SPSSX Discussion

Comparing samples on SPSS

Classic

List

Threaded

7 messages Options

Eugenio Grant

Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words, if
they are comparable with each other or not?

Regards,

Anthony Babinec

Re: Comparing samples on SPSS

Code a variable "1" if a case is from the first sample
and a "0" if it is from the second, then set up
a logistic regression that predicts this variable.
Your hope is for no significant or sizeable predictors
of SAMPLE.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Eugenio Grant
Sent: Thursday, February 22, 2007 4:48 PM
To: [hidden email]
Subject: Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words, if
they are comparable with each other or not?

Regards,

Swank, Paul R

Re: Comparing samples on SPSS

With emphasis on sizable since with this sample size, you will likely
find significant differences that may be trivial.

Paul R. Swank, Ph.D. Professor
Director of Reseach
Children's Learning Institute
University of Texas Health Science Center-Houston

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Anthony Babinec
Sent: Thursday, February 22, 2007 9:53 PM
To: [hidden email]
Subject: Re: Comparing samples on SPSS

Code a variable "1" if a case is from the first sample and a "0" if it
is from the second, then set up a logistic regression that predicts this
variable.
Your hope is for no significant or sizeable predictors of SAMPLE.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Eugenio Grant
Sent: Thursday, February 22, 2007 4:48 PM
To: [hidden email]
Subject: Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of
variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words,
if they are comparable with each other or not?

Regards,

Dennis Deck

Re: Comparing samples on SPSS

In reply to this post by Eugenio Grant

What Anthony describes is essentially a step in Propensity Score
Analysis. The approach encourages you to carefully model the sample
differences.

You can save the predicted score from the logistic regression and use it
in subsequent substantive analyses where you want to adjust for sample
differences (eg, as a covariate, weight, matching variable, etc).

Be sure to properly handle your categorical predictors.

Dennis Deck, PhD
RMC Research Corporation
[hidden email]

-----Original Message-----
From: Anthony Babinec [mailto:[hidden email]]
Sent: Thursday, February 22, 2007 7:53 PM
Subject: Re: Comparing samples on SPSS

Code a variable "1" if a case is from the first sample
and a "0" if it is from the second, then set up
a logistic regression that predicts this variable.
Your hope is for no significant or sizeable predictors
of SAMPLE.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Eugenio Grant
Sent: Thursday, February 22, 2007 4:48 PM
To: [hidden email]
Subject: Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of
variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words,
if
they are comparable with each other or not?

Regards,

Ornelas, Fermin

Re: Comparing samples on SPSS

In reply to this post by Eugenio Grant

It seems that the samples are not large enough to make results
meaningful in a logistic regression.

What it would make sense given the samples is some descriptive
statistics such as mean, standard deviation, median min and max values
for each sample, and frequencies on the categorical variables and
compare their values. You may also do a t-test comparison of the means.

Fermin Ornelas, Ph.D.
Management Analyst III, AZ DES
Tel: (602) 542-5639
E-mail: [hidden email]

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Dennis Deck
Sent: Friday, February 23, 2007 1:05 PM
To: [hidden email]
Subject: Re: Comparing samples on SPSS

What Anthony describes is essentially a step in Propensity Score
Analysis. The approach encourages you to carefully model the sample
differences.

You can save the predicted score from the logistic regression and use it
in subsequent substantive analyses where you want to adjust for sample
differences (eg, as a covariate, weight, matching variable, etc).

Be sure to properly handle your categorical predictors.

Dennis Deck, PhD
RMC Research Corporation
[hidden email]

-----Original Message-----
From: Anthony Babinec [mailto:[hidden email]]
Sent: Thursday, February 22, 2007 7:53 PM
Subject: Re: Comparing samples on SPSS

Code a variable "1" if a case is from the first sample
and a "0" if it is from the second, then set up
a logistic regression that predicts this variable.
Your hope is for no significant or sizeable predictors
of SAMPLE.

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Eugenio Grant
Sent: Thursday, February 22, 2007 4:48 PM
To: [hidden email]
Subject: Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of
variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words,
if
they are comparable with each other or not?

Regards,

NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR
CONFIDENTIAL information and is intended only for the use of the
specific individual(s) to whom it is addressed. It may contain
information that is privileged and confidential under state and federal
law. This information may be used or disclosed only in accordance with
law, and you may be subject to penalties under law for improper use or
further disclosure of the information in this e-mail and its
attachments. If you have received this e-mail in error, please
immediately notify the person named above by reply e-mail, and then
delete the original e-mail. Thank you.

Dennis Deck

Re: Comparing samples on SPSS

Rereading the original post I realize that the intended question was
probably: "How can I see whether the two samples differ on each of these
categorical variables". A conventional approach would be to use T-TEST
for continuous variables (age) and CROSSTABS for categorical variables
(marital status). With such large N's (9K-10K), even trivial
differences will likely be significant so you may want to also calculate
effect sizes or standardized differences.

However, in many fields the researcher must compare naturally occurring
groups which typically differ on multiple variables. Propensity Score
Analysis (generally using LOGISTIC) offers a useful multivariate
strategy for modeling the overall differences in the groups and
adjusting for these differences in subsequent analyses. If you have two
samples drawn in different years or from different populations and you
want to say something about any differences adjusting for sample
characteristics, this would be an appropriate approach.

Dennis Deck, PhD
RMC Research Corporation
[hidden email]

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Dennis Deck
Sent: Friday, February 23, 2007 1:05 PM
To: [hidden email]
Subject: Re: Comparing samples on SPSS

What Anthony describes is essentially a step in Propensity Score
Analysis. The approach encourages you to carefully model the sample
differences.

You can save the predicted score from the logistic regression and use it
in subsequent substantive analyses where you want to adjust for sample
differences (eg, as a covariate, weight, matching variable, etc).

Be sure to properly handle your categorical predictors.

Dennis Deck, PhD
RMC Research Corporation
[hidden email]

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Eugenio Grant
Sent: Thursday, February 22, 2007 4:48 PM
To: [hidden email]
Subject: Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of
variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words,
if
they are comparable with each other or not?

Regards,

NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR
CONFIDENTIAL information and is intended only for the use of the
specific individual(s) to whom it is addressed. It may contain
information that is privileged and confidential under state and federal
law. This information may be used or disclosed only in accordance with
law, and you may be subject to penalties under law for improper use or
further disclosure of the information in this e-mail and its
attachments. If you have received this e-mail in error, please
immediately notify the person named above by reply e-mail, and then
delete the original e-mail. Thank you.

Eugenio Grant

Re: Comparing samples on SPSS

Thanks Dennis that's exactly what I need: To establish by some manner if
both samples are correct (by correct I mean that if they were properly
taken, they should show some differences but not to the point of saying that
sample A is not comparable to sample B on the provided variables.

Is there a way to do this on SPSS, SPSS syntax or some place with info would
be great considering statistics is not my specialty.

Regards,

-----Mensaje original-----
De: SPSSX(r) Discussion [mailto:[hidden email]] En nombre de
Dennis Deck
Enviado el: Viernes, 23 de Febrero de 2007 03:34 p.m.
Para: [hidden email]
Asunto: Re: Comparing samples on SPSS

Rereading the original post I realize that the intended question was

probably: "How can I see whether the two samples differ on each of these

categorical variables". A conventional approach would be to use T-TEST

for continuous variables (age) and CROSSTABS for categorical variables

(marital status). With such large N's (9K-10K), even trivial

differences will likely be significant so you may want to also calculate

effect sizes or standardized differences.

However, in many fields the researcher must compare naturally occurring

groups which typically differ on multiple variables. Propensity Score

Analysis (generally using LOGISTIC) offers a useful multivariate

strategy for modeling the overall differences in the groups and

adjusting for these differences in subsequent analyses. If you have two

samples drawn in different years or from different populations and you

want to say something about any differences adjusting for sample

characteristics, this would be an appropriate approach.

Dennis Deck, PhD

RMC Research Corporation

[hidden email]

-----Original Message-----

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of

Dennis Deck

Sent: Friday, February 23, 2007 1:05 PM

To: [hidden email]

Subject: Re: Comparing samples on SPSS

What Anthony describes is essentially a step in Propensity Score

Analysis. The approach encourages you to carefully model the sample

differences.

You can save the predicted score from the logistic regression and use it

in subsequent substantive analyses where you want to adjust for sample

differences (eg, as a covariate, weight, matching variable, etc).

Be sure to properly handle your categorical predictors.

Dennis Deck, PhD

RMC Research Corporation

[hidden email]

-----Original Message-----

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of

Eugenio Grant

Sent: Thursday, February 22, 2007 4:48 PM

To: [hidden email]

Subject: Comparing samples on SPSS

Hello to all

I have results form a sample of 10.000 interviews for a series of

variables:

Age (1,2,3,4...99)

Marital Status (Single, married, divorced, etc)

Persons in home (1,2,3,4,5,6, more)

Income Level (High, Medium, Low)

And results form another sample of 9.000 interviews.

How can I establish if both samples are correctly taken, in other words,

if

they are comparable with each other or not?

Regards,

NOTICE: This e-mail (and any attachments) may contain PRIVILEGED OR

CONFIDENTIAL information and is intended only for the use of the

specific individual(s) to whom it is addressed. It may contain

information that is privileged and confidential under state and federal

law. This information may be used or disclosed only in accordance with

law, and you may be subject to penalties under law for improper use or

further disclosure of the information in this e-mail and its

attachments. If you have received this e-mail in error, please

immediately notify the person named above by reply e-mail, and then

delete the original e-mail. Thank you.