Time trends on pooled cross section data

[Abhishek Maheshwari]

How do I plot income over time by gender, by race, by education level, for a pooled cross section data?

[Fei Wang]

Abhishek, if you are going to plot average income over time by a group variable, you may generate the aggregate data for plotting using -collapse-.

collapse income, by(year gender)

Then you may plot with the aggregate data.

twoway line income year if gender == 1 || line income year if gender == 2

Above are example codes as I don't know about your actual data.

[Abhishek Maheshwari]

Thank you so much Fei Wang.
But then this will collapse the entire data, right? Is there no other way to do it? because once I collapse the data on mean values, I can't run regressions on individual data, right?

[Fei Wang]

Thank you so much Fei Wang.
But then this will collapse the entire data, right? Is there no other way to do it? because once I collapse the data on mean values, I can't run regressions on individual data, right?

Abhishek, there are several ways of handling this issue. For example, you may use -preserve- and -restore- to return to the original individual dataset after plotting with the aggregate data.

preserve
collapse income, by(year gender)
twoway line income year if gender == 1 || line income year if gender == 2
graph save income_by_gender, replace
restore

* Now you have returned to the original individual data and may collapse another time or do other analyses.

preserve
collapse income, by(year race)
twoway line income year if race == 1 || line income year if race == 2 || line income year if race == 3  ...
graph save income_by_race, replace
restore

[Abhishek Maheshwari]

This is extremely helpful! Thank you so much!
Let me try it...

[Abhishek Maheshwari]

Professor Wang, Could you also tell how to plot income by percentile? I found the following code from an article ( https://www.theanalysisfactor.com/cr...bserve-trends/ )

the code is:

gen ptl=0 // variable for percentile

preserve
forvalues x=10(10)90{
collapse (p`x’) wage1985-wage2005 (mean)ptl
replace ptl=`x’
save wage`x’,replace
restore,preserve
}
use wage10,clear
forvalues x=20(10)90{
append using wage`x’
save wage_ptl,replace
}

use wage_ptl,clear
order ptl, first // moving the variable “ptl” to the top of the list

....(after that they use the xtset command etc.)


Now, when I run the second part of it, all the wage values become zero (probably because of the CLEAR command)
Please suggest.

Thanks a lot.