Related Experiment Videos
When predictions don't predict
A L Speirs1, R H Asch, S J Silber
1Reproductive Biology Unit, Royal Women's Hospital, Melbourne, Australia.
The Australian & New Zealand Journal of Obstetrics & Gynaecology
|November 1, 1991
Summary
Misinterpreting statistical significance with multiple comparisons is a common issue. This study demonstrates how apparent significance can arise falsely, highlighting a critical flaw in statistical analysis.
Area of Science:
- Statistics
- Biostatistics
- Data Analysis
Background:
- Multiple comparisons are frequently applied in statistical analysis.
- Misinterpretation of statistical significance is a widespread problem.
- The 'multiple comparisons problem' can lead to erroneous conclusions.
Purpose of the Study:
- To highlight the common abuse of statistical methods.
- To demonstrate the misinterpretation of statistical significance.
- To show how false significance can appear with multiple comparisons.
Main Methods:
- Statistical analysis involving multiple comparisons.
- Demonstration of a scenario where significance is falsely detected.
- Illustrative examples of statistical misinterpretation.
Main Results:
- Apparent statistical significance was demonstrated where none truly existed.
- The study provides evidence of the pitfalls of uncorrected multiple comparisons.
- The findings underscore the need for careful statistical interpretation.
Conclusions:
- Misinterpretation of statistical significance is a critical issue.
- Proper statistical methods are essential to avoid false positives.
- Researchers must be cautious when interpreting results after multiple comparisons.