Is there a way to get eigenvalues from CATPCA?

[Art Kendall]

I am familiarizing myself with CATPCA.  Is there a way to have CATPCA produce eigenvalues so that I can can compare them to those from a PCA factor analysis of the same data.
I have 14 dichotomous variables coded 1 for no and 2 for yes. (CATPCA does not like 0,1.)
I have 51,000 cases.

So far I have this syntax what further specifications do I have to make?
catpca variables = vr46 to vr59
 /analysis = vr46 to vr59 (level=nume)
 /dimension = 3
 /plot = loading ndim(1,3)
 /PRINT = none.
catpca variables = vr46 to vr59
 /analysis = vr46 to vr59 (level=nume)
 /dimension = 3
 /plot = loading ndim(2,3)
 /PRINT = none.


Art Kendall
Social Research Consultants

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

[Dale Glaser]

Hi all......I have ran a multiplicative model (group x test score) using zero-inflated Poisson (ZIP) in Mplus given the DV (# of fails). The interaction term is significant for the Poisson component of the model, and now we are trying to plot the interaction  This is where I am stuck.  I couldn't find a way by which to graph this result via Mplus so I used the GENLIN option in SPSS with the log link function for the Poisson model, and thought if I saved the predicted scores I could plot the two groups x moderator interaction (as you can with OLS regression when you graph a scatterplot of the moderator and outcome and use the subgroup fit line option to super-impose the predicted lines for each group).  However, as most of you would expect when I graph the predicted scores from GENLIN against the moderator and plot the two group fit lines, it only shows the two lines since it is a perfect fit for each group.  So, does anyone have any suggestions how to overlay the predicted line for subgroups over the scatterplot based on the results from a Poisson model?

 

thank you very much....dale

 

Dale Glaser, Ph.D.
Principal--Glaser Consulting
Lecturer/Adjunct Faculty--SDSU/USD/Alliant
Past-President, San Diego Chapter of
American Statistical Association
3115 4th Avenue
San Diego, CA 92103
phone: [phone]
fax: [phone]
email: [hidden email]
website: www.glaserconsult.com

[Kooij, A.J. van der]

The eigenvalues are in the output: see the Model Summary Table (always displayed, even if /print none)

under the heading "Variance Accounted For", subheading "Total (Eigenvalues)".

 

Regards,

Anita van der Kooij,

Data Theory Scaling System Group,

Leiden University.

---

From: SPSSX(r) Discussion on behalf of Art Kendall
Sent: Thu 03-Feb-11 22:41
To: [hidden email]
Subject: Is there a way to get eigenvalues from CATPCA?

I am familiarizing myself with CATPCA.  Is there a way to have CATPCA produce eigenvalues so that I can can compare them to those from a PCA factor analysis of the same data.
I have 14 dichotomous variables coded 1 for no and 2 for yes. (CATPCA does not like 0,1.)
I have 51,000 cases.

So far I have this syntax what further specifications do I have to make?
catpca variables = vr46 to vr59
 /analysis = vr46 to vr59 (level=nume)
 /dimension = 3
 /plot = loading ndim(1,3)
 /PRINT = none.
catpca variables = vr46 to vr59
 /analysis = vr46 to vr59 (level=nume)
 /dimension = 3
 /plot = loading ndim(2,3)
 /PRINT = none.


Art Kendall
Social Research Consultants

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

**********************************************************************

This email and any files transmitted with it are confidential and

intended solely for the use of the individual or entity to whom they

are addressed. If you have received this email in error please notify

the system manager.

**********************************************************************

 

[Bruce Weaver]

How about adding lines from equations, or whatever that option is called (in the Chart Editor)?

Dale Glaser wrote

Hi all......I have ran a multiplicative model (group x test score) using zero-inflated Poisson (ZIP) in Mplus given the DV (# of fails). The interaction term is significant for the Poisson component of the model, and now we are trying to plot the interaction  This is where I am stuck.  I couldn't find a way by which to graph this result via Mplus so I used the GENLIN option in SPSS with the log link function for the Poisson model, and thought if I saved the predicted scores I could plot the two groups x moderator interaction (as you can with OLS regression when you graph a scatterplot of the moderator and outcome and use the subgroup fit line option to super-impose the predicted lines for each group).  However, as most of you would expect when I graph the predicted scores from GENLIN against the moderator and plot the two group fit lines, it only shows the two lines since it is a perfect fit for each group.  So, does anyone have any suggestions how
 to overlay the predicted line for subgroups over the scatterplot based on the results from a Poisson model?
 
thank you very much....dale
 

Dale Glaser, Ph.D.
Principal--Glaser Consulting
Lecturer/Adjunct Faculty--SDSU/USD/Alliant
Past-President, San Diego Chapter of
American Statistical Association
3115 4th Avenue
San Diego, CA 92103
phone: [phone]
fax: [phone]
email: [e-mail]
website: www.glaserconsult.com

[Ryan]

Dale,

 

I'm confused by your post. First you state that you fit a zero-inflated poisson (ZIP) model in MPLUS, presumably because you have count data with excess zeros, but you then reportedly fit a Poisson model on the same data in SPSS to obtain predicted values. Which model yields a superior fit? (You could construct a likelihood ratio test to compare overall fit of each model.) If the ZIP model is superior, then that is the model from which you should obtain predicted values.

 

Ryan

 

On Thu, Feb 3, 2011 at 7:26 PM, Dale Glaser <[hidden email]> wrote:

Hi all......I have ran a multiplicative model (group x test score) using zero-inflated Poisson (ZIP) in Mplus given the DV (# of fails). The interaction term is significant for the Poisson component of the model, and now we are trying to plot the interaction  This is where I am stuck.  I couldn't find a way by which to graph this result via Mplus so I used the GENLIN option in SPSS with the log link function for the Poisson model, and thought if I saved the predicted scores I could plot the two groups x moderator interaction (as you can with OLS regression when you graph a scatterplot of the moderator and outcome and use the subgroup fit line option to super-impose the predicted lines for each group).  However, as most of you would expect when I graph the predicted scores from GENLIN against the moderator and plot the two group fit lines, it only shows the two lines since it is a perfect fit for each group.  So, does anyone have any suggestions how to overlay the predicted line for subgroups over the scatterplot based on the results from a Poisson model?

 

thank you very much....dale

 

Dale Glaser, Ph.D.
Principal--Glaser Consulting
Lecturer/Adjunct Faculty--SDSU/USD/Alliant
Past-President, San Diego Chapter of
American Statistical Association
3115 4th Avenue
San Diego, CA 92103
phone: [phone]
fax: [phone]
email: [hidden email]
website: www.glaserconsult.com