A simple method for correcting for the Will Rogers phenomenon with biometrical applications
Mark Stander1, Julian Stander2
1AECOM, London, UK.
Biometrical Journal. Biometrische Zeitschrift
|January 21, 2020
Summary
The Will Rogers phenomenon can skew study results by creating false impressions of improvement. This paper introduces a new method to correct for this statistical bias in epidemiological and clinical studies.
Area of Science:
- Biostatistics
- Epidemiology
- Clinical Research
Background:
- The Will Rogers phenomenon describes a counterintuitive statistical effect where average values in two groups increase after an element moves between them, suggesting improvement without actual change.
- This phenomenon can compromise the reliability of epidemiological and clinical studies by creating misleading conclusions.
- Previous studies have been affected by this statistical bias, highlighting the need for correction methods.
Purpose of the Study:
- To describe the impact of the Will Rogers phenomenon on epidemiological and clinical studies.
- To present a novel and simple method for correcting the Will Rogers phenomenon in statistical analyses.
- To demonstrate the efficacy of the proposed correction method using real-world biometrical examples.
Main Methods:
- The proposed method involves incorporating a transition matrix to model the movement of elements between two sets.
- Probability-weighted expectations are utilized to adjust for the statistical bias.
- The method's effectiveness is illustrated with two biometrical examples: migration economics and breast cancer epidemiology, including estimation uncertainty in the latter.
Main Results:
- The Will Rogers phenomenon can lead to erroneous conclusions regarding improvements in study outcomes.
- The presented correction method effectively adjusts for this bias, demonstrating improved accuracy compared to naive analyses.
- The application in breast cancer epidemiology, accounting for estimation uncertainty, further validates the method's robustness.
Conclusions:
- The Will Rogers phenomenon poses a significant threat to the validity of findings in epidemiological and clinical research.
- The developed statistical method offers a reliable approach to correct for this bias.
- This correction is crucial for accurate interpretation of results in biometrical studies, particularly in fields like cancer epidemiology.
