Trouble with scalar drop_all

[Andry Haja]

Hi,

Can someone tell how to drop scalars?
I've tried:

scalar drop _all

but it's not working.

Thanks

[Mike Lacy]

What you have shown does not have syntactical problems, so there must be something else going on. Without any context for your problem, for others to guess t. Item 12 of the FAQ would be helpful to you in posing your question.

[Andry Haja]

Sorry... Here are some details with the code and results.

After -sts test, wilcoxon- I saved chi2, df, and p in scalars which I can show with ereturn list to copy and paste them in excel.
I have other variables that I would like to use in the same way but I wanted to drop the previous chi2, df and p with scalar drop _all but I still have the statistics from the previous test.

sts  test Secteurs,wilcoxon
...output ommited

estadd scalar chi2Secteurs = r(chi2), replace
added scalar:
       e(chi2Secteurs) =  64.57887
estadd scalar dfSecteurs = r(df), replace

added scalar:
         e(dfSecteurs) =  3
estadd scalar pSecteurs=( chi2tail(r(df), r(chi2))), replace

added scalar:
          e(pSecteurs) =  6.172e-14


ereturn list

scalars:
       e(chi2Secteurs) =  64.57886981891892
         e(dfSecteurs) =  3
          e(pSecteurs) =  6.17216083883e-14

scalar drop _all

 
sts  test Sexe,wilcoxon

estadd scalar chi2Sexe = r(chi2), replace
...

ereturn list

scalars:
       e(chi2Secteurs) =  64.57886981891892
         e(dfSecteurs) =  3
          e(pSecteurs) =  6.17216083883e-14
           e(chi2Sexe) =  19.54577121176661
             e(dfSexe) =  1
              e(pSexe) =  9.82179101923e-06

[Clyde Schechter]

The "scalars" you see in -ereturn list- are "virtual" scalars that are not effected by the -scalar drop- command. They persist until the next estimation command overwrites them with something else, or until you use the command -estimates drop-.

Clarification: all of the results of an estimation command that are stored in e() persist, collectively, until the next estimation command writes a new set of return values. If a particular estimation command returns e(this), and there is no such return value from the next estimation command, e(this) will, nevertheless, vanish with the next estimation command.

[Andry Haja]

Thanks Clyde,
I tried estimate drop _all but they still persist.

[Clyde Schechter]

Sorry, my mistake. It's -ereturn clear-, not -estimates drop-.

[Andry Haja]

Don't be sorry. It works fine. Thanks a lot Clyde.
God bless.