Journal article
Stepwise Regression: A Caution
Political methodology, Vol.5(2), pp.213-240
01/01/1978
Abstract
Although quantitative researchers in political science and sociology have used stepwise regression to select independent variables and to rank them in terms of their impact on a dependent variable, the technique should not be employed for these purposes. It is inappropriate for selecting independent variables because its choices do not necessarily accord with the theoretical structure linking the variables. It also rests on an improper partitioning of variance scheme. Researchers selecting independent variables should be guided largely by theory, rather than by stepwise regression. To judge the comparative impact of independent variables, researchers can rely on well-developed methods to estimate effects in single- and multi-equation models.
Details
- Title: Subtitle
- Stepwise Regression: A Caution
- Creators
- Michael S. Lewis-Beck
- Resource Type
- Journal article
- Publication Details
- Political methodology, Vol.5(2), pp.213-240
- Publisher
- Geron-X, Inc
- ISSN
- 0162-2021
- Language
- English
- Date published
- 01/01/1978
- Academic Unit
- Political Science
- Record Identifier
- 9984025639302771
Metrics
31 Record Views