Invalid name error

[Akanksha Aggarwal]

Hi

I am trying to run a quantile regression and then plot the ols and quantile regression coefficients for variable treatment with confidence intervals

 grqreg treatment , cons ci ols olsci

Ran the obove code but keeps showing the error below

invalid name
r(198);


Please help.

[Rich Goldstein]

since -grqreg- can only be used after a quantile regression command, please show us what the preceding command was as there may be an inconsistency between the commands

[Akanksha Aggarwal]

I am sorry

qreg fee_adjusted i.treatment i.state ib2007.year age i.sector , vce(robust)
grqreg treatment , cons ci ols olsci

Error:

0b.treatment invalid name
r(198);

[Carlo Lazzaro]

Akanksha:
the error rests on the way you called the community-contributed command (as you're kidly requested by FAQ to detail) -grqreg-, as you can see from the following toy-example:

webuse auto, clear
. qreg price mpg weight headroom
Iteration  1:  WLS sum of weighted deviations =  68749.068

Iteration  1: sum of abs. weighted deviations =   66982.32
Iteration  2: sum of abs. weighted deviations =  63996.616
Iteration  3: sum of abs. weighted deviations =  63956.522

Median regression                                   Number of obs =         74
  Raw sum of deviations  71102.5 (about 4934)
  Min sum of deviations 63956.52                    Pseudo R2     =     0.1005

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         mpg |  -77.84984   107.8287    -0.72   0.473    -292.9075    137.2078
      weight |    .456869   .8347793     0.55   0.586    -1.208046    2.121784
    headroom |  -160.2684   497.2776    -0.32   0.748    -1152.057    831.5205
       _cons |   5986.834   4584.871     1.31   0.196    -3157.402    15131.07
------------------------------------------------------------------------------

. grqreg, cons ci ols olsci

As you can see, there are no variables between -grqreg- and comma.

[Akanksha Aggarwal]

Hi Carlo. Thanks for the response.

 qreg fee_adjusted i.treatment i.state ib2007.year age i.sector , vce(robust)
grqreg  , cons ci ols olsci

When i run the above , the following error is still shown:

0b.treatment invalid name
r(198);

[Carlo Lazzaro]

Akankska:
what if you run:

 
 xi: qreg fee_adjusted i.treatment i.state ib2007.year age i.sector , vce(robust)  
 grqreg, cons ci ols olsci

[Akanksha Aggarwal]

2007b.year invalid name
r(198);


Now it shows the above error, Carlo.

[Rich Goldstein]

it appears that -xi- does not allow you to set the baseline (thus, "ibX" is not valid) and that grqreg appears to have a problem with factor variable notation; try just "i.year" instead)

[Carlo Lazzaro]

Akankska:
being a bit old-fashioned, the community-contributed command -grqreg- does not support -fvvarlist- notation.
Therefore, you cannot choose the reference category youeself, but should Stata do it and calculate what you're intrested in according to Stata outcome table.
That said, you should go:

 
  xi: qreg fee_adjusted i.treatment i.state i.year age i.sector , vce(robust)    grqreg, cons ci ols olsci

[Akanksha Aggarwal]

Thanks so much for your help Carlo and Rich. The following code worked.

xi: qreg fee_adjusted i.treatment i.state i.year age i.sector [fw=round(weight)] , vce(robust)
grqreg  , cons ci ols olsci

I have couple of follow up questions too:

1) The code shows 37 graphs and I am only interested in one of them which is barely even properly visible. Can something be done?
2) I want to also cluster the standard errors at the state level since the treatment is being given at the state level. Can that be done?


Thanks so much.

[Carlo Lazzaro]

Akankska:
1) as far as I know, there's no way to magnify the graph you're intrested in;
2) no, you can't, as -qreg- does not support the -vce(cluster clusterid)- option for non-default standard error.