Time series graphs of counts by different variables - which can be easily manipulated and changed

[Joshua Nealon]

Dear all,
I am trying to make a time-series graph showing count of events; by quarter; in groups who received; or did not receive an intervention (coded as a binary variable; this is the first step but there will be others). An example of my data is copied below.

Using some code from Nick Cox from this forum, this is possible by:

egen no_int = count(int) if int==0, by(date_quart)
egen yes_int = count(int) if int==1, by(date_quart)
egen tag = tag(date_quart)
twoway line no_int yes_int date_quart if tag, sort

But this is not really practical for making many exploratory graphs. I would like a solution which I can easily manipulate, something like:

twoway (line int date_quart if int == 0) (line int date_quart if int == 0)

but I cannot see how to do this with counts.

Ideally I would prefer not to collapse the data unless this is really the simplest solution.

Any advice would be much appreciated.

Josh.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str5 Agegroup str6 Sex float(intervention date_quart)
"20-24" "Female" 0 231
"00-04" "Male"   0 231
"00-04" "Female" 0 231
"00-04" "Female" 0 231
"00-04" "Female" 0 231
"70-74" "Male"   0 231
"15-19" "Male"   1 231
"40-44" "Male"   0 231
"85+"   "Female" 0 231
"00-04" "Male"   0 231
"25-29" "Female" 0 231
"20-24" "Male"   0 231
"25-29" "Male"   0 231
"25-29" "Male"   0 231
"75-79" "Female" 0 231
"00-04" "Male"   0 231
"60-64" "Male"   0 231
"00-04" "Female" 0 231
"60-64" "Female" 0 231
"80-84" "Female" 0 231
"25-29" "Male"   0 230
"15-19" "Male"   1 230
"15-19" "Male"   1 230
"20-24" "Female" 0 230
"00-04" "Female" 0 230
"00-04" "Male"   0 230
"75-79" "Male"   0 230
"15-19" "Male"   1 230
"00-04" "Male"   0 230
"25-29" "Male"   0 230
end
format %tq!Qq-CCYY date_quart

[Nick Cox]

I don't get what you want precisely. Are age and sex relevant or not?

histogram date_quart , by(inter, col(1)) freq

does the counting for you and shows the two categories of intervention separately..

[Joshua Nealon]

Thanks Nick.
I am looking for something as you propose but which plots lines (rather than histograms) and on the same graph.

Age and sex aren't yet relevant yet...but they will be in future. It's the reason I didn't want to collapse because I felt that limits flexibility in exploring the data. Probably I should just get better at using "collapse" and switching between wide and long data. Let me try that.

Thanks again.

Josh.

[Nick Cox]

spikeplot date_quart , by(inter, col(1)) recast(line)