14:30 - 15:30
Contributed Paper Session
Room: Upper Foyer
Chair:
Aleksandra BUJNOWSKA
Organiser/s:
Per Nymand-Andersen
Providin access to confidential microdata for research purposes comes with several challenges. This session will cover several related issues:
- On-line remote access to invisible microdata. Development which minimise information loss and maximise utility without compromising privacy of data subjects.
- Ethical issues in access to microdata.
- Scientific research should be reproducible. How can reproducibility be achieved without giving access to the microdata?
Research Reproducibility with Confidential Data: Certifying the Uncertifiable
christophe perignon 1, christophe hurlin 2
1 HEC Paris, Cernay la ville
2 University of Orleans, Orleans

A growing fraction of research is nowadays conducted using confidential data, such as

highly‐granular government data which require specific accreditation and controlled secure

access. Analyzing such rich datasets allows researchers to conduct extremely innovative

research programs and to address research questions that they could not address by only

relying on public data; hence significantly pushing the frontiers of knowledge and having a

positive impact on Society.

However, researchers using confidential data are inexorably challenged when it turns to

research reproducibility. Indeed, how can they show that that their research is reproducible

when their peers, referees, editors, etc cannot have access to these unique data? How can

they signal the reproducibility nature of their research?

In this paper, we present a joint initiative conducted in France between the CASD (French

Secure Data Access Center) and cascad (Certification Agency for Scientific Code and Data).

They jointly propose to CASD users to attest that the numerical results in a given scientific

article (tables and figures) can be reproduced from the code and confidential data used by

the researcher. This certification consists in a rigorous evaluation process jointly conducted

by a referee specialized in the software used by the researcher and by an expert from the

particular scientific field (an Editor). At the end of the process, a certification rating is

delivered to the researcher, with RRR being the highest potential rating. The researcher can

transmit the reproducibility certificate along with the manuscript when submitting a paper

to an academic journal. Hence, such certification process enriches the « peer review »

process of research.


Reference:
Th-CPS03-01
Session:
Challenges in providing access to microdata
Presenter/s:
christophe perignon
Presentation type:
Oral presentation
Room:
Upper Foyer
Chair:
Aleksandra BUJNOWSKA
Date:
Thursday, 18 October 2018
Time:
14:30 - 14:45
Session times:
14:30 - 15:30