Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Matching


From   Sophie Jensen <thorns.n.roses@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Matching
Date   Mon, 4 Oct 2010 23:54:37 +0100

Also, Im not sure Im meant to have that many observations... I think I
have too many... my matched sample should have less than a few hundred
observations... would you know why I have so many?

On Mon, Oct 4, 2010 at 11:51 PM, Sophie Jensen <thorns.n.roses@gmail.com> wrote:
> Btw, by neigherst neighbour we mean the most compositionally similar - right?
>
> On Mon, Oct 4, 2010 at 11:51 PM, Sophie Jensen <thorns.n.roses@gmail.com> wrote:
>> how do you ensure random sorting... just sort on some varibale?
>> How do you do area matching... radius matching... do you know
>>
>> Thanks
>>
>> S
>> On Mon, Oct 4, 2010 at 11:50 PM, Sophie Jensen <thorns.n.roses@gmail.com> wrote:
>>> Here's what I've done... how would you interpret that:
>>>
>>> . psmatch2 treatment white fsm male if year>2006, n(1)
>>>
>>> Probit regression                                 Number of obs   =      15471
>>>                                                  LR chi2(4)      =     322.40
>>>                                                  Prob > chi2     =     0.0000
>>> Log likelihood = -678.85535                       Pseudo R2       =     0.1919
>>>
>>> ------------------------------------------------------------------------------
>>>   treatment |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
>>> -------------+----------------------------------------------------------------
>>>   ewhite |    .172414   .1514341     1.14   0.255    -.1243915    .4692194
>>>         fsm |   1.650061   .2705032     6.10   0.000     1.119885    2.180238
>>> male |   .0000787   .0000105     7.50   0.000     .0000581    .0000992
>>>       _cons |   .4350676   .8062152     0.54   0.589    -1.145085     2.01522
>>> ------------------------------------------------------------------------------
>>>
>>>
>>> pstest fsm eth_white ks2mn outwage_male
>>>
>>> ----------------------------------------------------------------------------
>>>                        |       Mean               %reduct |     t-test
>>>    Variable     Sample | Treated Control    %bias  |bias| |    t    p>|t|
>>> ------------------------+----------------------------------+----------------
>>>         fsm  Unmatched | .32883   .13543    136.0         |  18.34  0.000
>>>                Matched | .32883     .328      0.6    99.6 |   0.04  0.968
>>>                        |                                  |
>>> white  Unmatched | .58394     .829    -89.1         | -12.70  0.000
>>>                Matched | .58394   .54751     13.2    85.1 |   1.00  0.318
>>>                        |                                  |
>>>  male  Unmatched |  28221    24729     92.7         |  11.82  0.000
>>>                Matched |  28221    28606    -10.2    89.0 |  -0.87  0.387
>>>                        |                                  |
>>>
>>>
>>>
>>> On Mon, Oct 4, 2010 at 11:35 PM, Santosh Kumar <santosh.uh@gmail.com> wrote:
>>>> This is an example of matching with 3 neighbors.
>>>>
>>>> psmatch2 improved, pscore(pscore) outcome(diarr) common caliper(.001)
>>>> neighbor(3)
>>>> There are observations with identical propensity score values.
>>>> The sort order of the data could affect your results.
>>>> Make sure that the sort order is random before calling psmatch2.
>>>> (4005 missing values generated)
>>>>
>>>> Variable     Sample     Treated     Controls   Difference         S.E.   T-stat
>>>>
>>>> diarr  Unmatched  .104890901   .125810627  -.020919727   .001373497   -15.23
>>>> ATT  .104893374   .110885821  -.005992448   .007151216    -0.84
>>>>
>>>> Note: S.E. for ATT does not take into account that the propensity
>>>> score is estimated.
>>>>
>>>> psmatch2:    psmatch2: Common
>>>> Treatment         support
>>>> assignment  Off suppo  On suppor      Total
>>>>
>>>> Untreated          0    159,136    159,136
>>>> Treated          2     84,829     84,831
>>>>
>>>> Total          2    243,965    243,967
>>>>
>>>> On Mon, Oct 4, 2010 at 6:28 PM, Sophie Jensen <thorns.n.roses@gmail.com> wrote:
>>>>> Could someone perhaps explain the process of this kind of matching please.
>>>>>
>>>>> Thanks
>>>>>
>>>>> S
>>>>>
>>>>> On Mon, Oct 4, 2010 at 11:26 PM, Sophie Jensen <thorns.n.roses@gmail.com> wrote:
>>>>>> Dear statlisters,
>>>>>>
>>>>>> What is the difference between a 1-1 to match and the nearest
>>>>>> neighbour match in stata. I would like to do a nearest neighbour
>>>>>> match, a 1-1 and then specifiy a distance to match on.
>>>>>>
>>>>>> Thanks
>>>>>>
>>>>>> S
>>>>>> *
>>>>>> *   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/statalist/faq
>>>>> *   http://www.ats.ucla.edu/stat/stata/
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>>
>>>> Santosh
>>>>
>>>> *
>>>> *   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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index