Run test by group

[Chinmay Sharma]

Hi,

I have data which resembles the following:

* Example generated by -dataex-. To install: ssc install dataex
clear
input double(group v1)
1 1
1 0
1 1
1 0
1 1
2 1
2 0
2 1
2 1
2 1
end

Here, I have a group ID variable, and a variable of interest, v1. Essentially, I wish to perform the Runs Test (https://www.stata.com/manuals/rttest.pdf) on Stata, to test whether or not the sequence of v1 are serially independent, within group. As far as I could tell from the documentation, I am unable to combine that -runtest- command by group. Ideally, I would simply do something like:

by group: runtest v1

which would be similar to how t-tests for different groups could be done. Unfortunately, the runtest does not seem to have this functionality. Is there an elegant way to do it? The brute-force method would be to do the test one group at a time, and store the results in either a separate dataset (or separate variable), by preserving, dropping all other groups, testing and then restoring.

Any insight on this is much appreciated!

Best,
CS

[Joro Kolev]

Try

. statsby, by(group) clear: runtest v
(running runtest on estimation sample)

      command:  runtest v
            p:  r(p)
            z:  r(z)
          Var:  r(Var)
         mean:  r(mean)
      N_above:  r(N_above)
      N_below:  r(N_below)
       n_runs:  r(n_runs)
            N:  r(N)
           by:  group

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

. list

     +-------------------------------------------------------------+
     | group   p   z   Var   mean   N_above   N_below   n_runs   N |
     |-------------------------------------------------------------|
  1. |     1   .   .     0      1         0         5        1   5 |
  2. |     2   .   .     0      1         0         5        1   5 |
     +-------------------------------------------------------------+

[Chinmay Sharma]

Thanks a lot, Joro Kolev !!