Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Dichotomizing continuous predictors in multiple regression: a bad idea.

Patrick Royston1, Douglas G Altman, Willi Sauerbrei

  • 1MRC Clinical Trials Unit, 222 Euston Road, London NW1 2DA, UK. patrick.royston@ctu.mrc.ac.uk

Statistics in Medicine
|October 12, 2005
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Population attributable fraction.

BMJ (Clinical research ed.)·2018
Same author

Reporting guidelines for oncology research: helping to maximise the impact of your research.

British journal of cancer·2018
Same author

Updating standards for reporting diagnostic accuracy: the development of STARD 2015.

Research integrity and peer review·2018
Same author

Do declarative titles affect readers' perceptions of research findings? A randomized trial.

Research integrity and peer review·2018
Same author

Choosing important health outcomes for comparative effectiveness research: An updated systematic review and involvement of low and middle income countries.

PloS one·2018
Same author

The INTERGROWTH-21<sup>st</sup> fetal growth standards: toward the global integration of pregnancy and pediatric care.

American journal of obstetrics and gynecology·2018
Same journal

Optimal Weighted Tests for Replication Studies and the 'Two-Trials Rule' With Multiple Hypotheses.

Statistics in medicine·2026
Same journal

Identifiable Copula-Double-Cox Models: A Fully Parametric Framework for Dependent Right-Censored Survival Data.

Statistics in medicine·2026
Same journal

Moving From Individualized Risk-Based Prevention to Benefit-Based Prevention: Estimating Individualized Life-Years Gained From Prevention Services as a Basis for Eligibility.

Statistics in medicine·2026
Same journal

A Mixture of Distributed Lag Non-Linear Models to Account for Spatially Heterogeneous Exposure-Lag-Response Associations.

Statistics in medicine·2026
Same journal

Practical Considerations for Gaussian Process Modeling for Causal Inference in Quasi-Experimental Studies With Panel Data.

Statistics in medicine·2026
Same journal

Covariate Adjustment for Wilcoxon Two Sample Statistic and Test.

Statistics in medicine·2026
See all related articles

Dichotomizing continuous variables in medical research, while seemingly simple, can lead to significant issues like reduced statistical power and bias. Researchers should avoid this practice, especially in regression models, to maintain data integrity.

Area of Science:

  • Medical Research
  • Biostatistics
  • Clinical Trials

Background:

  • Continuous variables are frequently converted into categorical variables in medical research.
  • Dichotomization, the creation of two categories from continuous data, is a common but problematic approach.
  • This simplification can introduce statistical challenges rather than solve them.

Purpose of the Study:

  • To detail the issues associated with dichotomizing continuous variables into two groups.
  • To highlight the negative impacts of this practice on statistical power and confounding.
  • To demonstrate the bias introduced by using data-derived 'optimal' cutpoints.

Main Methods:

  • Detailed case study analysis of a randomized trial in primary biliary cirrhosis.
  • Examination of the statistical consequences of dichotomizing continuous predictor variables.

Related Experiment Videos

  • Argument against the necessity and utility of dichotomization in statistical analysis.

    • Main Results:

    • Dichotomization results in a considerable loss of statistical power.
    • This practice can lead to residual confounding, masking true relationships.
    • Using data-derived cutpoints introduces significant bias into analyses.

    Conclusions:

    • Dichotomization of continuous data is unnecessary for robust statistical analysis.
    • The practice should be avoided for explanatory variables within regression models.
    • Maintaining continuous data preserves statistical power and reduces bias.