Description

In a lot of (high-stakes) situations the four-eyes principle is used: from flying a plane to grading a thesis, we tend involve two individuals to avoid making critical mistakes (among other things). However, this often does not seem to apply to data-processing and -analysis in the context of research. Perhaps unsurprisingly, studies have shown that a substantial percentage of published findings are not analytically reproducible: if someone else tries to perform the same analyses on the same data, they frequently obtain different results. In this course, we will cover a number of tools and approaches to avoid this issue. Most notably, we will introduce the co-piloting model, which involves having a collaborator independently verify a study’s results by redoing the analyses.

Schedule 
Day 1:

• What does reproducibility entail, why is it important, what is the state of different fields?
• Tools to verify reproducibility from relatively basic applications such as statcheck, to more comprehensive approaches like the  compreprev application.

Day 2:

• Tools and approaches to improve reproducibility (data management, working with syntax/code, version control, dynamic reports, and containers)
• Introduction to co-piloting and how to deal with reproducibility issues

Day 3: 

• Co-piloting each other’s work

Mode of instruction

Preparation
Fill in a short survey about the type of research you (plan to) conduct, the type of analyses you (plan to) perform use, the software you (plan to) use. Also, prepare to bring your own dataset/analyses to which we can apply (some of) the tools discussed during Day 1-2, and which can serve as the basis for a co-piloting exercise on Day 3.

Mode of instruction 
Mix of collective explanation, hands-on exercise prepared by the instructors, and application of the tools and approaches to your own dataset/analyses. In case you don’t have a suitable dataset (yet), we can arrange an alternative (we will ask this in the preparatory survey; see above). On Day 3 we will form duos or triplets based on your answers to the preparatory survey (looking to match on type of research, software etc.), and the goal is to practice co-piloting each other’s work.   

Reading list

Not applicable. We will cover some resources during the course, which will be shared with the participants.

Fees

Target group	One day	Two days	Three days
PhD candidates FSW	FREE	FREE	FREE
Staff FSW	€330	€440	€550
Other Leiden University PhD candidates	€235	€345	€455
Externals	€490	€655	€820

Entry requirements

The course will mostly focus on quantitative research, so some familiarity with statistical analyses of quantitative data is helpful.