How to cumulate the contents of string variables by id?

[Fred Lee]

As far as I know, sum is used for numberic variables, how about the string variables?
Here is the data:
----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. To install: ssc install dataex
clear
input str54 Mediator float tempid
"negAffect5"                                             1
"posAffect4"                                             1
"negAffect2 negAffect3"                                  1
"negAffect2 negAffect5"                                  1
"negAffect3 negAffect4"                                  1
"negAffect3 negAffect5"                                  1
"negAffect4 negAffect5"                                  1
"negAffect1 negAffect3 negAffect5"                       1
"negAffect2 negAffect3 negAffect4"                       1
"negAffect2 negAffect3 negAffect5"                       1
"negAffect2 negAffect4 negAffect5"                       1
"negAffect3 negAffect4 negAffect5"                       1
"negAffect1 negAffect2 negAffect3 negAffect4"            1
"negAffect1 negAffect2 negAffect3 negAffect5"            1
"negAffect1 negAffect3 negAffect4 negAffect5"            1
"negAffect2 negAffect3 negAffect4 negAffect5"            1
"negAffect1 negAffect2 negAffect3 negAffect4 negAffect5" 1
"negAffect5"                                             2
"posAffect4"                                             2
"negAffect2 negAffect3"                                  2
end

[Nick Cox]

This question was addressed in an article last year https://www.stata-journal.com/articl...article=pr0071

For string variables, technique starts with

gen wanted = given[1] in 1 

replace wanted = wanted[_n-1] + given in 2/L

with very often some variation such as

replace wanted = wanted[_n-1] + " " + given in 2/L

replace wanted = wanted[_n-1] + "," + given in 2/L

[Fred Lee]

This question was addressed in an article last year https://www.stata-journal.com/articl...article=pr0071

For string variables, technique starts with

gen wanted = given[1] in 1

replace wanted = wanted[_n-1] + given in 2/L

with very often some variation such as

replace wanted = wanted[_n-1] + " " + given in 2/L

replace wanted = wanted[_n-1] + "," + given in 2/L

However, how to combined with tempid?
It says ""'in' may not be combined with 'by' when using bys tempid.

[Nick Cox]

Under by: you need

...  if _n == 1 

... if _n > 1

[Fred Lee]

Under by: you need

... if _n == 1

... if _n > 1

Thanks Nick! But this is still not correct. In the bellow example, I want to generate a varaible which includes the same contents for each tempid.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str54 Mediator float tempid
"negAffect5"                                             1
"posAffect4"                                             1
"negAffect2 negAffect3"                                  1
"negAffect2 negAffect5"                                  1
"negAffect3 negAffect4"                                  1
"negAffect3 negAffect5"                                  1
"negAffect4 negAffect5"                                  1
"negAffect1 negAffect3 negAffect5"                       1
"negAffect2 negAffect3 negAffect4"                       1
"negAffect2 negAffect3 negAffect5"                       1
"negAffect2 negAffect4 negAffect5"                       1
"negAffect3 negAffect4 negAffect5"                       1
"negAffect1 negAffect2 negAffect3 negAffect4"            1
"negAffect1 negAffect2 negAffect3 negAffect5"            1
"negAffect1 negAffect3 negAffect4 negAffect5"            1
"negAffect2 negAffect3 negAffect4 negAffect5"            1
"negAffect1 negAffect2 negAffect3 negAffect4 negAffect5" 1
"negAffect5"                                             2
"posAffect4"                                             2
"negAffect2 negAffect3"                                  2
"negAffect2 negAffect5"                                  2
"negAffect3 negAffect4"                                  2
"negAffect3 negAffect5"                                  2
"negAffect4 negAffect5"                                  2
"negAffect1 negAffect2 negAffect3"                       2
"negAffect1 negAffect3 negAffect5"                       2
"negAffect2 negAffect3 negAffect4"                       2
"negAffect2 negAffect3 negAffect5"                       2
"negAffect2 negAffect4 negAffect5"                       2
"negAffect3 negAffect4 negAffect5"                       2
"negAffect1 negAffect2 negAffect3 negAffect4"            2
"negAffect1 negAffect2 negAffect3 negAffect5"            2
"negAffect1 negAffect3 negAffect4 negAffect5"            2
"negAffect2 negAffect3 negAffect4 negAffect5"            2
"negAffect1 negAffect2 negAffect3 negAffect4 negAffect5" 2
"negAffect5"                                             3
"posAffect4"                                             3
"negAffect2 negAffect3"                                  3
"negAffect2 negAffect5"                                  3
"negAffect3 negAffect4"                                  3
end

gen tempMediatorTotal = Mediator[1] in 1 if tempid == 1 
replace tempMediatorTotal = tempMediatorTotal[_n-1] + Mediator in 2/L if tempid>1

[Nick Cox]

Sorry, but I don't understand what you want. Please back up and pose a much simpler equivalent problem in which you


* show what you start with in terms of (say) A B C D

* show what you want to get from it

* explain your rules

The question How to cumulate the contents of string variables says nothing about cumulating conditionally on what is inside the strings.

[Fred Lee]

Sorry, but I don't understand what you want. Please back up and pose a much simpler equivalent problem in which you


* show what you start with in terms of (say) A B C D

* show what you want to get from it

* explain your rules

The question How to cumulate the contents of string variables says nothing about cumulating conditionally on what is inside the strings.

Thanks Nick!
I made a post here:
https://www.statalist.org/forums/for...ables-by-group

[Fred Lee]

This question was addressed in an article last year https://www.stata-journal.com/articl...article=pr0071

For string variables, technique starts with

gen wanted = given[1] in 1

replace wanted = wanted[_n-1] + given in 2/L

with very often some variation such as

replace wanted = wanted[_n-1] + " " + given in 2/L

replace wanted = wanted[_n-1] + "," + given in 2/L

Hi Nick, what does "L" mean in your syntax?
Do you have a guide for such systematic macros?

[Nick Cox]

For in L see the help:

help in

No Stata macros are used in #2 -- where this code appears -- so I don't understand the other question.

[Fred Lee]

For in L see the help:

help in

No Stata macros are used in #2 -- where this code appears -- so I don't understand the other question.

I thought L was a macro by mistake, thansk for your information!