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

Re: st: Equality of means and pweights

From   "DANIEL CARO" <>
Subject   Re: st: Equality of means and pweights
Date   Tue, 8 Jul 2003 18:34:32 -0500

Thanks a lot May. I've been able to test the significance using the dummies
and regress (xi).


                      May Boggess                                                                                  
                      <>             To:                    
                      Sent by:                         cc:                                                         
                      owner-statalist@hsphsun2.        Subject:  Re: st: Equality of means and pweights            
                      08/07/2003 05:01 p.m.                                                                        
                      Please respond to                                                                            


I am not sure what you mean when you say the dependent
variable is ordinal AND continuous,
so I am just going to assume it's a continuous numerical variable.
I'll be calling this variable y.

I'll call the independent variable x, which I'll assume
is numerical, but discrete (discrete because I am supposing
you want to test the means of y over the levels of the factor x).

I'll also suppose the pweights are in a variable called p.

To keep the observations between the first and forth quartile
then drop the observations you don't need by doing:

summarize y, detail
drop if y > r(p75)
drop if y < r(p25)

(By changing y to x in these lines of code you will
drop the observations that are between the first and forth quartile
of your independent variable x.)

To do an equality of means I am supposing you want an
ANOVA table, and (as you know) Stata's -anova- doesn't have a
pweight option. So, here is something you could try.

svyset [pweight=p]

will tell Stata that p is the variable with your pweights in there.

xi: svyregress y i.x

will do the regression. Then use -test- to get it to tell
you the significance of x:

test _Ix_1 _Ix_2

(this list should contain all the dummy variables created for x)

hope this is some help,

May Boggess

On Fri, 2003-07-04 at 16:43, DANIEL CARO wrote:
> Dear Statalisters,
>   I'm trying to test the equality of means of a dependent variable
>   and continuous) between the first and fourth quartile of an independent
>   variable (in quartiles). Do you know of any command to perform this
>   that can handle pweights?
>   Thanks a lot for your help,
>   Daniel
> *
> *   For searches and help try:
> *
> *
> *

*   For searches and help try:

*   For searches and help try:

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