[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

From |
"Barth Riley" <BarthRiley@comcast.net> |

To |
<statalist@hsphsun2.harvard.edu> |

Subject |
st: Regression question |

Date |
Wed, 13 Feb 2008 15:30:34 -0600 |

Hello statalist, I am conducting a CHAID model to examine potential interaction of several categorical variables in explaining variance in a dependent variable. CHAID models do not have the capability of forcing covariates into the model, i.e., examining the effect of the predictors on the dependent after controlling for the effect of the covariate(s). To get around this limitation, I want to perform a regression analysis of the same dependent variable with the covariate as the sole predictor, then use the dependent variable partialling out the effect of the covariate (call it y prime) as the target variable in the CHAID model. Would y prime be the value predicted by the regression model? What I'm struggling with is that if the relationship between the dependent and the covariate is 0, the values predicted from the regression will bear no resemblance to the original dependent variable. What I want instead is the dependent variable with the effect of the covariate partialled out; if the covariate has no relationship to the dependent, then, in effect, nothing should be partialled out, i.e.., y prime = y. Is it possible to do this? Barth * * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

**Follow-Ups**:**Re: st: Regression question***From:*Maarten buis <maartenbuis@yahoo.co.uk>

- Prev by Date:
**st: st predicting probabilities after ivprobit** - Next by Date:
**st: Accessing command line parameters** - Previous by thread:
**st: st predicting probabilities after ivprobit** - Next by thread:
**Re: st: Regression question** - Index(es):

© Copyright 1996–2015 StataCorp LP | Terms of use | Privacy | Contact us | What's new | Site index |