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 on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: Different results for 1:1-merging using the same variables (int & string)


From   Nick Cox <njcoxstata@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Different results for 1:1-merging using the same variables (int & string)
Date   Thu, 14 Feb 2013 10:09:17 +0000

A side detail but

gen newstrvar = strvar + string(numvar)

is a simpler way to do that in this situation where -numvar- contains
4-digit integers. Calling up -tostring- and creating a new variable
can be avoided.

More to the point, did you look closely at some or all of the
observations where you get a different result?

Nick

On Thu, Feb 14, 2013 at 8:17 AM, Hofbaur, Ulrich <Ulrich.Hofbaur@whu.edu> wrote:

> Jeff, thanks for your suggestions! "next_year" is an integer converted to string by using the tostring-command. So, I simply add two strings. I created the variables in both files and exactly the same way. Just validated that.

Jeph Herrin

> I do not understand how you can calculate length(next_year) if "next_year" is an integer.
>
> Do you create the variables in both files, or just the -master- file? If just the -master- check that the -using- file variables have been constructed in the same way.

On 2/13/2013 1:07 PM, Hofbaur, Ulrich wrote

>> I have an issue with conducting a 1:1-merge in Stata. The merge is
> based on two variables. The 1. variable (string) consists of exactly 6 digits. The 2. variable  (integer) consists of exactly 4 digits (no variation in the length of digits in either of the two variables). I tried two versions, and the they both yielded different results. Please, further note that I use the same file to merge and the variables differ
>>
>> Option 1: Defining a 10-digit string variable. Therefore, convert "var
> 2" to string and then sum var1 and var2. Hence, I obtain" var3" (which is a 10-digit string; again no variation w.r.t to the length of
> 10-digits) and merge (1:1) on "var3". → Results in 15,839 matches
>> Option 2:  Merge (1:1) on var1 and var2  as separate variables →
> Results in 14,227
>>
>> Does anybody know where this difference comes from. My gut feeling
> tells me that Option 2 is the more reliable one. However, I lack evidence on that. The abbreviated Do-File is attached.
>>
>> Thank you very much for your support!
>>
>> Best,
>> Ulrich
>>
>> ******* Do File **************
>>
>> use F:\001_Forschung\Daten\Cash&Acquisitions\file_A_prelim.dta, clear
>>
>>   * Option 1
>>       gen acquirorcusip_year=cusip_6dgt+next_year //corresponds to var 3 in
> the above description
>>       gen length_cusip_6dgt=length(cusip_6dgt)
>>       gen length_announcement_year=length(next_year)
>>       gen length_acquirorcusip_year=length(acquirorcusip_year)
>>       sum length_cusip_6dgt length_announcement_year
> length_acquirorcusip_year
>>
>>       Variable        Obs        Mean    Std. Dev.       Min        Max
>>       length_cus~t     196217           6           0          6          6
>>       length_ann~r     196217           4           0          4          4
>>       length_acq~r     196217          10           0         10         10
>>
>> * Option 2
>>        gen announcement_year=next_year // corresponds to var 2 in the above
> description. Rename due to file_B
>>               destring announcement_year, replace
>>        gen acquirorcusip=cusip_6dgt  // corresponds to var 1 in the above
> description.
>>       sort acquirorcusip announcement_year
>>
>> save file_A.dta, replace
>>
>>
>> * Option 1: Merge on the joint string variable
>>       use file_A.dta, clear
>>       merge 1:1 acquirorcusip_year using
> F:\001_Forschung\Daten\Cash&Acquisitions\file_B.dta
>>
>>       Result                           # of obs.
>>       -----------------------------------------
>>       not matched                       191,640
>>       from master                   180,378  (_merge==1)
>>       from using                     11,262  (_merge==2)
>>
>>       matched                            15,839  (_merge==3)
>>       -----------------------------------------
>>
>> * Option 2:  Merge on two separate variables
>>       use file_A.dta, clear
>>       merge 1:1 acquirorcusip announcement_year using
> F:\001_Forschung\Daten\Cash&Acquisitions\file_B.dta
>>
>>       Result                           # of obs.
>>       -----------------------------------------
>>       not matched                       194,864
>>       from master                   181,990  (_merge==1)
>>       from using                     12,874  (_merge==2)
>>
>>       matched                            14,227  (_merge==3)
>>       -----------------------------------------

*
*   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