What is it about?
The mechanical, ritualistic application of statistics is contributing to a crisis in science. Education, software and peer review have encouraged poor practice – and it is time for statisticians to fight back.
Featured Image
Why is it important?
The results of practically every scientific activity are expressed in the language and with the methods of statistics. These methods are difficult to learn and to apply. Additionally it is all to easy to be caried away by one's own bias, while believing to be following an objective procedure. For these reasons we have to rethink the way we learn, teach and practice statistics.
Perspectives
While different fields of science grapple with the present reproducibility crisis the blame if often put at the door of statistics and of its methods. In reality statistics is a vantage point from where to start a process of reformation, which needs to incorporate practical as well as ethical ingredients. How can a statistician help?
Professor Andrea Saltelli
University Pompeo Fabra, Barcelona School of Management
Read the Original
This page is a summary of: Cargo-cult statistics and scientific crisis, Significance, July 2018, Wiley,
DOI: 10.1111/j.1740-9713.2018.01174.x.
You can read the full text:
Resources
Cargo-cult statistics and scientific crisis
Online open access version of the paper.
Fixing statistics is more than a technical issue
Correspondence on Nature from Andrea Saltelli and Philip Stark.
Integrity must underpin quality of statistics
Correspondence on Nature from Jeroeme R. Ravetz
Young Statistician, You shall live adventurous times
Related article in Significance, by Andrea Saltelli, December 2016, Volume 13, Issue 6, (pages 38–41), draft copy
Before reproducibility must come preproducibility
Instead of arguing about whether results hold up, let’s push to provide enough information for others to repeat the experiments, says Philip Stark. Stark, P., 2018, Before reproducibility must come preproducibility, Nature.
Should statistics rescue mathematical modelling?
This is a discussion paper published on ArXiv and submitted to a statistical jounal. The main thesis is that mathematical modelling suffers from pathologies possibly worse than those afflicting statistical analysis. The highlights of the paper are as follows: The perceived misuse and possible abuse of statistical methods has been mentioned as a concause of the reproducibility crisis. The situation has analogies with mathematical modelling, where practitioners flag the absence of agreed quality assurance standards and poor practices. Technical, cultural and ethical dimensions are simultaneously at play in both statistics and mathematical modelling. Since mathematical modelling is not a discipline like statistics, its shortcomings risk remaining untreated longer. The tools of statistics and its disciplinary organisation might help by mainstreaming tools for quality assurance of mathematical models. As an example, techniques for uncertainty quantification, sensitivity analysis and sensitivity auditing could become part of statistical syllabuses and practices.
Excerpts on modelling from “Crisis? Surely you must be joking”, talk given to CERN Geneve June 2018
Slides on the subject of the crisis in mathematical modelling and how statistics could help.
Contributors
The following have contributed to this page
