To use -merge-, the two datasets SHOULD be unique and sorted by an index variable. In your stituation, I guess,

index   var1

1           14

2           18

3            23

Index     var2

1              10

1              12

2               2

2              4

....

For the non-unique dataset, you are supposedto  FIRSLTY know WHY YOU ARE MERGING THEM. To generate a unique dataset, you may use -collapse- or -reshape- to calculate the mean, sum, median . . . .. Then merge the two after sorting.

Lastly, one suggestion for those raising questions: describe your question(s) with example rather than just throwing questions out. The more detailed, the more likely people help you out.        

Thank you so much.

My question is: one database is about hospitals in California, with "county_name" which is not unique. I have to input county population data from another dataset, in which "county_name" is unique. Like this:

1. hospital  county_name other variables...

    XX          AA                   XXXXX

    xxxx        AA                    XXXXX

    XXX        BB                    XXXXX

......

2. county_name  population

    AA                   123345

    BB                    54321

......

I have to state market population(in this case county population) for every hospital. Thanks!

you can merge directly. Stata does not require the common varible to have unique values per observation.

. use data1, clear
. sort county_name
. save, replace

. use data2, clear
. sort county_name
. merge county_name using data1

Let me know if there is any problem

no, it does not work, just as #2 said. index should be unique to merge datasets.

In this case, you do not have to merge in a matching way. The way #4 suggested is supposed to be workable. Make sure you use the first dataset as the MASTER dataset.  If it still does not work, I guess the problem might be that the country_name is not consistently numeric. Make sure country_name has uniform label values.

[此贴子已经被作者于2008-6-17 23:17:54编辑过]