unable to compute sqrt function due to error message

Question

Trying to do a one-sided Wald test based on a simple quantile regression on Stata 15.0

qreg volume if period == 2010

I want to test whether the coefficient (constant) of my qreg <= 10

Based on this post , I wrote this code:

local to_test = sign(_b[_cons]-1 )
display "H0: volume <=10 p-value=" ttail(r(df_r),`to_test' * sqrt(r(F)))

I end up with this error message:

H0: volume <=10 p-value=unknown function *sqrt()
r(133);

Why is this error message appear for a simple sqrt function?

Please provide us with a [mcve] using Stata's `dataex` command so we can try to help you. Without sample data we cannot replicate anything. — , Aug 07 '18 at 20:24

score 1 · Accepted Answer · edited Aug 08 '18 at 01:08

To perform a one-sided test, you need to perform the corresponding two-sided Wald test first. Then you can use the results to calculate the test statistic and p-value for the one-sided test.

Your code does not work because you are not doing the Wald test, so r(df_r) and r(F) are not defined. When you try to take the square root of a quantity that has not been calculated, Stata gives you a cryptic error. There are other problems, like having a 1 instead of a 10 in the sign calculation.

I think breaking up the problem and displaying all of the pieces before doing the calculation that is giving you a problem is a good way to solve this.

Here's an example showing this on the cars dataset (an alternative to using dataex):

. sysuse auto, clear
(1978 Automobile Data)

. qreg price weight length i.foreign, nolog

Median regression                                   Number of obs =         74
  Raw sum of deviations  71102.5 (about 4934)
  Min sum of deviations 54411.29                    Pseudo R2     =     0.2347

------------------------------------------------------------------------------
       price |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      weight |   3.933588   1.328718     2.96   0.004     1.283543    6.583632
      length |  -41.25191   45.46469    -0.91   0.367    -131.9284    49.42456
             |
     foreign |
    Foreign  |   3377.771   885.4198     3.81   0.000     1611.857    5143.685
       _cons |   344.6489   5182.394     0.07   0.947     -9991.31    10680.61
------------------------------------------------------------------------------

. local sign_cons = sign(_b[_cons] - 10)

. display "`sign_cons'"
1

. display r(df_r)
.

. display r(F)
.

. test _b[_cons] = 10

 ( 1)  _cons = 10

       F(  1,    70) =    0.00
            Prob > F =    0.9487

. display r(df_r)
70

. display r(F)
.00416983

. display "Ho: _cons <= 10 p-value = " ttail(r(df_r),`sign_cons'*sqrt(r(F)))
Ho: _cons <= 10 p-value = .47434855

unable to compute sqrt function due to error message

1 Answers1