Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 187 - Theory and Methods for Building Successful Data Analyses
Type: Topic Contributed
Date/Time: Monday, August 8, 2022 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistics and Data Science Education
Abstract #323651
Title: Reproducibility: You Can Do Data Analysis Without it, but Should You?
Author(s): Tiffany A Timbers*
Companies: University of British Columbia
Keywords: Data Science; Education; Reproducibility; Analysis

Reproducibility, as defined by the National Academy of Sciences, is reaching the same result given the same input, computational methods, and conditions. In the context of data analysis, commonly the input is the data, the computational methods are the code needed to do the data analysis, and the conditions are the programming languages, packages and operating system dependencies (as well as their specific versions) needed to execute the code. Data analysis can be, and often is, performed without recording and sharing one or more of these components - resulting in an analysis that cannot be reproduced. Does this matter? What are, if any, the consequences of such a non-reproducible data analysis? And is omitting one of the components needed for a reproducible data analysis "worse" than another? This talk will try to answer these questions through discussions of real-life examples of non-reproducible data analyses and their consequences in an attempt to illustrate the critical importance of reproducibility in data analysis.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2022 program