SPSSX Discussion

Propensity score matching (318 versus 87 N)

Classic

List

Threaded

5 messages Options

Ergul, Emel A.

Propensity score matching (318 versus 87 N)

Dear all:
Planning to compare two treatment groups by using propensity score matching but total number of cases in treatment A= 318 and Treatment B=90, all my variables are categorical. Regret that I've never used propensity score macro before. Is there anyone knows the macro for how to create propensity scores between these two groups with significant unequal sample sizes?

Thank you,
Kind regards,
Emel


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.

Swank, Paul R

Re: Propensity score matching (318 versus 87 N)

Propensity score matching (318 versus 87 N)

Why use matching? Why not just do a logistic regression using the covariates as predictors of the treatment groups and then output the predicted values and use those as a covariate for your final analysis.

Dr. Paul R. Swank,

Professor and Director of Research

Children's Learning Institute

University of Texas Health Science Center-Houston

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ergul, Emel A.
Sent: Thursday, February 12, 2009 12:05 PM
To: [hidden email]
Subject: Propensity score matching (318 versus 87 N)

Thank you,
Kind regards,
Emel

The information in this e-mail is intended only for the person to whom it is

addressed. If you believe this e-mail was sent to you in error and the e-mail

contains patient information, please contact the Partners Compliance HelpLine at

http://www.partners.org/complianceline . If the e-mail was sent to you in error

but does not contain patient information, please contact the sender and properly

dispose of the e-mail.

Ergul, Emel A.

Re: Propensity score matching (318 versus 87 N)

Propensity score matching (318 versus 87 N)

Paul,

subjects were not assigned to the treatment groups randomly, my goal to use propensity score matching is simply to take care of background characteristics; and propensity score matching is quite new to me so I want to learn how to do it.

It would be interesting compare the two methodology and see where is the difference if any.

Do I make sense?

thanks,

emel

From: Swank, Paul R [mailto:[hidden email]]
Sent: Thursday, February 12, 2009 1:30 PM
To: Ergul, Emel A.; [hidden email]
Subject: RE: Propensity score matching (318 versus 87 N)

Dr. Paul R. Swank,

Professor and Director of Research

Children's Learning Institute

University of Texas Health Science Center-Houston

Thank you,
Kind regards,
Emel

The information in this e-mail is intended only for the person to whom it is

addressed. If you believe this e-mail was sent to you in error and the e-mail

contains patient information, please contact the Partners Compliance HelpLine at

http://www.partners.org/complianceline . If the e-mail was sent to you in error

but does not contain patient information, please contact the sender and properly

dispose of the e-mail.

Zetu, Dan

Re: Propensity score matching (318 versus 87 N)

Propensity score matching (318 versus 87 N)

Propensity score matching is nothing but a logistic regression with treatment/control group assignment as dependent variable and background characteristics as covariates. Then sort the predicted probabilities in descending order, divide the data into equal groups (such as deciles - in your case, given that the population is small, you should use maybe tertiles or quartiles), calculate the proportion of treatment and control subjects in each group, divide the proportions and these are your propensity weights.

I hope this gives you an idea on how to proceed.

-------------------------------

Dan Zetu

Analytical Consultant

R. L. Polk & Co.

248-728-7278

[hidden email]

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ergul, Emel A.
Sent: Thursday, February 12, 2009 3:27 PM
To: [hidden email]
Subject: Re: Propensity score matching (318 versus 87 N)

Paul,

It would be interesting compare the two methodology and see where is the difference if any.

Do I make sense?

thanks,

emel

From: Swank, Paul R [mailto:[hidden email]]
Sent: Thursday, February 12, 2009 1:30 PM
To: Ergul, Emel A.; [hidden email]
Subject: RE: Propensity score matching (318 versus 87 N)

Dr. Paul R. Swank,

Professor and Director of Research

Children's Learning Institute

University of Texas Health Science Center-Houston

Thank you,
Kind regards,
Emel

The information in this e-mail is intended only for the person to whom it is

addressed. If you believe this e-mail was sent to you in error and the e-mail

contains patient information, please contact the Partners Compliance HelpLine at

http://www.partners.org/complianceline . If the e-mail was sent to you in error

but does not contain patient information, please contact the sender and properly

dispose of the e-mail.

*****************************************************************
This message has originated from R. L. Polk & Co.,
26955 Northwestern Highway, Southfield, MI 48033.
R. L. Polk & Co. sends various types of email
communications. If this email message concerns the
potential licensing of a Polk product or service, and
you do not wish to receive further emails regarding Polk
products, forward this email to [hidden email]
with the word "remove" in the subject line.

The email and any files transmitted with it are confidential
and intended solely for the individual or entity to whom they
are addressed.

If you have received this email in error, please delete this
message and notify the Polk System Administrator at
[hidden email].
*****************************************************************

Swank, Paul R

Re: Propensity score matching (318 versus 87 N)

In reply to this post by Ergul, Emel A.

Propensity score matching (318 versus 87 N)

If you had random assignment, you would need propensity scores. Since you don’t have random assignment, they may be useful. Propensity scores may be used in two ways. One way is to match subjects in groups to make the groups more similar. In your case, however, you would have to do a many to one match since you have more than three times as many in group A as B. Even then you will have 48 cases left over. The other way to make use of propensity scores is to use them as you covariate. This allow you to control for many covariates with only one variable and also to use your entire subject base.

Dr. Paul R. Swank,

Professor and Director of Research

Children's Learning Institute

University of Texas Health Science Center-Houston

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ergul, Emel A.
Sent: Thursday, February 12, 2009 2:27 PM
To: [hidden email]
Subject: Re: Propensity score matching (318 versus 87 N)

Paul,

It would be interesting compare the two methodology and see where is the difference if any.

Do I make sense?

thanks,

emel

From: Swank, Paul R [mailto:[hidden email]]
Sent: Thursday, February 12, 2009 1:30 PM
To: Ergul, Emel A.; [hidden email]
Subject: RE: Propensity score matching (318 versus 87 N)

Dr. Paul R. Swank,

Professor and Director of Research

Children's Learning Institute

University of Texas Health Science Center-Houston

Thank you,
Kind regards,
Emel

The information in this e-mail is intended only for the person to whom it is

addressed. If you believe this e-mail was sent to you in error and the e-mail

contains patient information, please contact the Partners Compliance HelpLine at

http://www.partners.org/complianceline . If the e-mail was sent to you in error

but does not contain patient information, please contact the sender and properly

dispose of the e-mail.