Re: Propensity score matching in PSPP

[Alan Mead]

I don't know.  If the option doesn't appear in the GUI you could see if the option is available in syntax and, if so, create the syntax using the GUI and then paste it.  It would probably be the same syntax in PSPP as SPSS but the issue would be whether it's implemented and I don't know.

However, I have never used a multivariate routine to create "score" variables in either PSPP or SPSS.  I create new variables using compute statements, which is a little more work but (assuming you save the syntax) makes the process self-documenting and replicable. In general, I would strongly advocate for the exclusive use of syntax for analyses (by all means, use the GUI to create the syntax, but paste, run, and save the syntax). But maybe these may be less serious considerations for your context.

I'm not particularly familiar with logistic regression, but it looks like you would create predicted values with something like: compute ps = 1/(1+exp(-(<alpha>+<beta>*x))). where ps is the new propensity score, x is the matching variable and <alpha> and <beta> are the numerical values in the logistic output. 

I hope this helps.

-Alan

On 10/2/2013 10:04 AM, Suniya Farooqui wrote:

Thanks Alan! I can run a logistic regression but where can I find the option to create a new variable and save the predicted probabilities in PSPP (similar to the option in SPSS)?

On Tue, Oct 1, 2013 at 4:57 PM, Alan Mead <address@hidden> wrote:

Wikipedia describes these steps in propensity score matching:

1.Run logistic regression:
    Dependent variable: Y = 1, if participate; Y = 0, otherwise.
    Choose appropriate conditioning (instrumental) variables.
    Obtain propensity score: predicted probability (p) or log[p/(1 − p)].

2.Match each participant to one or more nonparticipants on propensity score:
    Nearest neighbor matching
    Caliper matching
    Mahalanobis metric matching in conjunction with PSM
    Stratification matching
    Difference-in-differences matching (kernel and local linear weights)

3.Multivariate analysis based on new sample
    Use analyses appropriate for non-independent matched samples

So, yes PSPP could potentially be used to implement these steps, but you would have to understand these steps and be able to complete each one.  OTOH, no, PSPP doesn't have a routine that will automatically perform all these steps.

-Alan

On 10/1/2013 4:37 PM, Suniya Farooqui wrote:

Hi,

Can propensity score matching be done in PSPP? If so, can someone please share those instructions?

I using PSPP 8.0 on Windows 7.

Thank you,

suniya

_______________________________________________
Pspp-users mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/pspp-users

-- 

Alan D. Mead, Ph.D.
Assistant Professor
Industrial and Organizational Psychology Program
Department of Psychology
Lewis College of Human Sciences
Illinois Institute of Technology
3101 South Dearborn, 2nd floor
Chicago IL 60616

+312.567.5933 (Campus)
+815.588.3846 (Home Office)
+267.334.4143 (Mobile)
+312.567.3493 (Fax)

http://www.iit.edu/~mead
http://www.alanmead.org

Announcing the Ideas in Testing Research Seminar, October 11, 2013.
Free registration. Complete details: 
http://mypages.iit.edu/~mead/ideas2013/

Announcing the Journal of Computerized Adaptive Testing (JCAT), a
peer-reviewed electronic journal designed to advance the science and
practice of computerized adaptive testing: http://www.iacat.org/jcat

--

Suniya Farooqui | Research Associate

Social IMPACT Research Center

a program of Heartland Alliance

33 W. Grand Avenue, Suite 500 | Chicago, IL 60654

P: 312.870.4952 F: 312.870.4950 M: 562.644.3472 

-- 

Alan D. Mead, Ph.D.
President, Talent Algorithms Inc.

+815.588.3846 (Office)
+267.334.4143 (Mobile)

http://www.alanmead.org

Announcing the Journal of Computerized Adaptive Testing (JCAT), a
peer-reviewed electronic journal designed to advance the science and
practice of computerized adaptive testing: http://www.iacat.org/jcat