Panel data: Plot line graph for 2 periods before and after state change

[Bianca de Jong]

Hi everyone,

I have a panel data set which looks approximately like this:

id	time	crime	income
1	1	0	2000
1	2	0	1800
1	3	1	1000
1	4	0	1200
1	5	0	1400
2	1	0	1500
2	2	1	1100
2	3	1	1400
2	4	0	1700
2	5	0	2000

For every person, I have data on 100 periods, with a variable saying whether they committed a crime in that period, and what their income was in that period.

I would like to plot a graph where I show for the average income of all offenders on t-2, t-1, t=0 (when crime = 1), t+1 and t+2. So basically, I want to show the average income of offenders two periods before and two periods after committing the crime.

It should look something like this:

where the line is at t=0 (where crime = 1), in this case one can see average income from t-8 until t+8.

I have no idea how to get to such a graph, so it would be great if someone could help me out!

[Nick Cox]

Your example is helpful but immediately raises the question of what you want done if there are crimes at two or more times for each individual. I am going to assume that you want to use the first time.

Here is some technique,

clear 
input id    time    crime    income
1    1    0    2000
1    2    0    1800
1    3    1    1000
1    4    0    1200
1    5    0    1400
2    1    0    1500
2    2    1    1100
2    3    1    1400
2    4    0    1700
2    5    0    2000
end 

egen origin = min(cond(crime == 1, time, .)), by(id)
gen T = time - origin 
egen mean = mean(income), by(T)

The only tricky bit is getting as new time origin the reference data for each individual. For the technique here see https://www.stata-journal.com/articl...article=dm0055 especially Section 9. But there are many ways to do it, as an earlier FAQ underlines: https://www.stata.com/support/faqs/d...ing-last-date/

[Bianca de Jong]

Thank you for your answer, Nick. I am mainly interested in the first offence, so happy to only look at that one for now. I have tried your technique, but still not 100% sure how to get to the plot. If I understand it correctly, origin should give me the time period t in which the first offence was committed, T a difference between current period and that period and mean would give me the mean income for every difference in time period. How would I then in best case get to the plot?

[Nick Cox]

It's just a line plot of mean against T. You showed in #1 that you knew how to get such plots.

[Bianca de Jong]

Great, thank you! I think it worked