Trends in the Distribution of P Values in Epidemiology Journals: A Statistical, P-Curve, and Simulation Study
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.Epidemiologists are shifting from P values to confidence intervals. Analysis shows P values have decreased, but this is likely due to increased statistical power, not reduced P hacking.
Area Of Science
- Epidemiology
- Biostatistics
Background
- Concerns exist regarding null-hypothesis statistical-significance testing, P hacking, and reproducibility in epidemiology.
- Epidemiologists have recommended reporting confidence intervals and reducing reliance on P values.
Purpose Of The Study
- To investigate whether efforts to de-emphasize P values have altered their distribution in epidemiological research.
- To analyze trends in P value distributions over time in major epidemiology journals.
Main Methods
- Scraped P values (N=25,288) from 21,332 abstracts published between 2000-2024 in four major epidemiology journals using ChatGPT's 4o model.
- Calculated P values from estimates and confidence intervals.
- Evaluated trends over time and fitted to expected P value distributions, simulating scenarios with and without changes in statistical power.
Main Results
- Average P values decreased from 2000 to 2024.
- The proportion of P values falling just below the 0.05 significance threshold also decreased.
- Model fits suggest an increase in statistical power over the study period.
Conclusions
- The observed trends in P value distribution are consistent with increased statistical power, rather than a reduction in P hacking.
- While the frequency of P values near the 0.05 threshold has modestly declined, this is attributed to enhanced statistical power.
Contact us if these videos are not relevant.
Contact us if these videos are not relevant.