t-test over time

[Sang-Bum Park]

Hi

I am looking for a way that can make it easy to provide the results of a series of t-tests. I want to see whether two sample groups are statistically different in their means by year (2010-2020).
Logically, I thought I can do this by running t-tests ten times by separating each sample by year. Using loops such as the foreach command is also one way to make it handy.
Is there any other way to do this? I wonder if there is already a command in Stata for this.

[Øyvind Snilsberg]

you can use regress to test all differences at once,

use https://www.stata-press.com/data/r17/nlswork.dta, clear
ttest ln_wage if year==70, by(collgrad)
reg ln_wage i.collgrad#i.year i.year 
di _b[1.collgrad#70.year]

[William Lisowski]

More generally, the statsby command provides a framework for running the same command repeatedly on different subgroups and accumulating the results..

help statsby

[Sang-Bum Park]

Øyvind Snilsberg Thank you. It works well for me.
William Lisowski Thanks although I have not yet understood how statsby can be used for my case.

[William Lisowski]

use https://www.stata-press.com/data/r17/nlswork.dta, clear
keep if year<=72
ttest ln_wage if year==70, by(collgrad)
statsby t=r(t) p=r(p), by(year) clear: ttest ln_wage, by(collgrad)
format %9.4f t p
list, noobs

. ttest ln_wage if year==70, by(collgrad)

Two-sample t test with equal variances
------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. err.   Std. dev.   [95% conf. interval]
---------+--------------------------------------------------------------------
       0 |   1,557    1.477278     .009687    .3822379    1.458277    1.496279
       1 |     129    1.866361    .0366441     .416197    1.793854    1.938867
---------+--------------------------------------------------------------------
Combined |   1,686    1.507048    .0097045    .3984743    1.488014    1.526082
---------+--------------------------------------------------------------------
    diff |           -.3890822    .0352667               -.4582534    -.319911
------------------------------------------------------------------------------
    diff = mean(0) - mean(1)                                      t = -11.0326
H0: diff = 0                                     Degrees of freedom =     1684

    Ha: diff < 0                 Ha: diff != 0                 Ha: diff > 0
 Pr(T < t) = 0.0000         Pr(|T| > |t|) = 0.0000          Pr(T > t) = 1.0000

. statsby t=r(t) p=r(p), by(year) clear: ttest ln_wage, by(collgrad)
(running ttest on estimation sample)

      Command: ttest ln_wage, by(collgrad)
            t: r(t)
            p: r(p)
           By: year

Statsby groups
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
.....

. format %9.4f t p

. list, noobs

  +--------------------------+
  | year          t        p |
  |--------------------------|
  |   68   -11.6224   0.0000 |
  |   69   -11.7128   0.0000 |
  |   70   -11.0326   0.0000 |
  |   71   -13.2678   0.0000 |
  |   72   -11.8362   0.0000 |
  +--------------------------+

[Sang-Bum Park]

William Lisowski Thank you very much for your help.