Dear community,
I have date strings exported from surveymonkey in the following format:
All I am trying to achieve is to transform this into a stata compatible format, i.e. %tc.
Here's my strategy and the description of the problems I encounter:
1. I am extracting the months, days, years, hours, minutes, and seconds via
So far, this works pretty well, as documented in this example (first observation):
2. I am trying to bring all of this together via the the complete date function, and the following happens (again example first observation):
As you can see, I am getting the correct date but incorrect minutes/seconds. More precisely, the resulted time is 57 seconds later than the original time. However, this is not a constant. For example, in the 1000th observation, the resulting time is 38 seconds before the original time. Therefore, I do not believe this is somehow related to the leap seconds, because it should be more or less constant across all observations (time span is less than 2 months).
3. I am suspecting that the problem is related to the day being not extracted accurately for some reason. If I extract "MDY" directly via the date() function and then transform the result into milliseconds via cofd(), I am not getting the exact day:
I am hoping to have provided enought details for you to understand my problem but I'd be happy to provide more information if necessary.
Thanks in advance,
Jakob
I have date strings exported from surveymonkey in the following format:
Code:
. list date1 in 1
+---------------------+
| date1 |
|---------------------|
1. | 10/11/2015 13:04:11 |
Here's my strategy and the description of the problems I encounter:
1. I am extracting the months, days, years, hours, minutes, and seconds via
Code:
gen month = substr(date1,1,2) gen day = substr(date1,4,2) gen year = substr(date1,7,4) gen hour = substr(date1,12,2) gen minute = substr(date1,15,2) gen second = substr(date1,18,2)
Code:
. list date1 month-second in 1
+-------------------------------------------------------------------+
date1 month day year hour minute second
-------------------------------------------------------------------
1. 10/11/2015 13:04:11 10 11 2015 13 4 11
+-------------------------------------------------------------------+
Code:
. gen datefull = mdyhms(month, day, year, hour, minute, second)
. format datefull %tc
. list date1 datefull in 1
+------------------------------------------+
| date1 datefull |
|------------------------------------------|
1. | 10/11/2015 13:04:11 11oct2015 13:05:08 |
+------------------------------------------+
3. I am suspecting that the problem is related to the day being not extracted accurately for some reason. If I extract "MDY" directly via the date() function and then transform the result into milliseconds via cofd(), I am not getting the exact day:
Code:
. ** a) transform date1 into days since 01/01/1960:
. gen d1 = date(date1, "MDYhms")
. ** b) transform days into milliseconds
. gen d2 = cofd(d1)
. format d2 %tc
. list d2 in 1, notrim
d2
1. 11oct2015 00:00:53
Thanks in advance,
Jakob