Difficulty working with time variable

[Matthew Breckons]

Hi,

I thought I had cracked this but seemingly not:

After having trouble importing variables containing dates from an SPSS file I used -usespss- which looked to have done the trick and after formatting the variables using:

format time1 %tcHH:MM

so times are displayed (11.05, 11.15 etc) and ultimately I was looking for a duration of an appointment which I was able to calculate as time2 - time 1 which gave a third var time3 which I formatted as %tcMM so it shows a time in minutes, in this case 10 minutes.

My problem is that I want this value to behave as a 'normal' number, so I want to multiply var3, as '10' by another variable but I get really strange numbers. I've been reading the 'datetime translation' help files but can't work what I need to do. Is there a way to use/ convert this variable?

Any thoughts much appreciated.

Matt

[Justin Niakamal]

You'll have to post an example of your data and what you've tried using dataex. See post #3 here as an example:

https://www.statalist.org/forums/for...e-in-two-times

[Nick Cox]

I agree with Justin Blasongame that we really need to see a data example.

Either way, clock times in Stata have units of milliseconds.The fact that you times displayed as hours and minutes does not affect. Your data are not thereby converted to units of minutes (or hours, for that matter).

If your times are just hours and minutes, I would myself be tempted just to use two versions, a string version and a numeric version in minutes. I would not use Stata datetimes without extra reason.

clear
input str5 (start end) 
"11:22" "22:33"
"08:00" "09:15" 
end 

gen stime = 60 * real(substr(start, 1, 2)) + real(substr(start, 4, 2)) 
gen etime = 60 * real(substr(end, 1, 2)) + real(substr(end, 4, 2)) 

gen duration = etime - stime

list 

     +------------------------------------------+
     | start     end   stime   etime   duration |
     |------------------------------------------|
  1. | 11:22   22:33     682    1353        671 |
  2. | 08:00   09:15     480     555         75 |
     +------------------------------------------+

An extra reason, however, might well be periods spanning two or more days.

[Matthew Breckons]

Thank you both for your responses. I was about to post an example though in the meantime a colleague directed me to:

https://www.stata.com/statalist/arch.../msg01148.html

where one of the posters describes:
gen minute = mm(time) which neatly did what I was trying to do.

[Nick Cox]

In #3 sorry about a mangled sentence

The fact that you times displayed as hours and minutes does not affect

which should have been

The fact that you have times displayed as hours and minutes does not affect that principle.