Categorizing Continuous Variables
Author:
- 29 September 2014
- book chapter
- Published by Wiley
Abstract
Continuous variables are often converted to categorical variables by grouping values. For descriptive purposes, this can be useful but problems can arise when this is done as part of a data analysis. This article discusses relevant issues with categorization of explanatory variables in regression. Effects of categorization are discussed, such as loss of information, power, and efficiency. This is considered first for categorizing into two groups and then for categorizing into several groups. Rather than group according to observed values with several variables, one might group according to a weighted sum of the variables–a prognostic index.Keywords
This publication has 17 references indexed in Scilit:
- Dose-Response and Trend Analysis in EpidemiologyEpidemiology, 1995
- Dangers of Using "Optimal" Cutpoints in the Evaluation of Prognostic FactorsJNCI: Journal of the National Cancer Institute, 1994
- Bivariate median splits and spurious statistical significance.Psychological Bulletin, 1993
- Efficiency Loss from Categorizing Quantitative Exposures into Qualitative Exposures in Case-Control StudiesAmerican Journal of Epidemiology, 1992
- Effects of mismodelling and mismeasuring explanatory variables on tests of their association with a response variableStatistics in Medicine, 1988
- Two issues concerning the analysis of grouped dataEuropean Journal of Epidemiology, 1987
- Effect of Categorizing a Continuous Covariate on the Comparison of Survival TimeJournal of the American Statistical Association, 1986
- The Cost of DichotomizationApplied Psychological Measurement, 1983
- Grouping for Testing Trends in Categorical DataJournal of the American Statistical Association, 1972
- Note on GroupingJournal of the American Statistical Association, 1957