Creating a new indicator variable

[Lydia Palumbo]

Dear Statalisters,

I am dealing with the following data, where pidnew denotes the identification variable of the respondent, wave the wave of the survey, mach the change in marital status and marital the marital status.

* Example generated by -dataex-. To install: ssc install dataex
clear
input long pidnew float(wave mach) byte marital
100022  1  . 1
100044  1  . 1
100044  2 11 1
100045  1  . 1
100045  2 11 1
100045  3 11 1
100078  1  . 4
100078  2 44 4
100078  3 44 4
100078  8 44 4
100078  9 44 4
100078 10 44 4
100078 11 44 4
100168  1  . 2
100168  2 22 2
100168  3 22 2
100168  4 22 2
100168  7 25 5
100168  8 55 5
100168  9 55 5
100168 10 55 5
100168 12 55 5
100168 14 54 4
end
label values pidnew pid
label values mach mach
label def mach 11 "single->single", modify
label def mach 13 "single->cohabiter", modify
label def mach 22 "married->married", modify
label def mach 25 "married ->divorced", modify
label def mach 32 "cohabitor->married", modify
label def mach 33 "cohabitor->cohabiter", modify
label def mach 44 "widowed->widowed", modify
label def mach 53 "divorced->cohabiter", modify
label def mach 54 "divorced->widowed", modify
label def mach 55 "divorced->divorced", modify
label values marital marital
label def marital 1 "Never married", modify
label def marital 2 "Married", modify
label def marital 3 "Living as a couple", modify
label def marital 4 "Widowed", modify
label def marital 5 "Split", modify

I would like to create a new ID in a new column (pid2) which modifies every time there is a change in the marital status, such that data become this way:

* Example generated by -dataex-. To install: ssc install dataex
clear
input long pidnew float(wave mach) byte marital pid2
100022  1  . 11
100044  1  . 12
100044  2 11 12
100045  1  . 13
100045  2 11 13
100045  3 11 13
100078  1  . 43
100078  2 44 43
100078  3 44 43
100078  8 44 43
100078  9 44 43
100078 10 44 43
100078 11 44 43
100168  1  . 24
100168  2 22 24
100168  3 22 24
100168  4 22 24
100168  7 25 54
100168  8 55 55
100168  9 55 55
100168 10 55 55
100168 12 55 55
100168 14 54 45
end
label values pidnew pid
label values mach mach
label def mach 11 "single->single", modify
label def mach 13 "single->cohabiter", modify
label def mach 22 "married->married", modify
label def mach 25 "married ->divorced", modify
label def mach 32 "cohabitor->married", modify
label def mach 33 "cohabitor->cohabiter", modify
label def mach 44 "widowed->widowed", modify
label def mach 53 "divorced->cohabiter", modify
label def mach 54 "divorced->widowed", modify
label def mach 55 "divorced->divorced", modify
label values marital marital
label def marital 1 "Never married", modify
label def marital 2 "Married", modify
label def marital 3 "Living as a couple", modify
label def marital 4 "Widowed", modify
label def marital 5 "Split", modify

I tried several times by constructing some indicators (
xtset pidnew wave
by pidnew: egen transitions=max(mach==25|mach==54), but I am getting stuck.
If you could help me with this issue, this would be very helpful.

Thank you very much,
Lydia

[Phil Bromiley]

Thank you for providing data and code. You should try to simplify your posting to the minimum necessary to demonstrate the problem - I'm not sure we need all the labels etc.

You may be able to do this with egen and group where you group by pid and mach. But I'm not sure I understand the logic for your codes in red. Why not start the second number at 1 for each pid? You can almost always do by statements with two variables so it may not be necessary to put the two together.