Graph for multiple rounds

[Federico Atzori]

Hi everybody,
This is my first post here

Let me give you some information on my data set.
For 15 rounds of my study, I have 15 variables for one treatment and 15 for the other, indicating if the player in that treatment was the player 1-2-3-4.

Now, I want to see if player 1, in my case, both 1 and 3 (because they were matched), are predictive of that player's choice. So far, I get here:

forval i = 1/15 {
  gen P1_h`i' = (hd`i'groupid == 1 | hd`i'groupid == 3)
  gen P1_s`i' = (sh`i'groupid == 1 | sh`i'groupid == 3)
}

Where "male" is the choice variable (0/1).

So, now I have 30 variables indicating if the subject was P1.

Now, I would like to graph it by treatment, and I get here, but I'm stuck in an error:

preserve
collapse (mean) choice_num, by("`P1_h'" "`P1_s'" block male treatment)
twoway (connected choice_num block if "`P1_h'" == 1, sort msymbol(D) lpattern(l) by(treatment)) ///
       (connected choice_num block if "`P1_s'" == 1, sort msymbol(T) lpattern(dash) by(treatment)), ///
       legend() ///
       xscale(range (1 5)) ///
       xlabels(1(1)5) ///
       yscale(range (0 1)) ///
       ylabels(0(0.1)1) ///
       ytitle("")
restore

choice_num is the choice between A and B that the player perform (0/1)
block is --> block = ceil(round / 3)

Any help?

Best