It's time to move on from the bell curve

Area of Science:

• Biostatistics
• Clinical Electrophysiology

Background:

• The bell curve, historically used for binomial events, was applied to biological and social measures.
• Electrophysiologic studies have long used the bell curve for reference value derivation.
• Empirical evidence suggests electrophysiologic measures do not follow a bell-curve distribution.

Purpose of the Study:

• To challenge the use of the bell curve for electrophysiologic reference values.
• To advocate for the adoption of non-parametric statistical methods.
• To align with recommendations from the American Association of Neuromuscular & Electrodiagnostic Medicine.

Main Methods:

• Critique of the application of the bell curve (mean ± 2 standard deviations) to electrophysiologic data.
• Proposal for using non-parametric analyses, specifically percentile values.
• Review of statistical principles and existing task force recommendations.

Main Results:

• Electrophysiologic measures do not conform to a bell-curve distribution.
• The use of mean ± 2 standard deviations for reference values is statistically unsound for these measures.
• Non-parametric methods provide a more accurate approach to establishing reference values.

Conclusions:

• The practice of using bell curve-derived reference values in electrophysiology should be abandoned.
• Non-parametric statistical analyses, such as percentile values, are recommended for deriving reference values.
• This approach aligns with current expert recommendations and sound statistical principles.