FY19 IRAC Work Group on Reproducibility and Replication Issues in Science:
Quantitative Analysis of Biases in Epidemiology and its Role in Risk Assessment
The purpose of this workgroup was to increase awareness of federal personnel involved in risk assessment, risk management, and risk communication of issues associated with reproducibility and replication of scientific research. One specific topic was interpretation of epidemiological effect size findings and their associated biases. The workgroup goals included:
- Enhancing the technical communication of the underlying science behind, rationale for, and implications of a more quantitative treatment of these biases.
- Providing information and background on the potential of more recently available techniques and software to better and more quantitatively evaluate biases in these estimates and their direction
The workgroup held workshops that provided a forum to educate IRAC member agencies on the types of scientific biases in epidemiology relevant to risk assessment, why this is important to consider, and types of analyses and currently available tools to evaluate the impact of these biases. The first workshop, held on September 13, 2018 included a more general introduction to the topic which would be of interest to those involved in risk assessment generally. The workshop consisted of 4 presentations: 1) Issues in Reproducing and Replication Science, 2) Effect Size Magnification for Underpowered Studies: The role of the “Winner’s Curse” in reproducibility issues, 3) Quantitative Bias Analysis (QBA) for Observational Studies, and 4) Reproducibility and Replication of Research: Benefits and Challenges. The second workshop was a continuation of the first but included more detailed information intended for those more heavily involved in the topic or with a deeper interest.
|
Work Group |