Matching/ Partnering in Stata

[Toni Moreno]

Dear all,

I am currently trying to figure out a way to match suitable partners/ individuals with the help of their individual characteristics.

I have one dataset containing panel information on women and men. Some of them do have partner information (partner ID), while others do not.

Now I need to assign a partner to an individual who is married in for example 2005 but has no partner ID. The potential donor does not have to be single and can also be in a relationship already. Furthermore, individuals can be used as partners multiple times.

In a first step, I imputed the expected age and educational level of an individuals partner with the help of chained imputation.

But now I do not know how to find a suitable partner from the dataset and assign their personal ID as an individuals partner ID.

I thought about using propensity score matching and therefore the psmatch2 command and came up with something like:

 psmatch2 treat if syear==2005,  mahalanobis (Female_age Female_education) neighbor(1)

where treat says more or less whether the individual is male or female and has therefore be matched with partners of the other sex (female_age female_education).

But I am not sure whether this really works and if yes, how to identify the ID´s of matched individuals since _id and _n1 do not show partner IDs.

Maybe there is also a better and more elegant way to deal with this problem in Stata?

Any help would be appreciated. Thank you very much in advance,
Toni

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata Code in code delimiters, readable Stata output, and sample data using dataex.

I don't do matching. You should first see how folks do matching in your discipline. Next, try to repost with additional code and sample data, and see if one of the folks who does matching can be induced to help.

[Mike Lacy]

For a recent project, I found -calipmatch- quite convenient. Don't know what your purposes ("suitable" for what?) are or what your data is like here, so I can't respond further, but you might try
-ssc describe calipmatch- .

[Toni Moreno]

Thank you very much Phil and Mike for your replies. I will try to specify my question and add additional code and sample data in a repost