Establishing reliability of biomechanical data using univariate and multivariate approaches

Area of Science:

• Biomechanics
• Kinematics
• Human Movement Analysis

Background:

• Accurate calculation of the total body center of mass (CM) is crucial for understanding human locomotion.
• Generalizability theory provides a framework for assessing the reliability of measurements in complex human movement studies.

Purpose of the Study:

• To evaluate the reliability of total body center of mass (CM) values derived from cinematographical data using generalizability theory.
• To differentiate between univariate and multivariate approaches in assessing measurement error.

Main Methods:

• Twenty-eight male volunteers' walking gait was filmed, and six key stride frames were digitized by three plotters over three repetitions.
• Generalizability theory, employing univariate and multivariate ANOVAs, was used to analyze CM coordinate reliability across subjects, plotters, and repetitions.
• Measurement facets (plotters, repetitions) were treated as random effects.

Main Results:

• Minimal error was observed for digitizing repetitions, indicating high intra-plotter reliability.
• Significant interplotter error was detected for most analyzed frames, suggesting variability between different digitizers.
• Multivariate Phi-coefficients for generalizing across plotters and repetitions ranged from 0.82 to 0.90, indicating acceptable overall reliability.

Conclusions:

• Interplotter variability is a primary source of error in cinematographical CM calculations.
• Multivariate generalizability analysis provides a more comprehensive assessment of reliability than univariate methods for CM data.
• Standardizing digitizing protocols and training is essential to improve the reliability of CM measurements in biomechanical research.