Promoting Statistical Practice and Collaboration in Developing Countries

O. Olawale Awe, Kim Love, Eric A. Vance, O. Olawale Awe, Kim Love, Eric A. Vance

• Publisher: Chapman and Hall/CRC
• Year: 2022
• eBook ISBN: 9781000594416
• Edition: 1
• Topic: Mathematics
• Subtopic: Probability & Statistics
• Index: Mathematics

"Rarely, but just often enough to rebuild hope, something happens to confound my pessimism about the recent unprecedented happenings in the world. This book is the most recent instance, and I think that all its readers will join me in rejoicing at the good it seeks to do. It is an example of the kind of international comity and collaboration that we could and should undertake to solve various societal problems.

This book is a beautiful example of the power of the possible. [It] provides a blueprint for how the LISA 2020 model can be replicated in other fields. Civil engineers, or accountants, or nurses, or any other profession could follow this outline to share expertise and build capacity and promote progress in other countries. It also contains some tutorials for statistical literacy across several fields. The details would change, of course, but ideas are durable, and the generalizations seem pretty straightforward. This book shows every other profession where and how to stand in order to move the world. I urge every researcher to get a copy!"

— David Banks from the Foreword

Promoting Statistical Practice and Collaboration in Developing Countries provides new insights into the current issues and opportunities in international statistics education, statistical consulting, and collaboration, particularly in developing countries around the world. The book addresses the topics discussed in individual chapters from the perspectives of the historical context, the present state, and future directions of statistical training and practice, so that readers may fully understand the challenges and opportunities in the field of statistics and data science, especially in developing countries.

Features

• Reference point on statistical practice in developing countries for researchers, scholars, students, and practitioners

• Comprehensive source of state-of-the-art knowledge on creating statistical collaboration laboratories within the field of data science and statistics

• Collection of innovative statistical teaching and learning techniques in developing countries

Each chapter consists of independent case study contributions on a particular theme that are developed with a common structure and format. The common goal across the chapters is to enhance the exchange of diverse educational and action-oriented information among our intended audiences, which include practitioners, researchers, students, and statistics educators in developing countries.

Part 1 Statistics Collaboration and Practice in Developing Countries Experiences, Challenges, and Opportunities

1 Statistics and Data Science Collaboration Laboratories: Engines for Development

Eric A. Vance

University of Colorado Boulder

Tonya R. Pruitt

Virginia Tech

DOI: 10.1201/9781003261148-2

CONTENTS

• 1.1 Introduction: Why Statistics and Data Science Have Extraordinary Potential for Data-Driven Development
• 1.2 Collaborative Statisticians and Data Scientists
• 1.3 Statistics and Data Science Collaboration Laboratories (“Stat Labs”)
  • 1.3.1 The Purpose of Stat Labs
  • 1.3.2 What Stat Labs Do
    • 1.3.2.1 Supporting Domain Experts
    • 1.3.2.2 Creating New Knowledge
    • 1.3.2.3 Transforming Evidence into Action (TEA)
    • 1.3.2.4 Training the Next Generation of Collaborative Statisticians and Data Scientists
  • 1.3.3 Stat Labs Produce Collaborative Statisticians and Data Scientists and Data-Capable Development Actors
• 1.4 Exemplar Stat Labs
  • 1.4.1 The Laboratory for Interdisciplinary Statistical Analysis (LISA) at Virginia Tech
  • 1.4.2 LISA at the University of Colorado Boulder
  • 1.4.3 The University of Ibadan LISA (UI-LISA)
• 1.5 Stat Labs Can Become Engines for Development
  • 1.5.1 Theory of Change
  • 1.5.2 Benefits and Impacts of a Stat Lab
• 1.6 Seven Steps for Creating a Stat Lab
• 1.7 What Makes a Stat Lab Strong and Sustainable
• 1.8 Conclusion
• References

1.1 Introduction: Why Statistics and Data Science Have Extraordinary Potential for Data-Driven Development

Statistical analysis and data science have enormous potential for enabling and accelerating data-driven development. The discipline of statistics has been around for more than 100 years, and its impacts are widespread and profound. Statistics has been driving development since even before the world’s first university statistics department was founded in 1911 at University College London (Hotelling 1988). For example, during the great cholera outbreak in London in 1854, John Snow collected and analyzed data to identify and close off a contaminated water pump (Tulchinsky 2018). In the same year, Florence Nightingale, the founder of modern nursing and a very early member of the International Statistical Institute and the American Statistical Association, used statistics and data visualization to show that simple hygiene measures could drastically reduce infection and death in hospitals (McDonald 1998). As Gunderman and Vance (2021) wrote, the ability to collect and learn from large amounts of data has been a major driver of innovation over recent decades. Everything from health care—including patient analytics (Adebanji et al. 2015; Awe et al. 2021; Kosorok and Laber 2019), wearable devices (Michie et al. 2017), and the COVID-19 response (Aidoo et al. 2021; Gayawan et al. 2020)—to energy (Malakar et al. 2021) to entertainment recommender systems (e.g., Netflix (Bell et al. 2007)), is now driven by data and statistics.

We consider data science to be the science of learning from data (Donoho 2017), which encompasses—at a minimum—statistics, computation, and ways of thinking about data. Clearly, the idea of learning from data is not new. The scientific method is based upon deriving and testing theories from empirical observation, i.e., data. Figure 1.1 shows how the wheel of statistics and data science turns through the scientific process to ultimately apply statistics and data science to solve problems and make decisions for the benefit of society. Starting with an understanding of domain experts’ research, business, or policy questions, statisticians and data scientists collaborate with domain experts to design an experiment or study to produce high-quality data, or, alternatively, design an algorithm to collect or scrape already existing data.

FIGURE 1.1 Statistics and data science enables and accelerates all aspects of data-driven research, business, or policy.

Statisticians can furthermore collaborate with data producers to collect high-quality data (Vance and Love 2021). All statisticians and data scientists have the responsibility to understand—and reduce when possible—the sources of variation and potential biases within the data. Statisticians and data scientists visualize the data and analyze data sets with models and algorithms to produce findings relevant to the original research, business, or policy questions. Statistics can help with interpreting the practical and statistical significance of results leading to conclusions about the data and a greater understanding of the underlying questions. This, in turn, may lead to collecting and analyzing more data in another turn of the wheel, in time progressing to solving the problem, making recommendations, and making decisions. The final step for a statistician or data scientist who wants to make a deep contribution is to collaborate with others to implement data-driven solutions, recommendations, and decisions to take action for the benefit of society.

Later in this chapter, Love et al. (2022) ask and answer why we need to develop statistical practice and collaboration in developing countries. In short, local capacity in statistics and data science can enable and accelerate innovative, sustainable solutions to local development challenges. When local statisticians and data scientists collaborate with local domain experts to transform evidence into action—especially when it occurs within a statistics and data science collaboration laboratory—a virtuous cycle of development emerges in which initial collaborations build the capacity of all parties to achieve even more positive development outcomes by applying statistics and data science.

The goal of this chapter is to explain the concept of the statistics and data science collaboration laboratory (“stat lab”) and how it can sustainably harness the extraordinary potential of statistics and data science for development. In Section 1.2, we introduce the generator of a stat lab—the collaborative statistician or data scientist. In Section 1.3, we explain what a stat lab is and what it does. We provide three examples of stat labs in Section 1.4. In Section 1.5, we discuss how stat labs can become engines for development and attempt to answer the question: Why should a university create and sustain a stat lab? We walk through the seven steps to create a stat lab in Section 1.6. Based on the lessons learned so far from the LISA 2020 Network, in Section 1.7, we characterize what makes stat labs strong and sustainable. We conclude in Section 1.8.

1.2 Collaborative Statisticians and Data Scientists

A technically trained statistician, who understands the theory and methods of statistics, is well positioned to do many things in academia, business and industry, or the policy domain. One option for the academic statistician is to teach the theory and methods of statistics to others and/or continue to develop new theory and methods. Another path available to all statisticians is to apply the theory and methods of statistics to solve problems that will help benefit society. For a statistician in business, it may be straightforward to identify problems that directly affect companies’ products or profits (Oliveira et al. 2016). In academia, on the other hand, it can be a challenge to identify or be exposed to problems that have real-world relevance. In addition, universities, especially in developing countries, may fail to prepare statisticians with appropriate applied statistics skills, including the ability to compute and code effectively (Awe 2012; Love et al. 2022).

Drawing upon a definition proposed by Halvorsen et al. (2020), we define a collaborative statistician or data scientist to be

someone possessing deep technical skills in the theory and methods of statistics or data science and effective collaboration skills who can move between theory and practice to work with domain experts to create solutions to research, business, and policy challenges and achieve research, business, and policy goals.

In other words, a collaborative statistician or data scientist can transform evidence (data) into action, i.e., the solving of research problems or the making of data-driven business and policy decisions.

Fortunately, we have found that those with a strong background in theory and methods of statistics can learn applied and collaborative skills while working on consulting projects and more involved collaborative projects within a statistical collaboration laboratory (Vance et al. 2016). Through mentorship from an experienced individual or via a formal program, a statistician can become a collaborative statistician or data scientist; collaborate with domain experts on problems in research, business, or policy (see Figure 1.2); and subsequently have a more positive impact on society (Love et al. 2017; Vance et al. 2017a, 2017b). An individual can also learn collaboration skills through self-study of Vance and Smith’s ASCCR (Attitude–Structure–Content–Communication–Relationship) Framework for Collaboration (2019).

FIGURE 1.2 A well-trained collaborative statistician or data scientist can enable and accelerate many research, business, and policy projects.

Working alone, a collaborative statistician or data scientist can enable and accelerate as many as ten projects or more per year. For example, collaborative statisticians can help understand the policy implications of uneven distribution of costly medical interventions (Vance et al. 2013), impacts of climate change on disease prevalence in Botswana (Alexander et al. 2013), the cost benefits of upgrading domestic water systems in Senegal (Hall et al. 2015a), and the productive uses of piped water in Africa (Hall et al. 2014b).

Yet another path awaits a collaborative statistician or data scientist, one with orders of magnitude more potential impact. A collaborative statistician or data scientist can create a statistics and data science collaboration laboratory.

1.3 Statistics and Data Science Collaboration Laboratories (“Stat Labs”)

A statistics and data science collaboration laboratory, or “stat lab,” is a collection of statisticians and data scientists who have been trained to collaborate with domain experts and then do so to enable and accelerate research and make data-driven business and policy decisions. Similar to a statistical consulting center or a university “Center of Excellence,” a stat lab is where members of the community can go to access statistics and data science expertise. Stat labs are not “rooms full of computers.” Instead, they are more like rooms full of collaborative statisticians and data scientists working with domain experts from multiple disciplines. The term stat lab refers to both the human resources and the physical space they occupy. The statisticians and data scientists who work in stat labs all have different relative strengths—like the comic book and movie superheroes “The Avengers”—and likewise work together to achieve a common purpose: to enable and accelerate data-driven development while training the next generation of collaborative statisticians and data scientists.

1.3.1 The Purpose of Stat Labs

Stat labs can have different missions based on the host institution’s statistics and data science needs. Vance and Laga (2017) found in a survey of North American stat labs that their primary mission was (giv...

Table of contents

• Cover
• Half Title Page
• Title Page
• Copyright Page
• Table of Contents
• Foreword
• Preface
• Editors
• Reviewers
• Contributors
• Part 1 Statistics Collaboration and Practice in Developing Countries: Experiences, Challenges, and Opportunities
• Part 2 Building Capacity in Statistical Consulting and Collaboration Techniques through the Creation of Stat Labs
• Part 3 Statistics Education and Women’s Empowerment
• Part 4 Statistical Literacy and Methods across Disciplines
• Part 5 New Approaches to Statistical Learning in Developing Countries
• Part 6 Importance of Statistics in Urban Planning and Development
• Part 7 Statistical Literacy in the Wider Society
• Bonus Chapter: Systematically Improving Your Collaboration Practice in the 21st Century
• Index