Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Re: A regression question.

From   "Michael Blasnik" <[email protected]>
To   <[email protected]>
Subject   st: Re: A regression question.
Date   Fri, 19 Mar 2004 16:29:44 -0500

In OLS, the coefficient on a dummy variable simply reflects the mean
difference in the dependent variable between cases where the dummy is 0 vs.
1.  Therefore, just like the overall mean of the residuals from OLS is 0,
the mean residuals will also be zero for any group defined by an included
dummy variable.

Michael Blasnik
[email protected]

----- Original Message ----- 
From: "Puneet K Singhal" <[email protected]>
To: <[email protected]>
Sent: Friday, March 19, 2004 1:47 PM
Subject: st: A regression question.

> I have expenditures post-stroke at one month intervals and want to predict
> these expenditures by number of months post stroke controlling for other
> covariates.  For this purpose, I am using dummy variables for the number
> months post stroke.  For example, if I have 12 months of follow-up I'm
> 11 dummies for each month up to 11 months and keeping 12th month as my ref
> category.  Now, after fitting this OLS I want to take the mean of
> values for each month and compare the predicted values across months.
When I
> do that what I find is that the mean predicted values are the same as the
> mean observed values for each month.  I just can't conceptualize why they
> the same.  When I stratify the predicted values by gender, I again get the
> same values as I would if I stratified the observed values by gender.
> Can someone help?
> ----------------------------------------------------
> Puneet K Singhal, M.S.
> Doctoral Candidate
> Pharmaceutical Health Services Research Program
> University of Maryland
> Baltimore, MD-21201
> Phone: 410-706-1805
> Fax: 410-706-1488
> ----------------------------------------------------

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index