T-Test Using Three Groups

[Farhan Hasnat]

I am trying to do t-test where I have three groups (AIC, GIC, Others) - accordingly I created a variable (Test1) which is equal to 1 if AIC and 2 if GIC and 3 if Others. However, when I run t-test I can see it only takes two groups.

Is there is any way I can do three groups by Ticker and by Year ?

Any advice would be highly helpful

 ttest DA_w, by(Test1) unequal welch

Two-sample t test with unequal variances
------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. err.   Std. dev.   [95% conf. interval]
---------+--------------------------------------------------------------------
       1 |   1,105    .0142281    .0045872    .1524843    .0052275    .0232286
       2 |   2,515    -.006903     .003283    .1646403   -.0133406   -.0004654
---------+--------------------------------------------------------------------
Combined |   3,620   -.0004528    .0026809       .1613    -.005709    .0048035
---------+--------------------------------------------------------------------
    diff |             .021131    .0056409                .0100691    .0321929
------------------------------------------------------------------------------
    diff = mean(1) - mean(2)                                      t =   3.7460
H0: diff = 0                             Welch's degrees of freedom =  2265.65

    Ha: diff < 0                 Ha: diff != 0                 Ha: diff > 0
 Pr(T < t) = 0.9999         Pr(|T| > |t|) = 0.0002          Pr(T > t) = 0.0001

* Example generated by -dataex-. For more info, type help dataex
clear
input str9 Ticker int Year byte(AIC GIC Others) float(Test1 DA_w)
"TH:2S"    2003 . 1 . 2           .
"TH:2S"    2004 . 1 . 2           .
"TH:2S"    2005 . 1 . 2           .
"TH:2S"    2006 . 1 . 2           .
"TH:2S"    2007 . 1 . 2  -.08454061
"TH:2S"    2008 . 1 . 2  -.25864583
"TH:2S"    2009 . 1 . 2    .6085715
"TH:2S"    2010 . 1 . 2   .09650405
"TH:2S"    2011 . 1 . 2   -.1017216
"TH:2S"    2012 . 1 . 2   .11378127
"TH:2S"    2013 . 1 . 2  -.05965737
"TH:2S"    2014 . 1 . 2  -.04307863
"TH:A"     2003 1 . . 1           .
"TH:A"     2004 1 . . 1    .6278349
"TH:A"     2005 1 . . 1   .06726224
"TH:A"     2006 1 . . 1   -.1024842
"TH:A"     2007 1 . . 1   .09083974
"TH:A"     2008 1 . . 1  -.02267651
"TH:A"     2009 1 . . 1  -.15725097
"TH:A"     2010 1 . . 1    .0957468
"TH:A"     2011 1 . . 1  .032091457
"TH:A"     2012 1 . . 1    .1184249
"TH:A"     2013 1 . . 1    .2663026
"TH:A"     2014 1 . . 1    .2288631
"TH:ABICO" 2003 . 1 . 2   -.1509448
"TH:ABICO" 2004 . 1 . 2   -.7443731
"TH:ABICO" 2005 . 1 . 2    .6278349
"TH:ABICO" 2006 . 1 . 2 -.013820863
"TH:ABICO" 2007 . 1 . 2  .066848315
"TH:ABICO" 2008 . 1 . 2   -.1630021
"TH:ABICO" 2009 . 1 . 2    .1760944
"TH:ABICO" 2010 . 1 . 2  .027172344
"TH:ABICO" 2011 . 1 . 2    .1821229
"TH:ABICO" 2012 . 1 . 2    .2805742
"TH:ABICO" 2013 . 1 . 2   .04285799
"TH:ABICO" 2014 . 1 . 2   .06678031
"TH:ACC"   2003 . 1 . 2   .13844515
"TH:ACC"   2004 . 1 . 2  -.23043732
"TH:ACC"   2005 . 1 . 2   -.1565992
"TH:ACC"   2006 . 1 . 2  -.09132891
"TH:ACC"   2007 . 1 . 2   .04345912
"TH:ACC"   2008 . 1 . 2   .18008757
"TH:ACC"   2009 . 1 . 2           .
"TH:ACC"   2010 . 1 . 2           .
"TH:ACC"   2011 . 1 . 2     .213304
"TH:ACC"   2012 . 1 . 2   .05535654
"TH:ACC"   2013 . 1 . 2  -.17010236
"TH:ACC"   2014 . 1 . 2           .
"TH:AFC"   2003 . 1 . 2   .04718301
"TH:AFC"   2004 . 1 . 2   .05981214
"TH:AFC"   2005 . 1 . 2   .23442598
"TH:AFC"   2006 . 1 . 2   -.1069364
"TH:AFC"   2007 . 1 . 2  -.04298536
"TH:AFC"   2008 . 1 . 2   .05517821
"TH:AFC"   2009 . 1 . 2 -.006168552
"TH:AFC"   2010 . 1 . 2   .01499469
"TH:AFC"   2011 . 1 . 2  .012949838
"TH:AFC"   2012 . 1 . 2   .04469505
"TH:AFC"   2013 . 1 . 2   .05154819
"TH:AFC"   2014 . 1 . 2           .
"TH:AGE"   2003 . 1 . 2           .
"TH:AGE"   2004 . 1 . 2           .
"TH:AGE"   2005 . 1 . 2           .
"TH:AGE"   2006 . 1 . 2           .
"TH:AGE"   2007 . 1 . 2           .
"TH:AGE"   2008 . 1 . 2   .26367554
"TH:AGE"   2009 . 1 . 2   .08931428
"TH:AGE"   2010 . 1 . 2   .25956905
"TH:AGE"   2011 . 1 . 2  -.09252103
"TH:AGE"   2012 . 1 . 2   .04484579
"TH:AGE"   2013 . 1 . 2   .13395208
"TH:AGE"   2014 . 1 . 2  -.27340496
"TH:AH"    2003 1 . . 1   -.2078416
"TH:AH"    2004 1 . . 1   -.1830832
"TH:AH"    2005 1 . . 1 .0010229541
"TH:AH"    2006 1 . . 1  .016149808
"TH:AH"    2007 1 . . 1   -.0128961
"TH:AH"    2008 1 . . 1  -.04499222
"TH:AH"    2009 1 . . 1   -.0351702
"TH:AH"    2010 1 . . 1  -.04360331
"TH:AH"    2011 1 . . 1 -.005175362
"TH:AH"    2012 1 . . 1  .030353934
"TH:AH"    2013 1 . . 1  -.05105714
"TH:AH"    2014 1 . . 1  -.07181978
"TH:AHC"   2003 . 1 . 2  -.04777425
"TH:AHC"   2004 . 1 . 2 -.019984696
"TH:AHC"   2005 . 1 . 2   .04643552
"TH:AHC"   2006 . 1 . 2 -.003667239
"TH:AHC"   2007 . 1 . 2   .01884266
"TH:AHC"   2008 . 1 . 2  -.01925189
"TH:AHC"   2009 . 1 . 2  .027770244
"TH:AHC"   2010 . 1 . 2 -.013661582
"TH:AHC"   2011 . 1 . 2  -.03151839
"TH:AHC"   2012 . 1 . 2  -.02424605
"TH:AHC"   2013 . 1 . 2   .02805965
"TH:AHC"   2014 . 1 . 2  .018061165
"TH:AI"    2003 . 1 . 2  .072685905
"TH:AI"    2004 . 1 . 2  -.10123187
"TH:AI"    2005 . 1 . 2  .012330643
"TH:AI"    2006 . 1 . 2   .04343369
end

[Clyde Schechter]

A t-test is by its very definition only used for comparison of two groups. The analogous way of contrasting means among three (or more) groups is, classically, analysis of variance (ANOVA), and in more modern style, regression on a categorical variable:

regress DA_w i.Test1

With all due respect, what I have told you here is normally taught in all introductory statistics courses. The fact that you ask that, and that you are also proposing to then do this separately for each ticker and each year (which is not a very good approach to panel data), suggests to me that you are dealing with data complexities that are beyond your current level of statistical training. While it is possible to get step-by-step guidance through a project like this on Statalist, it is an inefficient use of your time, which would be better invested in shoring up your knowledge of fundamental statistics so that you can approach complex data sets like this more readily in the future.

[Nick Cox]

This question is a small variation on one you started on 21August

[Nick Cox]

This is the link alluded to in #3, which cycled over the same broad advice.

https://www.statalist.org/forums/for...h-three-groups