A QICSS pilot learning project: reproducibility of research conducted with confidential data

Overview
This training will provide an overview of the issues around the reproducibility of research conducted using confidential data, and demonstrate the steps to follow in order to obtain a reproducible project. The training will be followed by an opportunity to put the knowledge acquired into practice by being paired with a research team that already has an active project at QICSS, which has reached the stage of being replicable.

The training will be offered virtually on Zoom. During the initial week of training, trainees should expect to spend about 30 hours working on their own, at their own pace, and participating in the Zoom sessions with the instructor.

The subsequent practical portion will occur in a QICSS laboratory. Participating individuals must have — or obtain — a Statistics Canada Deemed Employee status. Trainees should expect to have to put in about 20 hours of individual work in the laboratory in order to successfully complete their project on reproducibility.

The training will be bilingual. The course material will be in English. The language of instruction will be English or French depending on the preferred language of the group. Trainees may interact with the instructor in either French or English, depending on their preference.

The workshop is offered free of charge.

Initial training phase
This training will be virtual, and will consist of a combination of live meetings on Zoom and individual work outside of these meetings. First, a two-day training will be held via Zoom on July 12 and 13, 2021, from 9 a.m. to 4 p.m. It will be followed by additional sessions on July 14, 15 and 16, 2021, during which trainees will apply the knowledge they have acquired using practical (fictional) exercises on how to set up, develop and "publish" a reproducible research project. During these sessions, the instructor will assist the trainees in subgroups, live on Zoom.

Requirements
Trainees must have already carried out analyses of empirical data in the social sciences using statistical software, ideally in a QICSS laboratory research environment. Familiarity with at least one of the following software, Stata, Matlab, R, SPSS or SAS, is required, as is familiarity with the Windows Desktop environment.

Subsequent training in a QICSS laboratory
The Zoom training will be followed by a practicum in a QICSS laboratory, provided the trainee can find a research team to be paired with that already has an active project underway at QICSS, and has reached the stage where it can be reproduced. Note that trainees must have, or obtain, the status of Deemed Employee of Statistics Canada and be added as co-investigator on the research team with which they will be associated. The laboratory-based practicum must be completed in summer or early fall. The instructor will offer support via Zoom or email, as availability allows.

Scholarship for students:
Trainees who have completed the initial training and who have subsequently been paired with a research team for the purposes of carrying out their reproducibility project in a QICSS laboratory will be eligible for a scholarship of $1,000. The disbursement of the scholarship is conditional on the trainee having successfully completed the reproducibility project in a lab setting. A report confirming completion will be required. (Trainees who could not be paired with a research team to conduct a reproducibility project at QICSS will be offered another practical experience option to complete over the summer. However, this option is not eligible for the scholarship.)

Call for expressions of interest (May-June 2021)
Are you a Master’s or PhD candidate, or a postdoctoral fellow, registered for full-time study for the summer 2021 semester, and interested in benefiting from a two-day training session on the reproducibility of research, with a $1,000 bursary attached?

Do you have a current project at QICSS that has reached the stage of being replicable, and would you be interested in being paired with a student or fellow who has had QICSS training on reproducibility ?

In either cases, QICSS invites you to let [email protected] know that you are interested by no later than June 15.

Trainer

Dr. Lars Vilhuber holds a Ph.D. in Economics from Université de Montréal, Canada. He is the Data Editor of the American Economic Association, Managing Editor of the Journal of Privacy and Confidentiality, and on advisory boards of restricted-access data centers in Europe Canada, and the United States.