Unifying country names when merging data

[Mustapha Younis]

I have data from different sources, some use names for a country and other sources use different name. Such as Hong Kong and Hong Kong, China. I merged the data but I have the two countries separate with different IDs. I want to merge both together in one ID and unify the name to be Hong Kong and merge all the variables together under this name. Thanks for helping.

[Ali Atia]

Your best bet is to employ a standardized coding scheme. Take a look at the -kountry- command from SSC, which may be what you're looking for.

[Maarten Buis]

A good place to start, that will capture a lot of the common variations is to use Rafal Raciborski's kountry, see ssc desc kountry.

[Mustapha Younis]

Thanks so much Mr. Ali and Mr. Maarten for the reply.

Can I use it after I already merged the data or it should be before merging?

Thanks

[Ali Atia]

Should be used before merging. In both datasets, use

kountry country_var, from(other) stuck marker

This will generate an iso3n code variable which can then be used as the key when you merge. It will also generate a marker variable identifying successful standardizations. Before merging, you should check if any countries failed to standardize using

tab if marker==0

[Mustapha Younis]

Ok Thanks Mr. Ali. If I found a country that is failed to standardize and the marker==0 what should I do in this case? Thanks

[Ali Atia]

If the countries which failed to standardize are low in quantity, you may find it easiest to simply recode them manually by creating temporary four-digit codes for those countries, replacing the missing _ISO3N_ in both datasets:

replace _ISO3N_ = <new four digit code> if country == "nonstandardized"

If there are many non-standardized countries, you can create a new dataset containing one observation for each of the non-standardized countries and a variable called _ISO3N_. Then you can create new unique four digit codes (to ensure there is no clash with the preexisting three digit codes) temporary codes for each of those countries:

replace _ISO3N_ = _n+999

Merge this dataset with each of your master datasets using the update option, then merge both master datasets using _ISO3N_ as the key.

[Mustapha Younis]

Thanks Mr. Ali. That was really beneficial.

[Farza Chow]

Hi;
I am getting the following error after running tab code:

tab if marker == 0
varlist required
r(100);

[Clyde Schechter]

Yes, of course you are. Your command asks Stata to create a table, but you don't specify what variable(s) you want it to tabulate. That is what you need to fill in between -tab- and -if-.