comparing variables

[Sunil Sampath]

Hi, can someone can help me with this query ?
I have 12 numerical variables for each subject, 6 variables are before an intervention (pre1, pre2, pre3, pre4, pre5 and pre6) and 6 are after an intervention (post1, post2, post3, post4, post5 and post6). I need to work out for each subject if any 3 of the 6 post intervention variables has decreased by 30% and in the remaining 3/6 variables, if ≤1 variables has increased by 30%.
I have generated a new set of variables which are ± 30% of the pre-variable values to compare with corresponding post-variables. But since it is any 3 out of any 6 variables and then choosing the remaining 3 variables to compare, I am unable to write a script for this. I have tried searching for related previous posts here but couldn’t find any. I am relatively new to stata, any help is greatly appreciated.
Thanks
Sunil

[Clyde Schechter]

There are several ways one could do this. Here's one approach I might use. I assume you have a subject id variable called id.

reshape long pre post, i(id) j(meas_num)
gen rel_diff = (post-pre)/pre
gen byte incr_30 = (rel_diff > 0.3) if !missing(rel_diff)
gen byte decr_30 = (rel_diff < -0.3) if !missing(rel_diff)
by id, sort: gen byte num_incr_30 = total(incr_30)
by id: gen byte num_decr_30 = total(decr_30)
gen byte criterion = num_decr_30 >= 3 & num_incr_30 <= 1

// OPTIONALLY RESTORE WIDE FORMAT TO DATA
reshape wide pre post incr_30 decr_30, i(id) j(meas_num)

Caveat: not tested, but I think it's right.

The variable criterion will indicate the observations fitting your description.

[Nick Cox]

Before I saw Clyde's most helpful reply, my one thought was "you need to reshape". As he has done the hard bit, I will do the easy bit and underline that the kind of comparison you want to do will be much easier with a different structure.

What's magic about 30%?

[Sunil Sampath]

Hi Clyde and Nick
Thank you for taking time to respond and help with this. 30 % is just one of the levels of response to a treatment in rheumatic conditions , hence I was interested .
BW
Sunil