Re: st: elapsed time between incidents

[Nick Cox <njcoxstata@gmail.com>]

Here's another way to do it:

gen notincident = missing(incident)
bysort id (notincident day) : gen elapsed2 = day - day[_n-1] if !notincident

Nick

On Tue, Feb 19, 2013 at 11:19 PM, Nick Cox <njcoxstata@gmail.com> wrote:
> I don't think anyone can be expected to choose the same words as me,
> so searching for this inevitably seems difficult.
>
> Nick
>
> On Tue, Feb 19, 2013 at 9:34 PM, Steven Archambault
> <archstevej@gmail.com> wrote:
>> Okay, thanks for this. I apologize for not finding the tip prior to
>> posting. My chosen search words were off!
>>
>> On Tue, Feb 19, 2013 at 2:08 PM, Nick Cox <njcoxstata@gmail.com> wrote:
>>> This kind of problem was discussed in my posts
>>>
>>> http://www.stata.com/statalist/archive/2011-01/msg00999.html
>>>
>>> http://www.stata.com/statalist/archive/2012-08/msg00633.html
>>>
>>> and in a Tip
>>>
>>> Cox, N.J. 2011. Stata tip 101: Previous but different. Stata Journal
>>> 11(3): 472-473.
>>>
>>> No doubt there are other references.
>>>
>>> Here's some code:
>>>
>>> . clear
>>>
>>> . input id day incident
>>>
>>>             id        day   incident
>>>   1.  1  1      6
>>>   2.  1   2     .
>>>   3.  1   3     .
>>>   4.  1  4      .
>>>   5.  1   5     8
>>>   6.  1   6     .
>>>   7.  1   7     5
>>>   8. end
>>>
>>> . bysort id (day): gen prev = day[_n-1] if incident[_n-1] < .
>>> (5 missing values generated)
>>>
>>> . by id: replace prev = prev[_n-1] if missing(prev)
>>> (4 real changes made)
>>>
>>> . gen elapsed = day - prev if incident < .
>>> (5 missing values generated)
>>>
>>> . l, sep(0)
>>>
>>>      +--------------------------------------+
>>>      | id   day   incident   prev   elapsed |
>>>      |--------------------------------------|
>>>   1. |  1     1          6      .         . |
>>>   2. |  1     2          .      1         . |
>>>   3. |  1     3          .      1         . |
>>>   4. |  1     4          .      1         . |
>>>   5. |  1     5          8      1         4 |
>>>   6. |  1     6          .      5         . |
>>>   7. |  1     7          5      5         2 |
>>>      +--------------------------------------+
>>>
>>> The nub of the matter is that the previous event can be picked up in
>>> the following observation for the same identifier. After that we just
>>> copy down as required in a cascade.
>>>
>>> FAQ     . . . . . . . . . . . . . . . . . . . . . . . Replacing missing values
>>>         . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
>>>         2/03    How can I replace missing values with previous or
>>>                 following nonmissing values?
>>>                 http://www.stata.com/support/faqs/data/missing.html
>>>
>>> -by:- makes doing this by panels no more difficult than doing it if
>>> there were no panels.
>>>
>>> Steve said non-zero, but his example implies non-missing; whatever the
>>> exact condition is can no doubt be substituted easily.
>>>
>>> Nick
>>>
>>> On Tue, Feb 19, 2013 at 8:45 PM, Steven Archambault
>>> <archstevej@gmail.com> wrote:
>>>
>>>> I have unbalanced panel data where an incident (measured by a non-zero
>>>> value) occurs randomly in the panel. I want to calculate the time
>>>> (days_elapsed) between incidents. Here is an example:
>>>>
>>>> id day incident days_elapsed
>>>> 1  1      6             .
>>>> 1   2     .              .
>>>> 1   3     .              .
>>>> 1  4      .              .
>>>> 1   5     8             4
>>>> 1   6     .              .
>>>> 1   7     5             2
>>>>
>>>> Thanks for any help here.  I have not been able to find this.
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/