Computing time span between dates

[Vlad Surdea]

Hello,

I have a dataset that records a series of variables for a large number of observations, including the date at which the observation was made. I want to create a new variable, that measures the distance between the moment of observation and a given date, in this case, 13 May 2019.

What is the easiest way to implement this?

[Rich Goldstein]

if your dates are in Stata internal format (see "help datetime") you can just subtract (though the exact method depends on whether your data are in long or wide form); if your dates are not in the SIF, then you need to get them there; for help with exact code, please post an example using -dataex- (see the FAQ for advice)

[Nick Cox]

A daily date in Stata is the number of days since 1 January 1960, so time spans are obtained by subtraction. Here I used scalar constants, but the principle is the same for a daily date variable and a daily date scalar or for two daily date variables.

. di mdy(1, 1, 1960)
0

. di mdy(5, 16, 2020)
22051

. di mdy(5, 16, 2020) - mdy(5, 1, 2020)
15

See

help datetime

[Vlad Surdea]

Thank you so much!

I also managed to do this in a more convoluted way:

gen ds_day=.
replace ds_day=clock("2015-03-29: 03:00:00", "YMD hms") if year==2015
replace ds_day=clock("2016-03-27: 03:00:00", "YMD hms"), if year==2016
replace ds_day=clock("2016-03-27: 03:00:00", "YMD hms") if year==2016
replace ds_day=clock("2017-03-26: 03:00:00", "YMD hms") if year==2017
replace ds_day=clock("2018-03-25: 03:00:00", "YMD hms") if year==2018
replace ds_day=clock("2019-03-31: 03:00:00", "YMD hms") if year==2019

format ds_day %tc
replace ds_day=dofc(ds_day)
format ds_day %td

gen distance=.
replace distance=date-ds_day

[Nick Cox]

A time of 3 am is gratuitous if it is going to be discarded any way. Your code seems equivalent to this: .

gen ds_day = mdy(3, 29, year) if year == 2015 
replace ds_day = mdy(3, 27, year) if year == 2016
replace ds_day = mdy(3, 26, year) if year == 2017 
replace ds_day = mdy(3, 25, year) if year == 2018
replace ds_day = mdy(3, 31, year) if year == 2019 
format ds_day %td 

gen distance = date - ds_day