Trust me, I'm a Statistician

The theme for the 2024 World Economic Forum is “Rebuilding Trust”, focusing on  “the fundamental principles driving trust, including transparency, consistency and accountability”. But (how) can professional statistics, (how) can we as professional statisticians contribute to rebuilding trust in a world of misinformation, disinformation, and statistical illiteracy?

“The judgment how trustworthy people are requires us to look at three things. Are they competent? Are they honest? Are they reliable? And if we find that a person is competent in the relevant matters, and reliable and honest, we'll have a pretty good reason to trust them, because they'll be trustworthy.”

Onora O'Neill, British philosopher and politician

“Ethical guidelines for statistics" are not just for people who hold the job title "statistician" or "data scientist". The International Statistical Institute (ISI) Declaration on Professional Ethics , the  American Statistical Association (ASA) Ethical Guidelines are explicit about it, and the Association of Computing Machinery (ACM) Code of Ethics is as well for computing.

Anyone who uses the tools, techniques, and technologies of statistics is obliged to do so ethically - even if they're a medical doctor.

When treating patients, a medical doctor owes the duty of care to the patient; when collecting, analyzing, interpreting, or drawing conclusions from data, the medical doctor owes the duty of care to the practice of statistics. Just as the duty of care for patients is guided by the Hippocratic oath, the duty of care for data should be guided by ethical principles for statistical practice.

A "Hippocratic oath" for data

The entire scientific community relies on that specific duty of care (about arguments with, or from, data). As data becomes more ubiquitous, handling, collecting, managing, analyzing, and making arguments with or from data also becomes more ubiquitous. Statistical ethics are becoming even more important as the increased availability of large data sets (big data), including linked data sets, and computationally intensive method (e.g., machine learning) raises more issues associated with privacy and social license.

Therefore, we do need to raise awareness of, and commitment to, ethical statistical practices - not only for statisticians.

Accreditation: The action or process of officially recognizing someone as having a particular status or being qualified to perform a particular activity.

Professional qualifications like CStat (Chartered Statistician, RSS), PStat (Accredited Professional Statistician, ASA) and AEUStat (Accredited  European Statistician) “testify that there is a body of knowledge known as statistics acquired through formal education, work experience, and ongoing professional development activities. Accreditation provides a measure of assurance to employers, contractors, and collaborators of statisticians and a mark of accomplishment to society at large.”

Compliance with ethical standards

They also testify that the Accredited Statistician knows and adheres to ethical principles, such as:

• ISI Declaration on Professional Ethics,

• ASA Ethical Guidelines for Statistical Practice,

• Toronto Declaration: Protecting the Rights to Equality and Non-Discrimination in Machine Learning Systems,

• ACM Code of Ethics and Professional Practice,

• UN Fundamental Principles of Official Statistics,

• OECD Recommendation on Good Statistical Practice,

• European Statistics Code of Practice,

• U.S. Statistical Practice Directive #1,

• CNSTAT Principles and Practices for Federal Statistical Agencies and Recognized Statistical Units,

• U.S. Federal Data Strategy: Data Ethics Tenets,

• UK Government: Data Ethics Framework,

• Other local and regional ethical guidelines, including those for specific areas, for example the ESOMAR Code of Conduct and the IEEE Code of Ethics.

Trust in statistics requires trustworthiness of those who produce, provide, and deal with data and statistics. Trustworthiness requires knowledge, dissemination and application of ethical values and principles and standards in the increasingly diverse landscape of data communities and the developing world.

“More trust is not an intelligent aim. Intelligently placed and intelligently refused trust is the proper aim. Trustworthiness is what we have to judge.”

Accreditation gives evidence of trustworthiness

The Federation of European Statistical Associations (FENStatS) launched a standard for accreditation of statisticians. Each European national association participating has decided to make this available to its members. FENStatS is an independent and non-profit European scientific association organised as a Federation of 27 European National Statistical Societies (SNEE) and institutions related to or interested in statistical sciences, such as the European Central Bank.

The accreditation is voluntary and is intended as a measure to enhance the quality and importance of statistics in a world in need of facts and high-level statistical literacy. It is also intended as an encouragement for statisticians to develop their roles, stand by ethical standards and promote the use of statistical data.

“So the moral of all this is, we need to think much less about trust, much more about being trustworthy, and how you give people adequate, useful and simple evidence that you're trustworthy.”

The accreditation is a complementary step in your career as a professional statistician. AEUStat is the European version of the qualifications:

• Accredited Professional Statistician (PStat) of the American Statistical Association

• Accredited Statistician (AStat) of the Australian Statistical Association.

AEUStat has mutual recognition with both titles. Accredited statisticians have demonstrated that they have

1. at least an MSc in statistics or equivalent,

2. at least five years of work experience,

3. invested in professional development during this time,

4. proven communication skills,

5. comply with ethical standards, and

6. are member of a FENStatS member organization.

Are you interested in rebuilding trust in statistics?

See https://www.fenstats.eu/accreditation for details or contact the president of the Accreditation Committee, Magnus Pettersson ([email protected]).

This article is based on contributions from Walter J. Radermacher, Dennis Trewin, and Rochelle E. Tractenberg.