Re: st: Reclink: high matching score, but no match

[Devra Golbe <dgolbe@gmail.com>]

Dorothy and all,

My apologies for not closing this thread properly.   Michael solved my 
problem in private correspondence, and I failed to report back to the list.
The problem was that in  successive runs of my do-file I had managed to 
save the idusing variable to the master dataset.  Once I dropped it from 
the master file reclink ran fine.
Devra

On 1/24/2014 1:43 PM Dorothy Bridges wrote:
> Hello everyone (especially Devra and Michael): Was this ever resolved?
> I'm having the exact same problem. Code and (partial) output copied
> below. I usually use reclink without any problems.
>
> reclink entidad municipio using
> "ESDATA/dta/MunicipioLevel/ESDATAMun_Jan2014.dta", ///
>
>          idu(idu) idm(idm) gen(match) required(entidad)
>
> listing the output:
>
> entidad                    municipio
> Umunicipio              match
>                 ZULIA               VALMORE RODRIGUEZ          SAN
> FRANCISCO   0.9961
>         NUEVA ESPARTA               ANTOLIN DEL CAMPO
> SOTILLO   0.9961
>               YARACUY               JOSE ANTONIO PAEZ
> BRUZUAL   0.9961
>                 ZULIA           LA CANADA DE URDANETA       FRANCISCO J
> PULG   0.9974
>                ARAGUA          MARIO BRICENO IRAGORRY   M.OCUMARE D LA
> COSTA   0.9977
>                 ZULIA               ROSARIO DE PERIJA
> MARA   0.9982
>                ARAGUA     FRANCISCO LINARES ALCANTARA   FRANCISCO
> LINARES A.   0.9992
>               BARINAS        ALBERTO ARVELO TORREALBA
> ZAMORA   0.9995
>                 SUCRE                          RIBERO
> MEJIA   1.0000
>              CARABOBO                          BEJUMA
> SIFONTES   1.0000
>               MIRANDA                        URDANETA           RIVAS
> DAVILA   1.0000
>               YARACUY                    MANUEL MONGE
> INDEPENDENCIA   1.0000
>         DELTA AMACURO                       CASACOIMA
> TINACO   1.0000
>                 SUCRE                           SUCRE
> MONTES   1.0000
>         NUEVA ESPARTA                           GOMEZ
> GARCIA   1.0000
>                ARAGUA                JOSE ANGEL LAMAS       JOSE ANGEL
> LAMAS   1.0000
>               BARINAS                    CRUZ PAREDES
> BARINAS   1.0000
>               MIRANDA                           SUCRE
> RANGEL   1.0000
>               MIRANDA                   SIMON BOLIVAR           PUEBLO
> LLANO   1.0000
>                 ZULIA                            PAEZ         MACHIQUES
> DE P   1.0000
>
> On Wed, Dec 28, 2011 at 10:04 AM, Devra Golbe <dgolbe@gmail.com> wrote:
> > Michael,
> >
> > student_name is non-numeric.  After some additional data cleaning and the
> > resulting reduction of the set that needed a fuzzy match  reclink succeeded
> > with student_name as the idusing variable, so my original problem is solved.
> >
> > But working with a smaller data set, I have an example where the non-numeric
> > identifier and a numeric identifier fail, but a different numeric identifier
> > succeeds.  I'll send those data and the do-file to you off-list.
> >
> > Thanks and happy new year.
> >
> > Devra
> >
> >
> > On 12/28/2011 11:49 AM Michael Blasnik wrote:
> > > It looks like this is a bug -- is student_name numeric?  If not, you
> > > may want to try encoding it and trying again.  If that isn't the
> > > problem, it might be best if you either send me the data or a trace
> > > log off-list to see if i can figure it out, but I may not get a chance
> > > to figure it out until after the holidays.
> > >
> > > Michael
> > >
> > > On Sat, Dec 24, 2011 at 4:10 PM, Devra Golbe<dgolbe@gmail.com>  wrote:
> > > > I am using  Michael Blasnik's reclink (from SSC) to match records.  I get
> > > > extremely high matching scores, and yet the records do not match.  Can
> > > > anyone help?    My code and relevant output are pasted below.
> > > >
> > > > Thanks and happy holidays,
> > > > Devra
> > > >
> > > > ******
> > > > . sort lname fname
> > > >    . gen idmaster=_n
> > > >    .tempfile ps1a
> > > >    .save `ps1a', replace
> > > >    . clear
> > > >    .use roster100f11Sep7.dta
> > > >    .sort lname fname
> > > >    .save, replace
> > > >    .clear
> > > >    .use `ps1a'
> > > >
> > > >    .reclink lname fname using roster100f11Sep7.dta, ///
> > > >     idmaster(idmaster) idusing(student_name) gen(link)
> > > >
> > > > 0 perfect matches found
> > > >
> > > >
> > > > Added: student_name= identifier from roster100f11Sep7.dta   link =
> > > > matching
> > > > score
> > > > Observations:  Master N = 26    roster100f11Sep7.dta N= 182
> > > >    Unique Master Cases: matched = 0 (exact = 0), unmatched = 26
> > > >
> > > > .list link _merge in 1/5, clean
> > > >
> > > >           link   _merge
> > > >    1.   0.9933        1
> > > >    2.   0.9933        1
> > > >    3.        .        1
> > > >    4.   0.6420        1
> > > >    5.   0.9988        1
> > > >
> > > > _______
> > > > Devra Golbe
> > > > Professor of Economics
> > > > Hunter College, CUNY
> > > > NY, NY
> > > > *
> > > > *
> > *
> > *   For searches and help try:
> > *   http://www.stata.com/help.cgi?search
> > *   http://www.stata.com/support/statalist/faq
> > *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/