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

RE: adjust after anova (was: st: Date: Wed, 18 May 2005 11:00:06 -0400)

From   "Visintainer, Paul" <[email protected]>
To   <[email protected]>
Subject   RE: adjust after anova (was: st: Date: Wed, 18 May 2005 11:00:06 -0400)
Date   Thu, 19 May 2005 12:22:22 -0400

Thanks, Ken.


Paul F. Visintainer, PhD
Professor and Program Director 
Health Quantitative Sciences
School of Public Health
New York Medical College 

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of
[email protected]
Sent: Thursday, May 19, 2005 11:22 AM
To: [email protected]
Subject: adjust after anova (was: st: Date: Wed, 18 May 2005 11:00:06

Paul Visintainer <[email protected]> asks:

> I'm using Stata v9.0.  I noticed that I don't get any output from
> -adjust- after running -anova-.  If I run the same model using
> -regress-, the -adjust- command works.  

I assume that you did something like

    sysuse auto
    anova price foreign
    adjust foreign

and saw an empty table.  And when you instead typed

    regress price foreign
    adjust foreign

-adjust- produced output.

With -adjust- after -anova- you would instead want to say

    anova price foreign
    adjust foreign=0


    adjust foreign=1

What is happening?  Why the difference?

Saying -adjust foreign- is really the same as saying
-adjust foreign = 0.2972973-.  In other words you are asking for
the prediction at the mean of the variable -foreign-.

In -anova- (except for variables specified in the -continuous()-
option), all right-hand-side variables are treated as
categorical.  With -anova price foreign-, the -foreign- variable
is treated as a 2 valued variable taking on values 0 and 1.  Any
value other than 0 or 1 for foreign when making predictions
(using the -predict- command) would produce a missing value.

-regress- is different.  It knows nothing about categorical
variables.  It treats all variables as if they were continuous.
So -predict- after -regress- allows a value of 0.2972973 for

-adjust- is really just a convenience tool for running -predict-.
The differences you are seeing are really due to how -predict-
works after -anova- versus -regress-.

You can

Though not as directly related, you might also want to see 

Ken Higbee    [email protected]
StataCorp     1-800-STATAPC

*   For searches and help try:

*   For searches and help try:

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