[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: Years to dates
It is important to understand that
Stata is not being difficult here,
as there are two basic requirements
that have to be met.
First, there must be scope for daily dates
to be represented by numeric variables.
(You _could_ use a string variable, but
you couldn't do much with it.)
That requires some convention about origin
(what is date 0), and most software writers seem
agreed that some relatively recent origin
is most practical. That though is bound to
seem arbitrary. I see that today's date is
to Stata 16699, which looks a fun number,
but otherwise doesn't have any meaning for me.
Second, there must be scope for different
formats. So, for example, you use examples
1/1/1998 and 12/31/1998. I guess the first
is fairly clear to most Statalist members
but I have to work hard at 12/31/1998 because
it is more common locally to use dates
like 31 December 1998 and astronomers
and some others would write 1998 December 31,
which is the only really logical convention!
And with dates like 5/6/1998 one has to know
which convention is being used. So, naturally,
Stata provides a wide range of formats for
you to choose, so no grumbling!
> Thanks, Sebastian. I got it to work using your method and
> then formating
> the new variable:
> format variable1 %dN/D/Y
> format variable2 %dN/D/Y
> so that it would make sense to the human reader.
> Terra Curtis asked for an easy way to create a certain date
> within a known year:
> I think of one way, eventhough I am not very familiar with
> time-series data,
> I guess it should still work:
> (let the variable "year" be the container of your years than
> it should work
> the following way:)
> gen variable1=mdy(1,1,year)
> gen variable2=mdy(12,31,year)
* For searches and help try: