Re: st: merge: "variable does not uniquely identify observations in the master data"

[Phil Schumm <pschumm@uchicago.edu>]

On Jul 8, 2008, at 6:46 PM, Jeff Pflueger wrote:
> If I do this, I get the error I would expect:
>
> use "$people", clear
> merge householdno using "$household", uniqmaster sort
>
> the error is in red on the command line and it terminates my do-file.
> It says: "variable householdno does not uniquely identify  
> observations in the master data"
>
> Makes sense, the master dataset does not contain unique  
> householdno's  but the household dataset does.
>
> But if I do this:
>
> use "$people", clear
> merge householdno using "$household", uniqusing sort
>
> I get the same error: "variable householdno does not uniquely  
> identify observations in the master data" but in this case, the do- 
> file continues to run, and the error message is in green, not red.

In the first case, you're getting an error because you specified the - 
uniqmaster- option and the variable householdno does not uniquely  
identify the observations in $people.  In the second case, you're  
merely getting an informative message; -merge- always gives such a  
message when the observations in either dataset are not uniquely  
identified by the variable(s) you are merging on.  You may ignore  
this message if you are intentionally performing a one-to-many or  
many-to-one merge.



> But householdno is unique in the master dataset....
>
> When I:
> use "$household", clear
> duplicates list householdno
>
> I get this:
>
> Duplicates in terms of householdno
>
> (0 observations are duplicates)

This is entirely consistent with what you have shown us above -- in  
those calls to -merge-, $household is not the master dataset, $people  
is.


-- Phil

*
*   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/