SPSSX Discussion

Print correlations only?

Classic

List

Threaded

12 messages Options

Mario Giesel

Print correlations only?

* Hello, SPSS friends.
* I've got a question on correlations output.
* Let's say these are the data.

DATA LIST FREE (" ")/v1 v2 v3.
BEGIN DATA
1 2 3
2 1 1
3 1 3
2 3 2
END DATA.

* Default correlations are calculated.

CORRELATIONS
  /VARIABLES=v1 v2 v3
  /PRINT=TWOTAIL NOSIG
  /MISSING=PAIRWISE .

* Question 1: How can I suppress rows "significance" and "N"? (SPSS 15).
* Question 2: Is it possible to suppress columns "v2" and "v3" as well - if I'm interested in but the first variables' correlations?.

* Thanks for any help.
* Mario.

Mario Giesel
Munich, Germany

Marta Garcia-Granero

Re: Print correlations only?

Hi Mario:

Regarding question 1, significance and N are part of the output, and can't be suppressed, unless you save the correlation matrix as a dataset, delete the unwanted rows and print it again. Workaround: You can use FACTOR to print a correlation matrix without N or Sigs:

FACTOR
/VARIABLES v1 v2 v3
/MISSING LISTWISE
/ANALYSIS v1 v2 v3
/PRINT CORRELATION
/ROTATION NOROTATE
/METHOD=CORRELATION .

Question 2 I don't understand clearly, but, assuming you only want the correlation of V1 with V2, and V1 with V3, but not the correlation between V2&V3, this modification of the correlation command will do it:

CORRELATIONS
/VARIABLES=v2 v3 WITH v1
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

Unfortunately, both solutions can't be combined (you can either have the correlation matrix without N&Sigs, or the non square table, with some correlation coefficients suppressed, but not both).

HTH,
Marta GG

El 16/05/2012 11:53, Mario Giesel escribió:

* Hello, SPSS friends.
* I've got a question on correlations output.
* Let's say these are the data.

DATA LIST FREE (" ")/v1 v2 v3.
BEGIN DATA
1 2 3
2 1 1
3 1 3
2 3 2
END DATA.

* Default correlations are calculated.

CORRELATIONS
/VARIABLES=v1 v2 v3
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

* Question 1: How can I suppress rows "significance" and "N"? (SPSS 15).
* Question 2: Is it possible to suppress columns "v2" and "v3" as well - if I'm interested in but the first variables' correlations?.

* Thanks for any help.

* Mario.

Mario Giesel

Re: Print correlations only?

Thanks Marta, 
 you understood correctly. I tried the factor workaround and this is a good solution for casual purposes. 
I can delete columns manually after that. Thanks for that.

However - addressed to the other list users - for automated reports a compact display of correlations v2 v3 WITH v1 
would be desirable. Maybe MATRIX can help somehow? 

Thanks,
 Mario

    Von: Marta García-Granero <[hidden email]>
 An: [hidden email] 
 Gesendet: 12:49 Mittwoch, 16.Mai 2012
 Betreff: Re: Print correlations only?
  
    Hi Mario:

    Regarding question 1, significance and N are part of the output, and
    can't be suppressed, unless you save the correlation matrix as a
    dataset, delete the unwanted rows and print it again. Workaround:
    You can use FACTOR to print a correlation matrix without N or Sigs:

    FACTOR

      /VARIABLES v1 v2 v3

      /MISSING LISTWISE

      /ANALYSIS v1 v2 v3

      /PRINT CORRELATION

      /ROTATION NOROTATE

      /METHOD=CORRELATION .

    Question 2 I don't understand clearly, but, assuming you only want
    the correlation of V1 with V2, and V1 with V3, but not the
    correlation between V2&V3, this modification of the correlation
    command will do it:

    CORRELATIONS

      /VARIABLES=v2 v3 WITH v1

      /PRINT=TWOTAIL NOSIG

      /MISSING=PAIRWISE .

    Unfortunately, both solutions can't be combined (you can either have
    the correlation matrix without N&Sigs, or the non square table,
    with some correlation coefficients suppressed, but not both).

    HTH,

    Marta GG

    El 16/05/2012 11:53, Mario Giesel escribió:
    * Hello, SPSS friends.

          * I've got a question on correlations output.

          * Let's say these are the data.

          DATA LIST FREE (" ")/v1 v2 v3.

          BEGIN DATA

          1 2 3

          2 1 1

          3 1 3

          2 3 2

          END DATA.

          * Default correlations are calculated.

          CORRELATIONS

            /VARIABLES=v1 v2 v3

            /PRINT=TWOTAIL NOSIG

            /MISSING=PAIRWISE .

          * Question 1: How can I suppress rows "significance" and "N"?
          (SPSS 15).

          * Question 2: Is it possible to suppress columns "v2" and "v3"
          as well - if I'm interested in but the first variables'
          correlations?.

          * Thanks for any help.
* Mario.

Mario Giesel
Munich, Germany

Jon K Peck

Re: Print correlations only?

In reply to this post by Mario Giesel

There is no way to suppress the N and sig rows in CORRELATIONS. There is a custom dialog, Format Correlation Matrix, that can do this and much more, but it will not work with old versions of SPSS. However, you could write a script or autoscript in SaxBasic to do these tasks.

Regards,

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
new phone: 720-342-5621

From: Mario Giesel <[hidden email]>
To: [hidden email]
Date: 05/16/2012 03:57 AM
Subject: [SPSSX-L] Print correlations only?
Sent by: "SPSSX(r) Discussion" <[hidden email]>

* Hello, SPSS friends.
* I've got a question on correlations output.
* Let's say these are the data.

DATA LIST FREE (" ")/v1 v2 v3.
BEGIN DATA
1 2 3
2 1 1
3 1 3
2 3 2
END DATA.

* Default correlations are calculated.

CORRELATIONS
/VARIABLES=v1 v2 v3
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

* Question 1: How can I suppress rows "significance" and "N"? (SPSS 15).
* Question 2: Is it possible to suppress columns "v2" and "v3" as well - if I'm interested in but the first variables' correlations?.

* Thanks for any help.
* Mario.

Gerard van Meurs-2

Re: Print correlations only?

In reply to this post by Mario Giesel

Hi Mario,

Perhaps this workaround:

DATA LIST FREE (" ")/v1 v2 v3.

BEGIN DATA

1 2 3

2 1 1

3 1 3

2 3 2

END DATA.

CORRELATIONS

/VARIABLES=v1 v2 v3

/matrix = out( *).

select if (rowtype_ = 'CORR').

list cases

/var = varname_ v1.

HTH, Gerard van Meurs

Van: SPSSX(r) Discussion [mailto:[hidden email]] Namens Mario Giesel
Verzonden: woensdag 16 mei 2012 14:41
Aan: [hidden email]
Onderwerp: Re: Print correlations only?

Thanks Marta,

you understood correctly. I tried the factor workaround and this is a good solution for casual purposes.

I can delete columns manually after that. Thanks for that.

However - addressed to the other list users - for automated reports a compact display of correlations v2 v3 WITH v1

would be desirable. Maybe MATRIX can help somehow?

Thanks,

Mario

Von: Marta García-Granero <[hidden email]>
An: [hidden email]
Gesendet: 12:49 Mittwoch, 16.Mai 2012
Betreff: Re: Print correlations only?

* Mario.

Bruce Weaver

Re: Print correlations only?

Administrator

In reply to this post by Mario Giesel

You can also get the correlations via RELIABILITY. E.g.,

RELIABILITY
/VARIABLES=v1 v2 v3
/SCALE('ALL VARIABLES') ALL
/STATISTICS=CORR.

This gives the complete 3x3 matrix, so you'll have to delete the last two columns to get what you want.

HTH.

Mario Giesel wrote

Thanks Marta,

you understood correctly. I tried the factor workaround and this is a good solution for casual purposes.

I can delete columns manually after that. Thanks for that.

However - addressed to the other list users - for automated reports a compact display of correlations v2 v3 WITH v1

would be desirable. Maybe MATRIX can help somehow?

Thanks,
Mario

________________________________
Von: Marta García-Granero <[hidden email]>
An: [hidden email]
Gesendet: 12:49 Mittwoch, 16.Mai 2012
Betreff: Re: Print correlations only?

Hi Mario:

Regarding question 1, significance and N are part of the output, and
can't be suppressed, unless you save the correlation matrix as a
dataset, delete the unwanted rows and print it again. Workaround:
You can use FACTOR to print a correlation matrix without N or Sigs:

FACTOR
/VARIABLES v1 v2 v3
/MISSING LISTWISE
/ANALYSIS v1 v2 v3
/PRINT CORRELATION
/ROTATION NOROTATE
/METHOD=CORRELATION .

Question 2 I don't understand clearly, but, assuming you only want
the correlation of V1 with V2, and V1 with V3, but not the
correlation between V2&V3, this modification of the correlation
command will do it:

CORRELATIONS
/VARIABLES=v2 v3 WITH v1
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

Unfortunately, both solutions can't be combined (you can either have
the correlation matrix without N&Sigs, or the non square table,
with some correlation coefficients suppressed, but not both).

HTH,
Marta GG

El 16/05/2012 11:53, Mario Giesel escribió:
* Hello, SPSS friends.
>* I've got a question on correlations output.
>* Let's say these are the data.
>
>DATA LIST FREE (" ")/v1 v2 v3.
>BEGIN DATA
>1 2 3
>2 1 1
>3 1 3
>2 3 2
>END DATA.
>
>* Default correlations are calculated.
>
>CORRELATIONS
> /VARIABLES=v1 v2 v3
> /PRINT=TWOTAIL NOSIG
> /MISSING=PAIRWISE .
>
>* Question 1: How can I suppress rows "significance" and "N"?
(SPSS 15).
>* Question 2: Is it possible to suppress columns "v2" and "v3"
as well - if I'm interested in but the first variables'
correlations?.
>
>* Thanks for any help.
>* Mario.

--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).

Bruce Weaver

Re: Print correlations only?

Administrator

Another way:

REGRESSION
/STATISTICS ZPP
/DEPENDENT v1
/METHOD=ENTER v2 v3.

The two correlations you want are in the zero-order column.

Bruce Weaver wrote

You can also get the correlations via RELIABILITY. E.g.,

RELIABILITY
/VARIABLES=v1 v2 v3
/SCALE('ALL VARIABLES') ALL
/STATISTICS=CORR.

This gives the complete 3x3 matrix, so you'll have to delete the last two columns to get what you want.

HTH.

Mario Giesel wrote

Thanks Marta,

you understood correctly. I tried the factor workaround and this is a good solution for casual purposes.

I can delete columns manually after that. Thanks for that.

However - addressed to the other list users - for automated reports a compact display of correlations v2 v3 WITH v1

would be desirable. Maybe MATRIX can help somehow?

Thanks,
Mario

________________________________
Von: Marta García-Granero <[hidden email]>
An: [hidden email]
Gesendet: 12:49 Mittwoch, 16.Mai 2012
Betreff: Re: Print correlations only?

Hi Mario:

Regarding question 1, significance and N are part of the output, and
can't be suppressed, unless you save the correlation matrix as a
dataset, delete the unwanted rows and print it again. Workaround:
You can use FACTOR to print a correlation matrix without N or Sigs:

FACTOR
/VARIABLES v1 v2 v3
/MISSING LISTWISE
/ANALYSIS v1 v2 v3
/PRINT CORRELATION
/ROTATION NOROTATE
/METHOD=CORRELATION .

Question 2 I don't understand clearly, but, assuming you only want
the correlation of V1 with V2, and V1 with V3, but not the
correlation between V2&V3, this modification of the correlation
command will do it:

CORRELATIONS
/VARIABLES=v2 v3 WITH v1
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

Unfortunately, both solutions can't be combined (you can either have
the correlation matrix without N&Sigs, or the non square table,
with some correlation coefficients suppressed, but not both).

HTH,
Marta GG

El 16/05/2012 11:53, Mario Giesel escribió:
* Hello, SPSS friends.
>* I've got a question on correlations output.
>* Let's say these are the data.
>
>DATA LIST FREE (" ")/v1 v2 v3.
>BEGIN DATA
>1 2 3
>2 1 1
>3 1 3
>2 3 2
>END DATA.
>
>* Default correlations are calculated.
>
>CORRELATIONS
> /VARIABLES=v1 v2 v3
> /PRINT=TWOTAIL NOSIG
> /MISSING=PAIRWISE .
>
>* Question 1: How can I suppress rows "significance" and "N"?
(SPSS 15).
>* Question 2: Is it possible to suppress columns "v2" and "v3"
as well - if I'm interested in but the first variables'
correlations?.
>
>* Thanks for any help.
>* Mario.

Marta Garcia-Granero

Re: Print correlations only?

In reply to this post by Mario Giesel

El 16/05/2012 14:41, Mario Giesel escribió:

Thanks Marta,

you understood correctly. I tried the factor workaround and this is a good solution for casual purposes.

I can delete columns manually after that. Thanks for that.

However - addressed to the other list users - for automated reports a compact display of correlations v2 v3 WITH v1

would be desirable. Maybe MATRIX can help somehow?

You can work it from here:

MATRIX.
GET data /VAR=v1 v2 v3 /NAMES=vnames.
COMPUTE k=NCOL(data).
COMPUTE n=NROW(data).
COMPUTE datadev=data - MAKE(n,n,1)*data/n.
COMPUTE CovMat=T(datadev)*datadev/(n-1).
COMPUTE rtdiagS=SQRT(DIAG(CovMat)).
COMPUTE rtD=MDIAG(rtdiagS).
COMPUTE invrtD=INV(rtD).
COMPUTE CorrMat=invrtD*CovMat*invrtD.
PRINT CorrMat.
END MATRIX.

You can modify it to select the columns you want to print.

HTH,
Marta

Thanks,

Mario

Von: Marta García-Granero [hidden email]
An: [hidden email]
Gesendet: 12:49 Mittwoch, 16.Mai 2012
Betreff: Re: Print correlations only?

Hi Mario:

Regarding question 1, significance and N are part of the output, and can't be suppressed, unless you save the correlation matrix as a dataset, delete the unwanted rows and print it again. Workaround: You can use FACTOR to print a correlation matrix without N or Sigs:

FACTOR
/VARIABLES v1 v2 v3
/MISSING LISTWISE
/ANALYSIS v1 v2 v3
/PRINT CORRELATION
/ROTATION NOROTATE
/METHOD=CORRELATION .

Question 2 I don't understand clearly, but, assuming you only want the correlation of V1 with V2, and V1 with V3, but not the correlation between V2&V3, this modification of the correlation command will do it:

CORRELATIONS
/VARIABLES=v2 v3 WITH v1
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

Unfortunately, both solutions can't be combined (you can either have the correlation matrix without N&Sigs, or the non square table, with some correlation coefficients suppressed, but not both).

HTH,
Marta GG

El 16/05/2012 11:53, Mario Giesel escribió:

* Hello, SPSS friends.
* I've got a question on correlations output.
* Let's say these are the data.

DATA LIST FREE (" ")/v1 v2 v3.
BEGIN DATA
1 2 3
2 1 1
3 1 3
2 3 2
END DATA.

* Default correlations are calculated.

CORRELATIONS
/VARIABLES=v1 v2 v3
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE .

* Question 1: How can I suppress rows "significance" and "N"? (SPSS 15).
* Question 2: Is it possible to suppress columns "v2" and "v3" as well - if I'm interested in but the first variables' correlations?.

* Thanks for any help.

* Mario.

David Marso

Re: Print correlations only?

Administrator

In reply to this post by Marta Garcia-Granero

Either CORR using WITH then script Statistics into the layer.
or FACTOR.
or
MATRIX.
GET X / VAR v1 TO v3/ FILE */NAMES=LBL.
COMPUTE CCP= SSCP(X)-T(CSUM(X))*CSUM(X)/NROW(X).
COMPUTE D=GINV(MDIAG(SQRT(DIAG(CCP)))).
COMPUTE R=D*CCP*D.
COMPUTE X1=LBL(1).
PRINT R(:,1)/RNAMES=LBL/CNAMES=X1.
END MATRIX.

Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me.
---
"Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis."
Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?"

Mario Giesel

Re: Print correlations only?

Thanks to Marta, Jon, Gerard, Bruce and David for your work and comments!
I'm curious to explore the MATRIX solutions a bit more so that I can understand what's being done there.
They work fine, however. I also like Gerards pretty solution.

For quick views I'm probably going to work with this FACTOR approach:

DEFINE !cor(!POS !CMDEND)
FACTOR
  /VARIABLES !1
  /ANALYSIS !1
  /PRINT CORRELATION
  /ROTATION NOROTATE.
!ENDDEFINE.
!cor v1 v2 v3.

Best wishes,
 Mario

   
 Von: David Marso <[hidden email]>
 An: [hidden email] 
 Gesendet: 18:48 Mittwoch, 16.Mai 2012
 Betreff: Re: Print correlations only?
  

Either CORR using WITH then script Statistics into the layer.
or FACTOR.
or
MATRIX.
GET X / VAR v1 TO v3/ FILE */NAMES=LBL.
COMPUTE CCP= SSCP(X)-T(CSUM(X))*CSUM(X)/NROW(X).
COMPUTE D=GINV(MDIAG(SQRT(DIAG(CCP)))).
COMPUTE R=D*CCP*D.
COMPUTE X1=LBL(1).
PRINT R(:,1)/RNAMES=LBL/CNAMES=X1.
END MATRIX.


--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Print-correlations-only-tp5710657p5710997.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
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

Mario Giesel
Munich, Germany

Jon K Peck

Re: Print correlations only?

You could clean the output up a little by wrapping the FACTOR command in this.
oms /exceptif subtypes = ['Correlation Matrix'] /destination viewer=no.
FACTOR ...
omsend.

Jon Peck (no "h") aka Kim
Senior Software Engineer, IBM
[hidden email]
new phone: 720-342-5621

From: Mario Giesel <[hidden email]>
To: [hidden email]
Date: 05/17/2012 10:50 AM
Subject: Re: [SPSSX-L] Print correlations only?
Sent by: "SPSSX(r) Discussion" <[hidden email]>

Thanks to Marta, Jon, Gerard, Bruce and David for your work and comments!
I'm curious to explore the MATRIX solutions a bit more so that I can understand what's being done there.
They work fine, however. I also like Gerards pretty solution.

For quick views I'm probably going to work with this FACTOR approach:

DEFINE !cor(!POS !CMDEND)
FACTOR
/VARIABLES !1
/ANALYSIS !1
/PRINT CORRELATION
/ROTATION NOROTATE.
!ENDDEFINE.
!cor v1 v2 v3.

Best wishes,
Mario

Von: David Marso <[hidden email]>
An: [hidden email]
Gesendet: 18:48 Mittwoch, 16.Mai 2012
Betreff: Re: Print correlations only?

Either CORR using WITH then script Statistics into the layer.
or FACTOR.
or
MATRIX.
GET X / VAR v1 TO v3/ FILE */NAMES=LBL.
COMPUTE CCP= SSCP(X)-T(CSUM(X))*CSUM(X)/NROW(X).
COMPUTE D=GINV(MDIAG(SQRT(DIAG(CCP)))).
COMPUTE R=D*CCP*D.
COMPUTE X1=LBL(1).
PRINT R(:,1)/RNAMES=LBL/CNAMES=X1.
END MATRIX.

--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Print-correlations-only-tp5710657p5710997.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (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

Mario Giesel

Re: Print correlations only?

Great, Jon!

I never was aware of that OMS command.
Pretty method to suppress viewer output.

Thanks,
 Mario

    Von: Jon K Peck <[hidden email]>
 An: Mario Giesel <[hidden email]> 
CC:
 [hidden email] 
 Gesendet: 18:59 Donnerstag, 17.Mai 2012
 Betreff: Re: [SPSSX-L] Print correlations only?
  
You could clean the output up a little
by wrapping the FACTOR command in this.

oms /exceptif subtypes = ['Correlation
Matrix'] /destination viewer=no.

FACTOR ...

omsend.

Jon Peck (no "h") aka Kim

Senior Software Engineer, IBM

[hidden email]

new phone: 720-342-5621

From:      
 Mario Giesel <[hidden email]>

To:      
 [hidden email]

Date:      
 05/17/2012 10:50 AM

Subject:    
   Re: [SPSSX-L]
Print correlations only?

Sent by:    
   "SPSSX(r)
Discussion" <[hidden email]>

Thanks to Marta, Jon, Gerard, Bruce and
David for your work and comments!

I'm curious to explore the MATRIX solutions a bit more so that I can understand
what's being done there.

They work fine, however. I also like Gerards pretty solution.

For quick views I'm probably going to work with this FACTOR approach:

DEFINE !cor(!POS !CMDEND)

FACTOR

  /VARIABLES !1

  /ANALYSIS !1

  /PRINT CORRELATION

  /ROTATION NOROTATE.

!ENDDEFINE.

!cor v1 v2 v3.

Best wishes,

 Mario

Von: David Marso <[hidden email]>

An: [hidden email] 

Gesendet: 18:48 Mittwoch, 16.Mai 2012

Betreff: Re: Print correlations only?

Either CORR using WITH then script Statistics into the layer.

or FACTOR.

or

MATRIX.

GET X / VAR v1 TO v3/ FILE */NAMES=LBL.

COMPUTE CCP= SSCP(X)-T(CSUM(X))*CSUM(X)/NROW(X).

COMPUTE D=GINV(MDIAG(SQRT(DIAG(CCP)))).

COMPUTE R=D*CCP*D.

COMPUTE X1=LBL(1).

PRINT R(:,1)/RNAMES=LBL/CNAMES=X1.

END MATRIX.

--

View this message in context: http://spssx-discussion.1045642.n5.nabble.com/Print-correlations-only-tp5710657p5710997.html

Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================

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

Mario Giesel
Munich, Germany