Quantcast

xtmixed - standardized resuduals plot

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

xtmixed - standardized resuduals plot

Hobst
I have used xtreg to fit a 3-level hierarchical model. To test if the residuals on the different levels are in fact nomally distributed i would like to plot histograms of the standardized residuals for level-1 r(ijk), level-2 r(jk) and level-3 r(k).  I know a way , using gllapred after gllamm and then plot the histogram, but is there something similar to gllapred to use after xtmixed?

Thank you in advance for yur support
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: xtmixed - standardized resuduals plot

Maarten buis
--- On Fri, 3/9/10, Hobst wrote:
> I have used xtreg to fit a 3-level hierarchical model. To
> test if the residuals on the different levels are in fact
> nomally distributed i would like to plot histograms of
> the standardized residuals for level-1 r(ijk), level-2
> r(jk) and level-3 r(k).

You can get the residuals using -predict-, see the example
below or -help xtmixed postestimation-. I used -hangroot-
as a more convenient way of comparing the histogram with
the theoretical normal distribution. To install it, type in
Stata -ssc install hangroot-.

*------------ begin example ---------------------
webuse productivity, clear
xtmixed gsp private emp hwy water other unemp ///
        || region: || state:, mle

predict re, re
predict re*, reff

// these variables are used to make sure that each
// higher level unit (region or state) contributes
// only one observation when potting their level
// residuals
bys region (re1): gen byte regionmark = _n == 1
bys state (re2) : gen byte statemark  = _n == 1

hangroot re, ci
hangroot re1 if regionmark, ci
hangroot re2 if statemark, ci
*---------------- end example --------------------
(For more on examples I sent to the Statalist see:
http://www.maartenbuis.nl/example_faq )

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


     

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: xtmixed - standardized resuduals plot

Hobst
Thank you very much Maarten, thats exactly what i was looking for!

Could you maybe, if you have time, have a very quick look at another of my questions: "-xtmixed- Crossed effets & hierarchy"

I dont know anybody else to ask, and i cant find any more theory about my problem. This would be really nice.

This is the link: http://statalist.1588530.n2.nabble.com/xtmixed-Crossed-effets-hierarchy-td5491619.html#a5491619

regards, Toby

Tobias Friedli
Bachelor Student at the University of Zurich



On 03.09.2010, at 13:22, Maarten buis wrote:

> --- On Fri, 3/9/10, Hobst wrote:
>> I have used xtreg to fit a 3-level hierarchical model. To
>> test if the residuals on the different levels are in fact
>> nomally distributed i would like to plot histograms of
>> the standardized residuals for level-1 r(ijk), level-2
>> r(jk) and level-3 r(k).
>
> You can get the residuals using -predict-, see the example
> below or -help xtmixed postestimation-. I used -hangroot-
> as a more convenient way of comparing the histogram with
> the theoretical normal distribution. To install it, type in
> Stata -ssc install hangroot-.
>
> *------------ begin example ---------------------
> webuse productivity, clear
> xtmixed gsp private emp hwy water other unemp ///
>        || region: || state:, mle
>
> predict re, re
> predict re*, reff
>
> // these variables are used to make sure that each
> // higher level unit (region or state) contributes
> // only one observation when potting their level
> // residuals
> bys region (re1): gen byte regionmark = _n == 1
> bys state (re2) : gen byte statemark  = _n == 1
>
> hangroot re, ci
> hangroot re1 if regionmark, ci
> hangroot re2 if statemark, ci
> *---------------- end example --------------------
> (For more on examples I sent to the Statalist see:
> http://www.maartenbuis.nl/example_faq )
>
> Hope this helps,
> Maarten
>
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
> http://www.maartenbuis.nl
> --------------------------
>
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: xtmixed - standardized resuduals plot

flori
This post has NOT been accepted by the mailing list yet.
In reply to this post by Maarten buis
Hi Maarten,

I found this link while I was looking for STATA code to predict the level-1 and level-2 residuals after xtmixed.

I figured out that in xtmixed, the two examples that you gave (predict re, re / predict re*, reff ) do not provide the level-1 and level-2 residauls.

See below:

 -- re is the residual option in xtmixed and it provides you w/ an overall residual of the model, it includes both the fixed and the random piece, so far I have not  found the corresponding prediction in gllapred

 -- reff is the reffects option in xtmixed and it provides you w/ the BLUPs. In gllapred you obtain the same thing if you use the option "u"

 -- you can get level-1 residuals in xtmixed if you type "rstandard" and you can get the same level-1 residuals in gllapred if you type "pearson"

 -- however, in xtmixed I have not found way to get just the level 2 residuals BUT in gllapred you can type "ustd" and you will get them

Best,

Flori
Loading...