
When the answers do not match
Sometimes students reach exactly the same results as the published study. Sometimes they do not. That is where the learning begins.
Students work together to understand why their findings are different. Did they overlook something? Did the researchers make different choices? Could the original study have limitations? Rather than being rewarded for finding the 'correct' answer, students are assessed on how they investigate the problem and explain their reasoning.
According to Zicha, this changes the way students see science. 'Science is a human enterprise. It's wonderful, but you also need to understand its grey areas.'
Different perspectives, better questions
Because LUC brings together students from different disciplines, every group approaches the same research differently.
A social science student may focus on the theory behind the study, while someone from environmental sciences looks more closely at the statistical model. As they compare ideas, students discover that different disciplines have different research traditions and ways of interpreting evidence.
Those discussions are just as valuable as the calculations themselves. For many students, this is also the first time they experience research as something they actively participate in rather than something they simply read about.
Looking back
Three years after receiving the Comenius grant, Zicha believes the project has achieved much more than improving statistical education.
Students leave the course with the confidence to question evidence, consider different perspectives and make careful judgements based on data. Those skills remain valuable whether they continue into research, develop public policy, work in business or address other complex societal challenges.
As Zicha puts it, 'there's no such thing as the science.' Scientific knowledge develops through discussion, testing ideas and asking better questions. By bringing real research into the classroom, students learn that they are not just studying science. They are learning how to contribute to it.
On 10 November, a hands-on workshop ‘Teaching Statistical Replication' will take place in Wijnhaven for educational staff.