Reliability Statistics in Strength and Conditioning

Reliability Statistics in Strength and Conditioning

Introduction

• Reliability statistics are crucial for strength and conditioning coaches to assess the consistency and accuracy of measurements.
• These statistics help determine if a real change in performance has occurred.

Types of Reliability Statistics

• Intraclass Correlation Coefficient (ICC)
• Coefficient of Variation (CV)
• Standard Error of the Measurement (SEM)
• Typical Error (TE)

Intraclass Correlation Coefficient (ICC)

• Assesses the relative consistency of a measure.
• Indicates how consistent participants are relative to the sample.
• Commonly required by journals for research publication.
• Multiple types exist (6-10), but type 3,1 is generally used in strength and conditioning for test-retest within the same group.
• Interpretation:
  • It's a correlation coefficient, so it estimates relative consistency.
  • Interpret based on the 95% confidence interval to assess where the true value may lie.
  • Use a scale proposed by Koo and Lee (2016) based on the lower bound of the 95% confidence interval.
  • Qualitative scale:
    • $< 0.5$: Poor relative consistency/reliability.
    • $> 0.9$: Excellent relative consistency/reliability.
• Limitation: Doesn't provide an intuitive understanding of what relative reliability means.

Standard Error of the Measurement (SEM)

• Calculates the absolute consistency of a test.
• Used in conjunction with ICC to determine test reliability.
• Related to ICC, calculated as: $SEM = \text{Standard Deviation} \times \sqrt{1 - ICC}$
• Commonly used to calculate other statistics like the coefficient of variation.

Coefficient of Variation (CV)

• Most common reliability statistic in strength and conditioning.
• Represents the absolute consistency of the measurement.
• Calculated as a percentage: $CV = (\frac{\text{Standard Deviation}}{\text{Mean}}) \times 100$
• Interpreted alongside 95% confidence intervals.
• Thresholds exist to determine acceptable levels of measurement error (e.g., < 10% is considered reliable).
• Example: Force-time curve characteristics in isometric mid-thigh pull and isometric squat.
• Importance of considering confidence intervals: Don't rely solely on point estimates.

Within-Individual Reliability

• CV can assess the reliability of an athlete within a test.
• Indicates the error an athlete displays relative to their actual performance.
• Helps determine if a change in performance is meaningful, regardless of whether monitoring fatigue or performance enhancement.
• Meaningful change: A change in performance exceeding the level of within-individual reliability.
• Calculation:
  • Repeated trials for the athlete.
  • Calculate the average (true value) and standard deviation.
  • $CV = (\frac{\text{Standard Deviation}}{\text{Mean}}) \times 100$
• Example: Athlete A performs countermovement jumps (47 cm, 48 cm, 47.8 cm, 47.6 cm, 45.6 cm).
  • Average = 47 cm.
  • Standard Deviation = 1.2 cm.
  • $CV = (\frac{1.2}{47}) \times 100 = 2.57 \%$.
  • Any change > 2.57% is likely meaningful.

Typical Error (TE)

• Similar to the standard error of the measure.
• Represents the typical variation expected for an athlete between trials or sessions.
• Expressed in the real unit of measurement (e.g., meters per second for barbell velocity).
• Used to determine within-athlete or between-session variability.
• Calculation:
  1. Conduct a series of repeated trials.
  2. Calculate the standard deviation of the test outcome.
  3. Divide the standard deviation by the square root of two: $TE = \frac{\text{Standard Deviation}}{\sqrt{2}}$.
• Can be used to calculate a form of the coefficient of variation.

Conclusion

• Understanding and careful interpretation of reliability statistics and their associated confidence intervals are essential when reading sport science literature.
• Close reading of method sections of studies is essential to properly interpret the results.