st: merge question

[Kit Baum <baum@bc.edu>]

Amadou has a randomness problem:

 I am trying to merge 2 datasets.
 But everytime, I get different results
 (_m==3 has 83 observations in the
 first time, 97 in the second, 100 in the
 third and 96 in the fourth, and so on).
 I tried to set seed and made my sort, stable.
 With no success. I also tried to recast double
 my merging identifier. No success. I tried to
 tostring it. No success either.
 Any hints why I obtain these various results?
 I verified in both Stata and Excel.
 I do not understand why Stata marked 3 to some
 observations that belonged to both datasets in the
 first trial and not in the second time.
 Best regards.
 Amadou.

 PS: When I work interractivelly, I do not have that problem.
 I have 96 observations that matched. So what I am doing
 wrong in my stata do file?




This is almost surely the result of a many-to-many merge which will  
create exactly what he finds: a do-file that, when rerun, yields  
different results (in terms of the number of obs.) every time.

Use the unique, uniqmaster or uniqusing options on merge, whichever  
is appropriate. The merge key should be unique in one file or the  
other, if not both.

"The dangers of many-to-many merges", p. 58 of the book cited below.


Kit Baum, Boston College Economics
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/