If you want to minimize the risk for errors, pay attention to the design phase of your research study

Author: 
Emilia Chiscop-Head, PhD, Scientific Integrity Associate, ASIST

During our Town Hall “Fundamentals of the Scientific Process: From Formulating the Research Question to Study Design and Data Collection”, researchers discussed good research practices when working with data

 

What are the common trends and sources of concern related to study design and data errors? – Dr. Steven Grambow, Director of the Duke Clinical Research Training program and Assistant Professor in the Department of Biostatistics and Bioinformatics, asked the members of the second panel of the January research town hall, attended by 408 researchers via Zoom. Grambow, who defines himself as “an advocate for radical data transparency”, shared with the audience that what he learned in the graduate school - “that statisticians were coming only after data collection”, as “data rescuers” – was contradicted shortly by the reality of research practice. He learned over the years about “the importance of team science, clearly formulating the research hypothesis, research questions, and expectations”.

Dr. David Corcoran, Director of Genomic Analysis and Bioinformatics Shared Resource, echoed the importance of robust study design through engaging the data analyst in the process. “Data landscape is changing so rapidly that the technology you use to generate data is not going to be the same. For example, two years ago a team did a big study, found some variants important for their disease status; then two years later, they did a follow up cohort. After generating all the data, they realized that the new kit they used did not even cover the region where those variants were located”, Corcoran recalls, while advising research teams to invest time in team building and proper consultation in the design phase of a research study.

Dr. Matthew Hirschey, Associate Professor in the Departments of Medicine, Division of Endocrinology, Metabolism, and Nutrition, and Pharmacology and Cancer Biology, talked about the importance of research integrity and honesty for the future of science; the courage to admit that your research was a waste of time when your data does not confirm the hypothesis; and the need to bring data driven hypotheses to be efficient, objective and more impactful in your work. Dr. Dana Pasquale, postdoctoral fellow in the Department of Sociology focused on two selection biases when working with data: censoring and truncation. Dr. Rasheed Adebayo Gbadegesin, MD, MBBS, Professor of Pediatrics and Medicine, highlighted the importance of data provenance and data sharing to minimize errors and maximize research reproducibility. Institutional resources, like the Pilot Training Program that inspired the town hall are very important to support researchers embrace good data practices consistently.  “The investment in resources, open source tools are more important than ever before. The tools and resources that allow sciences to analyze data, to find experts in order to do complicated things are much more available than ever before”, Hirschey concluded.

 

 

LINK TO EVENT RECORDING

Many thanks to our distinguished panelists:

Panel I: Designing Online Education for Researchers

Jessica Cranfill, Instructional Designer and E-Learning Developer at Duke Office of Clinical Research

Clemontina A Davenport, PhD, Duke CTSI Biostatistics Epidemiology and Research Design (BERD) Methods Core, Course Instructor, Department of Biostatistics and Bioinformatics

Anika Lucas Sylvester, MD, MTS, Nephrology Fellow at Duke University Health System

Gina-Maria Pomann, PhD, Statistical Research Scientist and the Director of the Duke Biostatistics, Epidemiology and Research Design (BERD) Methods Core

Panel II: Minimizing the Risk for Errors When Working with Data

David Corcoran, PhD, Director, Genomic Analysis and Bioinformatics Shared Resource

Rasheed Adebayo Gbadegesin, MD, MBBS, Professor of Pediatrics, Professor in Medicine, Affiliate of Duke Molecular Physiology Institute

Steven Grambow, PhD, Director, Duke Clinical Research Training program, Associate Chair of Education and Assistant Professor, Department of Biostatistics and Bioinformatics; Co-Director Duke Clinical and Translational Science Institute (CTSI) Education and Workforce Development Core

Matthew Hirschey, PhD, Associate Professor, Departments of Medicine, Division of Endocrinology, Metabolism, and Nutrition, and Pharmacology and Cancer Biology

Dana Pasquale, PhD, Postdoctoral Fellow, Department of Sociology, Trinity College of Arts and Sciences