Skip to content

Retrieve regression summary results after rq

6 messages · John Kane, Camila Mendes, David Winsemius

Camila Mendes
# 
Hi all,

I am using quantile regression with svy design. I want to retrieve
summary regression statistics (std error, p-value), since I don't have
any in my output:

Commands:

clus1_d<- svydesign(id=~cd002_co, weights=~wtper, strata=~str, data=data)
bclus1<-as.svrepdesign(clus1_d,type="bootstrap",replicates=100)
fit1<- withReplicates(bclus1,quote(coef(rq(newm428b~sch_new_2+sch_new_3+sch_new_4,
tau=c(0.05,0.25,0.5,0.75,0.95),weights=wtper))))

Output:

coefficients(fit1)
            tau= 0.05 tau= 0.25     tau= 0.50 tau= 0.75 tau= 0.95
(Intercept)      2340      2980  3.300000e+03      3750 4360.0000
sch_new_2        185       -30 -5.500000e+01      -250 -360.0000
sch_new_3        155        20  5.032317e-16      -200 -245.0000
sch_new_4         65        20 -1.500000e+01      -200 -309.9998

Any help will be appreciated.

Best,

C.
John Kane
# 
summary(fit1) perhaps/

John Kane
Kingston ON Canada
____________________________________________________________
FREE 3D MARINE AQUARIUM SCREENSAVER - Watch dolphins, sharks & orcas on your desktop!
Camila Mendes
# 
This does not work at all...
On Wed, Sep 26, 2012 at 12:04 PM, John Kane <jrkrideau at inbox.com> wrote:
John Kane
# 
Strange, I have never used it but presumably you are using the package ?survey?? The manual suggests summary(xx) should work. 

What results or error messages do you get?


John Kane
Kingston ON Canada
____________________________________________________________
GET FREE SMILEYS FOR YOUR IM & EMAIL - Learn more at http://www.inbox.com/smileys
Works with AIM?, MSN? Messenger, Yahoo!? Messenger, ICQ?, Google Talk? and most webmails
David Winsemius
#
On Sep 26, 2012, at 8:50 AM, John Kane wrote:

            

      
      
      
    

        
She didn't return a model object as fit1. She return only the results of the coef() on the rq(model) object.

Perhaps:

summ1 <- withReplicates(bclus1,
                        quote( summary( rq( newm428b~sch_new_2+sch_new_3+sch_new_4,
                                      tau=c(0.05,0.25,0.5,0.75,0.95),weights=wtper))))

Untested, and I've never used 'survey' and 'rq' packages together so this is based entirely on analogous reasoning.)

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Camila Mendes
    

    

      
      #
    

  


  

      
Hi all, I appreciate your help.

Here are a sample of my script. I appreciate any help.

David, I will go ahead and try your suggestion as well.

Thanks.

----------------------------------------------------
library(survey)
clust<- svydesign(id=id, weights=wtper, strata=strat, data=data)
summary(clust)
library(quantreg)
library(lmtest)
library(car)
library(stats)

#OLS Model
fitols<-(svyglm(y~sch_2+sch_3+sch_4, clust))

# Quantile reg

bclus1<-as.svrepdesign(clust,type="bootstrap",replicates=100)
fit1<- withReplicates(bclus1,quote(coef(rq(y~sch_2+sch_3+sch_4,
tau=c(0.05,0.25,0.5,0.75,0.95),weights=pesonorm))))

# Model diagnosis
coefficients(fit1)
            tau= 0.05 tau= 0.25     tau= 0.50 tau= 0.75 tau= 0.95
(Intercept)      2340      2980  3.300000e+03      3750 4360.0000
sch_2        185       -30 -5.500000e+01      -250 -360.0000
sch_3        155        20  5.032317e-16      -200 -245.0000
sch_4         65        20 -1.500000e+01      -200 -309.9998

confint(fit1,level=0.95)
2.5 % 97.5 %

summary(fit1)
   tau= 0.05        tau= 0.25        tau= 0.50        tau= 0.75
 Min.   :  65.0   Min.   : -30.0   Min.   : -55.0   Min.   :-250.0
 1st Qu.: 132.5   1st Qu.:   7.5   1st Qu.: -25.0   1st Qu.:-212.5
 Median : 170.0   Median :  20.0   Median :  -7.5   Median :-200.0
 Mean   : 686.2   Mean   : 747.5   Mean   : 807.5   Mean   : 775.0
 3rd Qu.: 723.8   3rd Qu.: 760.0   3rd Qu.: 825.0   3rd Qu.: 787.5
 Max.   :2340.0   Max.   :2980.0   Max.   :3300.0   Max.   :3750.0
   tau= 0.95
 Min.   :-360.0
 1st Qu.:-322.5
 Median :-277.5
 Mean   : 861.3
 3rd Qu.: 906.2
 Max.   :4360.0