Program like outreg2 but for t-test outputs

[Victoria Rogers]

Is there a useful program like outreg2 but for t-test outputs?

I came across one way to enter every value manually but there's probably a much better program.

[Attaullah Shah]

Unfortunately, there is nothing perfect available as outreg2 for ttest. For one sample ttest, a work around is to use reg command and then use outreg2. For example the following two commands give the same results

ttest returns==0
reg returns

If you are interested in two sample ttest, then you might use _^estpost see, http://fmwww.bc.edu/repec/bocode/e/estpost.html

[Victoria Rogers]

Thank you Attaullah! I'll look into -estpost-. While I'm trying to work with -estpost- I'd also appreciate other suggestions to tabulate t-test outputs

[Victoria Rogers]

Is it normal that the t-test output converts to this: http://i.imgur.com/quXoEJG.png by using estpost and esttab?

Is there a nicer way to tabulate t-test results?

[Roman Mostazir]

I use "putexcel" for my t-test.

putexcel set "T-test.xlsx", sheet("t_test") replace
putexcel A1=("Variable") B1=("Control (Mean)") C1=("Control (SD)") D1=("Treatment (Mean)") E1=("Treatment(SD)") F1=("p value")

local row=2
foreach var of varlist x1 x2 x3 x4 {
   qui ttest `var',by(treatment)
putexcel A`row' = ("`var'")
putexcel B`row' = (r(mu_1))
putexcel C`row' = (r(sd_1))
putexcel D`row' = (r(mu_2))
putexcel E`row' = (r(sd_2))
putexcel F`row' = (r(p))
local ++row
}
****

[Victoria Rogers]

I still haven't figured out how to work with "putexcel" but thank you anyways for the nice suggestion.
I've also seen that a normal regression has the same outcomes as the t-test outcome, in the case of equal variances, which isn't equal in my case.

Does anyone know how to convert a "ttest alpha, by(gender) unequal" into a regression with the same t-value etc. So, that I can use the easy outreg2 instead of "putexcel"

[Victoria Rogers]

Putexcel is for Stata13 only and I have version 12. Does anyone have a nice suggestion?

[Attaullah Shah]

Just to update this post, I am listing asdoc as an option to report ttest results. asdoc can be installed from

ssc install asdoc

and can work with almost all Stata command. The following examples are borrowed from the asdoc's help file

Example 67 : Appending t-test results with rowappend option
`

    sysuse auto, clear
        asdoc ttest rep78==0, replace title(T-test results : mean == 0)
        asdoc ttest price==0, rowappend
        asdoc ttest mpg==0, rowappend
        asdoc ttest turn==0, rowappend
        asdoc ttest weight==0, rowappend
        asdoc ttest length==0, rowappend

Example 68 : Repeat the above tests, requesting specific statistics

      sysuse auto, clear
        asdoc ttest rep78==0, replace title(T-test results : mean == 0) stat(obs mean se df t)
        asdoc ttest price==0, rowappend stat(obs mean se df t)
        asdoc ttest mpg==0, rowappend stat(obs mean se df t)
        asdoc ttest turn==0, rowappend stat(obs mean se df t)
        asdoc ttest weight==0, rowappend stat(obs mean se df t)
        asdoc ttest length==0, rowappend stat(obs mean se df t)

8.2 Two-sample t-test using groups i.e. with option the by()

Example 69: Two-sample t-test using groups i.e. with the option by()
`
asdoc ttest mpg, by(foreign) replace


8.3 Two-sample t-test using variables

Example 70: Two-sample t-tes using variables
`

 asdoc ttest mpg==price, replace

And to add similar tests to the same table, we can use rowappend

Example 71: append more tests
`

       asdoc ttest trunk==price, rowappend

8.4 Two sample test over groups

Example 72: Two sample test over groups
`

   bysort foreign: asdoc ttest mpg == price, replace

[Ingrid Lambert]

I have two groups by dummy variables.
I need to create a table with difference in means for a variable in treated and control groups and report it to outreg table.
In the first column you should include the means and, in parenthesis, the standard deviation of the treated and control firms respectively, corresponding to each variable.
The second column should include the median.
The third column should include the difference in the two means.
And the fourth column should include the t-statistic of the mean difference.

The table below is incorrect.
What I need is Mean1-Mean2 which will be TobinsQMeanfromTreatedGroup - TobinsQMeanfromControlGroup.

. bysort CUT: asdoc ttest TobinsQ == CUT, replace

Paired t test : TobinsQ CUT 

 obs
  Mean1
    Mean2
    dif
   St_Err
   t_value
   p_value

0
 219400
  8.145
    8.145
     0
   1.44
      0
      1

1
 910
  1.552
    1.552
     0
   .058
      0
      1

[Attaullah Shah]

I think there is some parsing issue, which I have noted and shall correct in the next update of asdoc. At the moment, please run the paired t-test separately for each group of CUT.
Here is an example

sysuse auto, clear  
asdoc ttest price = turn if foreign == 1, replace  
asdoc ttest price = turn if foreign == 0, rowappend

In your example, the variable CUT seems to have only two groups, i.e. 0 and 1, so if you are conducting ttest of TobinsQ = CUT if CUT == 0, then why not just

 ttest TobinsQ, by(CUT)

[Ingrid Lambert]

The wonderful table below sadly lacks the following:

and, in parenthesis, the standard deviation of the treated and control firms respectively, corresponding to each variable.
The second column should include the median

asdoc ttest TobinsQ,by(CUT)
  asdoc ttest FirmSize,by(CUT) rowappend
  asdoc ttest CFtoAssets,by(CUT) rowappend
  asdoc ttest CashtoAssets,by(CUT) rowappend
  asdoc ttest LTLev,by(CUT) rowappend
  asdoc ttest Inv,by(CUT) rowappend

Two-sample t test with equal variances   

     obs1
     obs2
     Mean1
    Mean2
     dif
     St_Err
    t_value
    p_value

 TobinsQ by CUT: 0 1
   219400
     910
   8.145
   1.563
   6.583
   22.351
   .3
   .768

 FirmSize by CUT: 0 1
   294702
     1143
   4.641
   3.821
   .82
   .081
   10.15
   0

 CFtoAssets by CUT:~1
   253058
     1115
   -.713
   -.009
   -.705
   1.933
   -.35
   .716

 CashtoAssets by CU~1
   272978
     1142
   .148
   .106
   .043
   .006
   6.85
   0

 LTLev by CUT: 0 1
   292995
     1142
   .281
   .18
   .101
   .34
   .3
   .768

 Inv by CUT: 0 1
   225241
     1045
    253.694
   104.993
   148.701
   726.377
   .2
   .838

[Attaullah Shah]

It is really hard for developers to foresee all possible variations needed by users. The formatting and contents of the ttest output from the asdoc command were really puzzling for me at the time of development.

Inside the help file of asdoc, you can see several options that can be used with the ttest command, specifically, read Section 8 of the help file. Still, if you want to further customize your table, you can read materials related to option row(). To get started, first read this article https://fintechprofessor.com/2018/09...stata-ms-word/

Then you may like to read https://fintechprofessor.com/2019/09...f-asdoc-stata/

[Ingrid Lambert]

then why not just

ttest TobinsQ, by(CUT)

Dear prof. Shah, I wish the life was that easy. ^_^
I need to report results using nice output.

I also have a question about asdoc. Is it possible to remove the columns I do not need? Say, obs and St_Err.
I need to add standard deviation of the treated and control firms and the medians instead...

In the first column you should include the means and, in parenthesis, the standard deviation of the treated and control firms respectively, corresponding to each variable.
The second column should include the median.
The third column should include the difference in the two means.
And the fourth column should include the t-statistic of the mean difference.

[Ingrid Lambert]

It is really hard for developers to foresee all possible variations needed by users. The formatting and contents of the ttest output from the asdoc command were really puzzling for me at the time of development.

Inside the help file of asdoc, you can see several options that can be used with the ttest command, specifically, read Section 8 of the help file. Still, if you want to further customize your table, you can read materials related to option row(). To get started, first read this article https://fintechprofessor.com/2018/09...stata-ms-word/

Then you may like to read https://fintechprofessor.com/2019/09...f-asdoc-stata/

Dear prof. Shah, I find asdoc to be the most comprehensive in terms of exporting the results of T Test.

putexcel worked for me but is it not suitable simply because I require the table to be suitable for a research paper.

Your comment on an option row() is indeed helpful.

I sincerely hope such feedback from users helps you to improve asdoc!

[Attaullah Shah]

I suggested reading the help file, which I think you have not noticed. So let me copy and paste from the help file. The point is you can restrict the output to your desired statistics using option stat().

6.stat(mean se df obs t p sd dif): Without stat() option, asdoc reports the number of observations
(obs), mean, standard error, t-value, and p-value with t-tests. However, you can select all or a few
statistics using the stat option.

    With ttest command, we can use the following options of asdoc to control asdoc behavior.  
(1) replace/append
(2) save(filename)
(3) title(text)
(4) fs(#)
(5) hide.
(6) stats()
(7) rowappend.
These options are discussed in detail in  Section 1.. Option stats and rowappend are discussed   below:


    ----+ Statistics +------------------------------------------------------
       n        Number of observations
       mean     Arithmetic mean
       se       Standard error
       df       degrees of freedom
       obs      Number of observations
       t        t-value
       p        p-value
       sd       standard deviation
       dif      difference in means if two-sample t-test
 -----------------------------------------------------------------------------