Scientific Measurement Uncertainties Across 41,000 Observations

41,000 measurements of 3,200 quantities from medicine, nuclear physics, particle physics, and interlaboratory comparisons were analyzed by David C. Bailey in 2020. The dataset studies reported scientific uncertainties, finding outliers and heavy-tailed distributions of uncertainty-normalized differences.

Use Cases

• Model the distribution of uncertainty-normalized differences between measurements to test for consistency with heavy-tailed Student's t-distributions.
• Analyze the frequency and magnitude of 5σ outlier disagreements across the 3,200 measured quantities.
• Compare the evaluation and improvement rates of measurement uncertainty between medical research and physics domains.
• Investigate the power-law distributions of contributions to measurement uncertainty from mistakes and unknown systematic errors.

Strengths

• Large scale analysis of 41,000 individual measurements.
• Covers a diverse set of 3,200 scientific quantities from multiple disciplines.
• Provides empirical evidence on outlier frequency and uncertainty distribution shapes.
• Data supports direct comparisons between uncertainty evaluation in medicine and physics.

Limitations

• The underlying raw measurement data and specific column structure are not provided, limiting direct analytical use.
• The dataset is a meta-analysis of reported uncertainties; original experimental context may be missing.
• Temporal coverage of the individual measurements is not specified, potentially limiting time-series analysis.

Provenance

Source

Dryad, authored by David C. Bailey.

Collection Method

Analysis of reported scientific uncertainties from published research and interlaboratory comparisons.

Freshness

Last updated in June 2020.