Related Experiment Videos
How to detect effects: statistical power and evidence-based practice in occupational therapy research
1School of Allied Health Sciences, University of Texas Medical Branch, Galveston 77555-1028, USA. kottenba@utmb.edu
Summary
Occupational therapy research often suffers from low statistical power, leading to missed significant findings and hindering evidence-based practice. Improving study power is crucial for advancing rehabilitation sciences.
Area of Science:
- Rehabilitation Sciences
- Occupational Therapy Research
- Statistical Methodology
Background:
- Occupational therapy interventions require rigorous evaluation to establish effectiveness.
- Statistical conclusion validity is a critical component of research quality.
- Low statistical power can lead to inaccurate conclusions about treatment effects.
Purpose of the Study:
- To assess the statistical power of published occupational therapy intervention studies.
- To identify the prevalence of low statistical power in this research area.
- To understand the implications of low power for evidence-based practice.
Main Methods:
- A systematic review and analysis of 30 occupational therapy intervention studies.
- Calculation of post hoc power coefficients for statistical hypothesis tests.
- Examination of median power values for small, medium, and large effect sizes.
Main Results:
- Median power values were .09 for small, .33 for medium, and .66 for large effect sizes.
- These low median power values indicate a high probability of Type II errors.
- Low power increases the likelihood of failing to detect true treatment effects.
Conclusions:
- The reviewed occupational therapy research is characterized by insufficient statistical power.
- Low power contributes to an increase in false negative findings, impeding scientific progress.
- Addressing low statistical power is essential for establishing evidence-based practice guidelines in occupational therapy.