Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Loop year variable


From   Nick Cox <njcoxstata@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Loop year variable
Date   Mon, 22 Oct 2012 10:23:38 +0100

You did not do what I suggested.

I suggested

bysort iso_o iso_d (year) :
by iso_o iso_d :

but your code is

bysort iso_o iso_d year :
by iso_o iso_d year :

That's quite wrong for your problem. Study the documentation for -by:-
if you don't see why.

By the way, the code might be slightly better as

bysort iso_o iso_d (year FT_A) :
by iso_o iso_d :

Nick

On Mon, Oct 22, 2012 at 10:05 AM, emanuele mazzini
<madsoenistata@gmail.com> wrote:
> Many thanks Nick. That worked well in the subsample I showed you.
> However, in my general sample I do not get as many as changes as I
> thought.
>
> I typed the following:
>
> bysort iso_o iso_d year: gen date_won = year if FT_A > FT_B | FT_A<FT_B
> by iso_o iso_d year: replace date_won = date_won[_n-1] if missing(date_w)
>
> that seems correct to me, but I only obtain 14 real changes, that are
> a way too low. That seems actually very strange to me, why I can't get
> the replace? May it be due to the structure of the dataset? I mean,
> even though it is not a panel in the proper sense, because I can have
> more than 1 observation per year, that should not be a matter of
> concern, because the date_won variable always takes the same value,
> doesn't it?
>
> Many thanks in advance for your help,
>
> Regards,
>
> Emanuele.
>
>
>
> 2012/10/22 Nick Cox <njcoxstata@gmail.com>:
>> See for technique here
>>
>> 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
>>
>> bysort iso_o iso_d (year) : gen datewon = year if !missing(FT_A, FT_B)
>> & FT_A > FT_B
>> by iso_o iso_d : replace datewon = datewon[_n-1] if missing(datewon)
>> gen yrs_won = date - datewon
>>
>> Nick
>>
>> On Sun, Oct 21, 2012 at 10:37 PM, emanuele mazzini
>> <madsoenistata@gmail.com> wrote:
>>> Ok, let's ignore this additional issue for the moment. I can be
>>> clearer I guess with the following example:
>>>
>>>  +---------------------------------------------------------------------+
>>>      | year   iso_o   iso_d   team_A       team_B   FT_A   FT_B   date_won |
>>>      |---------------------------------------------------------------------|
>>>   1. | 1982     ITA     DEU    Italy   Germany FR      3      1       1982 |
>>>   2. | 1983     ITA     DEU                            .      .          . |
>>>   3. | 1984     ITA     DEU                            .      .          . |
>>>   4. | 1985     ITA     DEU                            .      .          . |
>>>   5. | 1986     ITA     DEU                            .      .          . |
>>>      |---------------------------------------------------------------------|
>>>   6. | 1987     ITA     DEU                            .      .          . |
>>>   7. | 1988     ITA     DEU                            .      .          . |
>>>   8. | 1989     ITA     DEU                            .      .          . |
>>>   9. | 1990     ITA     DEU                            .      .          . |
>>>  10. | 1991     ITA     DEU                            .      .          . |
>>>      |---------------------------------------------------------------------|
>>>  11. | 1992     ITA     DEU                            .      .          . |
>>>  12. | 1993     ITA     DEU                            .      .          . |
>>>  13. | 1994     ITA     DEU                            .      .          . |
>>>  14. | 1995     ITA     DEU                            .      .          . |
>>>      +---------------------------------------------------------------------+
>>>
>>> As you can see, the variable date_won is =1982 only for the year in
>>> which the match was played. Now, I would like to generate a new
>>> variable that would tell me how many years have gone by since Italy's
>>> last won (e.g, =0 for 1982, =1 for 1983, =2 for 1984, etc.).
>>> As I said, I tried with: gen yrs_won=year-date_won but this is the result:
>>>
>>>   +---------+
>>>      | yrs_won |
>>>      |---------|
>>>   1. |       0 |
>>>   2. |       . |
>>>   3. |       . |
>>>   4. |       . |
>>>   5. |       . |
>>>      |---------|
>>>   6. |       . |
>>>   7. |       . |
>>>   8. |       . |
>>>   9. |       . |
>>>  10. |       . |
>>>      |---------|
>>>  11. |       . |
>>>  12. |       . |
>>>  13. |       . |
>>>  14. |       . |
>>>      +---------+
>>>
>>> As you can see, such command doesn't produce the expected result. That
>>> explains why I was looking for a way to expand the value of date_won
>>> for all cells. I thought that was the reason for the command not to
>>> work. Do you have any suggestion for overcoming my problem?
>>>
>>> Thank you in advance,
>>>
>>> Regards,
>>>
>>> E.M.
>>>
>>>
>>>
>>>
>>> 2012/10/21 Nick Cox <njcoxstata@gmail.com>:
>>>> I gathered from your original posting that you had yearly data. I
>>>> understand that matches may be more frequent than yearly, but I can't
>>>> advise further without seeing what your data look like as the trade
>>>> data and match data are not obviously compatible.
>>>>
>>>> Nick
>>>>
>>>> On Sun, Oct 21, 2012 at 1:00 PM, emanuele mazzini
>>>> <madsoenistata@gmail.com> wrote:
>>>>> Dear Mr Cox,
>>>>>
>>>>> thank you for your reply. The point, however, is that I cannot declare
>>>>> my dataset to be time-series (how would like but I cannot) due to the
>>>>> fact that some matches may take place once, twice or even more times
>>>>> in the same year. That is why I was looking form a manual command. I
>>>>> may better handle this first? And how can I do that in your opinion?
>>
>>
>> 2012/10/21 Nick Cox <njcoxstata@gmail.com>:
>>>>>> This should yield to a treatment in terms of spells in the sense of
>>>>>> -tsspell- (SSC) and of
>>>>>>
>>>>>> SJ-7-2  dm0029  . . . . . . . . . . . . . . Speaking Stata: Identifying spells
>>>>>>         . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J. Cox
>>>>>>         Q2/07   SJ 7(2):249--265                                 (no commands)
>>>>>>         shows how to handle spells with complete control over
>>>>>>         spell specification
>>>>>>
>>>>>> Look at -tsspell- first; it is self-contained. That is, your panels
>>>>>> are defined by country pairs (dyads) and the start of a spell is
>>>>>> defined by a match taking place (or a particular result; I don't
>>>>>> follow exactly what you want, e.g what about draws?). Then a variable
>>>>>> defined by -tsspell- gives you sequence in spell. You may need to
>>>>>> subtract 1.
>>>>>>
>>>>>> I don't see that loops are needed.
>>>>>>
>>>>>> Nick
>>>>>>
>>>>>> On Sun, Oct 21, 2012 at 12:07 PM, emanuele mazzini
>>>>>> <madsoenistata@gmail.com> wrote:
>>>>>>
>>>>>>> I am not a new Stata user, but I have to admit I still encounter some
>>>>>>> problems with writing loops. A loop may not be the solution to the
>>>>>>> issue I am going to describe, but I thought it is tough.
>>>>>>>
>>>>>>> I do have a dataset where I have country-pair trade observations for
>>>>>>> years 1950-2006 as well as another information, which is football
>>>>>>> match result for some dyads. I therefore need to generate a variable
>>>>>>> that indicates how many years are passed since the last won/lost of a
>>>>>>> country versus another. Suppose FT_A and FT_B are the final results
>>>>>>> for team A and B. I thought to generate a variable as follows:
>>>>>>>
>>>>>>> gen int date_won=year if FT_A>FT_B | FT_A<FT_B
>>>>>>>
>>>>>>> in order to get a variable that is = year for every observation I have
>>>>>>> information about, which reports the year of the won/loss. I then
>>>>>>> thought to generate another variable (let's say yrs_won) to tell me
>>>>>>> how many years have gone by as follows:
>>>>>>>
>>>>>>> gen int yrs_won=year-date_won
>>>>>>>
>>>>>>> Unfortunately, I do not get what I want: the only value yrs_won
>>>>>>> assumes is 0, but for all years from 1983 until 2006 I would like it
>>>>>>> to be 1, 2,3 and so on.
>>>>>>>
>>>>>>> At the beginning I didn't realize the problem, now it may be occur due
>>>>>>> to the fact that date_won = year only for the year in which the match
>>>>>>> was played. I guess I need to do something such as writing a loop to
>>>>>>> let it spread over all cells of such variable, but I don't really know
>>>>>>> how to make it. Does anyone of you has any suggestion?
>> *
>> *   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/
> *
> *   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/
*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index