Related Experiment Videos
Choosing the correct unit of analysis in Medical Care experiments
Medical Care
|December 1, 1984
Summary
A common statistical error in health research design was identified, affecting 71% of studies. This design flaw inflates experimental power, potentially leading to incorrect conclusions about healthcare provider performance.
Area of Science:
- Health Services Research
- Biostatistics
- Medical Research Methodology
Background:
- Analysis of statistical methodology in prominent medical journals from 1975-1980.
- Focus on experimental design and analysis errors in health research.
Purpose of the Study:
- To examine the prevalence of a specific statistical error in health research.
- To understand the impact of this error on study findings and conclusions.
Main Methods:
- Review of 28 health care experiments published in Lancet, NEJM, and Medical Care.
- Identification and quantification of a specific analytical error in experimental design.
Main Results:
- The identified error was present in 20 out of 28 (71%) reviewed health care experiments.
- This error typically inflates the statistical power of experiments, making it easier to detect differences.
Conclusions:
- The prevalent error involves using patient observations inappropriately as the unit of analysis for provider-level conclusions.
- Explicitly defining hypotheses and target populations prospectively may prevent this common experimental design error.