Related Experiment Videos
Why carve up your continuous data?
Steven V Owen1, Robin D Froman
1Department of Pediatrics, School of Medicine, and Center for Epidemiology & Biostatistics, University of Texas Health Science Center at San Antonio, USA.
Research in Nursing & Health
|November 16, 2005
Summary
Researchers often unnecessarily categorize continuous data, leading to flawed statistical power and errors. This study highlights the risks of data carving and offers better alternatives for social, biophysical, and health research.
Area of Science:
- Social Sciences
- Biophysics
- Health Research
Background:
- Continuous data are frequently used in social, biophysical, and health research.
- Researchers sometimes divide continuous data into ordered categories, a practice known as data carving.
- This categorization can lead to significant information loss and statistical inaccuracies.
Purpose of the Study:
- To examine the impact of data carving on statistical results.
- To identify instances of data carving in nursing literature.
- To propose alternative methods to data categorization.
Main Methods:
- Analysis of selected nursing research articles to identify data carving practices.
- Illustrative examples demonstrating how categorization affects statistical outcomes.
- Development of alternative data analysis strategies.
Main Results:
- Data carving was found in selected nursing literature.
- Unnecessary categorization can lead to reduced or inflated statistical power.
- Erroneous statistical results, including Type I and Type II errors, can arise from data carving.
Conclusions:
- Data carving is a problematic practice that distorts research findings.
- Researchers should avoid unnecessary categorization of continuous data.
- Alternative analytical approaches offer more accurate and reliable results.