RE: st: Tracking & documenting Change

[Rituparna Basu <basur@pamfri.org>]

Ok, I see what you are saying. The reason I say 'changed from A to B' because the person was 'A' in 04 and then changed to 'B' in 05 but stayed 'B' in 06 (if you see the data). Exact dates would have been more precise but to keep it simpler I am tracking yearly. 

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
Sent: Thursday, April 05, 2012 11:52 AM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Tracking & documenting Change

You must, or least should, have a variable that tells you the date more exactly.

Why say

1        05      B             Changed A-B
1        05       B           Changed A-B
1        05       B           Changed A-B

and not

1        05      B            Changed A-B
1        05       B           Stayed B
1        05       B           Stayed B

Naturally, the results for transitions will depend on what you define as a possible transition. How could it be otherwise?

On Thu, Apr 5, 2012 at 7:30 PM, Rituparna Basu <basur@pamfri.org> wrote:
> Thank you Nick!
> Here is the rule:  The dataset is about healthcare service records....so there are duplicates bc one can have more than one doctor's visit in a given year. Also, Insurance is numeric. Do you think the code will be different bc of the duplicates?
> I will let you know if the code worked.
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
> Sent: Thursday, April 05, 2012 11:21 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: Re: st: Tracking & documenting Change
>
> There seems here an indication that -Ins- changes Year to Year, but in that case why the duplicates? Note that the duplicates get in the way of counting transitions.
>
> But I have to take your example literally because you really don't explain the precise rules.
>
> Also, I assume -Ins- is string because you don't specify and it makes the code easier. But if not, then use -decode- first and use the result of that, not the numeric variable with labels.
>
> gen Status = ""
> sort ID Year
> by ID: replace Status = "Stayed" if Year > Year[_n-1] & Ins == 
> Ins[_n-1] by ID: replace Status = "Changed" if Year > Year[_n-1] & Ins 
> != Ins[_n-1] by ID: replace Status = Status + " " +  Ins  + " " +  
> Ins[_n-1] if
> !missing(Status)
> by ID: replace Status = Status[_n-1] if missing(Status)
>
> Code not tested.
>
> Nick
>
> On Thu, Apr 5, 2012 at 7:05 PM, Rituparna Basu <basur@pamfri.org> wrote:
>
>> I would like to track the change of insurance status of a person from year to year. After tracking, I would like to document them as  %'stayed-same' , %'changed from A-B', etc.
>> Here is the sample data:
>> ID    Year  Insurance Status
>> 1        04       A
>> 1        04       A
>> 1        04       A
>> 1        04       A
>> 1        05      B             Changed A-B
>> 1        05       B           Changed A-B
>> 1        05       B           Changed A-B
>> 1        06      B            Stayed B
>> 1        06       B            Stayed B
>> 1        06       B            Stayed B
>> 2        04       A
>> 2        04       A
>> 2        04       A
>> 2        05       A         Stayed A
>> 2        05       A        Stayed A
>> 2       05       A         Stayed A
>> 2        05       A        Stayed A

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/