SPSSX Discussion

Likert Scales Analysis

Classic

List

Threaded

8 messages Options

Steve-41

Likert Scales Analysis

Hi All,

A simplified explanation of the project I'm working on is that I'm
evaluating the association between the level of stress and depression in
individuals and the severity of and handicap due to their terminal illness.
Both the independent (disease characteristics) and dependent (psychological
status) variables are assessed via questionnaires involving Likert scales
(1-5 scaling). There are separate set of questions for stress and for
depression.

I've been going over all the posts related to analysis of Likert scales and
am even more confused than I started. :) Suggestions for analysis has been
to use CATPCA. Is this performed using syntax only or is there a
point-and-click method in the menubar? Also, can this be performed without
the SPSS Categories Add-on??

Would CATPCA be the best method to analyze this dataset? Should I try to
get some overall "psychological status score" and "disease severity score"
and try to perform a regression? I worry about doing the latter due to the
ordinal nature of the each of the scales.

Also, if someone knows of some good resource for Likert Scale Analysis,
please let me know. I've been browsing different websites for a week and am
no closer to an answer than before. Thanks to all in advance.

- Steve

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

John Fiedler

Re: Likert Scales Analysis - CATPCA

Hi Steve.

Yes, it is part of Categories. Not sure of the version history, maybe only
13 and later. However it may be available outside SPSS.

Yes, it can be done mostly in point and click though not every option is
available that way.

Yes, it is probably as good or better than any other approach.

Lastly, one neat thing is that you can see the scale values calculated for
each point of ordinal Likert scale which will confirm that it is not an
interval scale.

Good luck!

JOHN

----- Original Message -----
From: "Steve" <[hidden email]>
To: <[hidden email]>
Sent: Wednesday, August 13, 2008 6:22 PM
Subject: Likert Scales Analysis

> Hi All,
>
> A simplified explanation of the project I'm working on is that I'm
> evaluating the association between the level of stress and depression in
> individuals and the severity of and handicap due to their terminal
> illness.
> Both the independent (disease characteristics) and dependent
> (psychological
> status) variables are assessed via questionnaires involving Likert scales
> (1-5 scaling). There are separate set of questions for stress and for
> depression.
>
> I've been going over all the posts related to analysis of Likert scales
> and
> am even more confused than I started. :) Suggestions for analysis has
> been
> to use CATPCA. Is this performed using syntax only or is there a
> point-and-click method in the menubar? Also, can this be performed
> without
> the SPSS Categories Add-on??
>
> Would CATPCA be the best method to analyze this dataset? Should I try to
> get some overall "psychological status score" and "disease severity score"
> and try to perform a regression? I worry about doing the latter due to
> the
> ordinal nature of the each of the scales.
>
> Also, if someone knows of some good resource for Likert Scale Analysis,
> please let me know. I've been browsing different websites for a week and
> am
> no closer to an answer than before. Thanks to all in advance.
>
> - Steve
>
> =====================
> 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
>
>

Steve-41

Re: Likert Scales Analysis - CATPCA

On Thu, 14 Aug 2008 09:23:05 -0600, John Fiedler <[hidden email]> wrote:

>Hi Steve.
>
>Yes, it is part of Categories. Not sure of the version history, maybe only
>13 and later. However it may be available outside SPSS.
>

Thanks John. I have version 16 and I think it is part of "Categories".
What did you mean by "it may be available outside SPSS"?? As an addon? Or
you mean as a stand alone application?

>Yes, it can be done mostly in point and click though not every option is
>available that way.
>
I couldn't see any point and click options under Analyze so am assuming that
it would be visible only if I had Categories - which I'm afraid I do not.
Going to check up with the IT people but I think that wasn't included in our
license. :(

>Yes, it is probably as good or better than any other approach.
>
Any opinions about OLS regression?

>Lastly, one neat thing is that you can see the scale values calculated for
>each point of ordinal Likert scale which will confirm that it is not an
>interval scale.
>
That's good. I did not want to analyze it as an interval scale although the
person this data belongs to was quite confident that the spread would be
uniform between each Likert value. I know there's a lot of stuff out there
as to when one can and cannot assume that a Likert is an interval scale but
I wanted to stick to an ordinal scale.

Thanks again for your help. Again, if someone has an idea for good
resources (online or print) for Likert analysis I'd appreciate it. Thanks.

- Steve

>Good luck!
>
>JOHN
>
>
>
>----- Original Message -----
>From: "Steve" <[hidden email]>
>To: <[hidden email]>
>Sent: Wednesday, August 13, 2008 6:22 PM
>Subject: Likert Scales Analysis
>
>
>> Hi All,
>>
>> A simplified explanation of the project I'm working on is that I'm
>> evaluating the association between the level of stress and depression in
>> individuals and the severity of and handicap due to their terminal
>> illness.
>> Both the independent (disease characteristics) and dependent
>> (psychological
>> status) variables are assessed via questionnaires involving Likert scales
>> (1-5 scaling). There are separate set of questions for stress and for
>> depression.
>>
>> I've been going over all the posts related to analysis of Likert scales
>> and
>> am even more confused than I started. :) Suggestions for analysis has
>> been
>> to use CATPCA. Is this performed using syntax only or is there a
>> point-and-click method in the menubar? Also, can this be performed
>> without
>> the SPSS Categories Add-on??
>>
>> Would CATPCA be the best method to analyze this dataset? Should I try to
>> get some overall "psychological status score" and "disease severity score"
>> and try to perform a regression? I worry about doing the latter due to
>> the
>> ordinal nature of the each of the scales.
>>
>> Also, if someone knows of some good resource for Likert Scale Analysis,
>> please let me know. I've been browsing different websites for a week and
>> am
>> no closer to an answer than before. Thanks to all in advance.
>>
>> - Steve
>>
>> =====================
>> 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
>>
>>
>
>=====================
>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

Reutter, Alex

finding which commands go with which options in v16 [was Re: Likert Scales Analysis - CATPCA]

This may not be well known, so note that v16 includes help for all options, even the ones that are not licensed. So if you go to Help > Topics... then

A) If you know the name of the command or type of analysis but are not sure of the option:

On the Index tab, type catpca; by the time you get to "catp", it will have found the entry for CATPCA (command). Click on the "syntax chart" subentry and it will take you to the main page for the CATPCA syntax help. Right under the heading the text notes: "CATPCA is available in the Categories option."

Below the syntax chart (just above the release history) the text notes: "Syntax for the CATPCA command can be generated from the Categorical Principal Components Analysis (CATPCA) dialog." The link takes you to the main page for the GUI help.

Open the "To Obtain a Categorical Principal Components Analysis" collapsible block and that will show you how to navigate the menu system to get to the dialog. If that item doesn't appear on the menu, then you know that option isn't licensed.

There is also a "Show me" link just below the "Statistics and plots" paragraph that shows some examples of how to use the procedure (these are the same as those in the Categories PDF on the Manuals CD).

B) If you know the option but not the name of the command, or just want to browse the options, on the Contents tab, open up the Categories option and browse through the subfolders and topics.

The Search tab is also a good tool, especially for terms that fall through the indexing cracks, but it's a big help system and can return a lot of hits (i.e., searching on CATPCA isn't as useful as checking the index).

Cheers,
Alex

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Steve
Sent: Thursday, August 14, 2008 10:57 AM
To: [hidden email]
Subject: Re: Likert Scales Analysis - CATPCA

On Thu, 14 Aug 2008 09:23:05 -0600, John Fiedler <[hidden email]> wrote:

>Hi Steve.
>
>Yes, it is part of Categories. Not sure of the version history, maybe only
>13 and later. However it may be available outside SPSS.
>

Thanks John. I have version 16 and I think it is part of "Categories".
What did you mean by "it may be available outside SPSS"?? As an addon? Or
you mean as a stand alone application?

>Yes, it can be done mostly in point and click though not every option is
>available that way.
>
I couldn't see any point and click options under Analyze so am assuming that
it would be visible only if I had Categories - which I'm afraid I do not.
Going to check up with the IT people but I think that wasn't included in our
license. :(

>Yes, it is probably as good or better than any other approach.
>
Any opinions about OLS regression?

>Lastly, one neat thing is that you can see the scale values calculated for
>each point of ordinal Likert scale which will confirm that it is not an
>interval scale.
>
That's good. I did not want to analyze it as an interval scale although the
person this data belongs to was quite confident that the spread would be
uniform between each Likert value. I know there's a lot of stuff out there
as to when one can and cannot assume that a Likert is an interval scale but
I wanted to stick to an ordinal scale.

Thanks again for your help. Again, if someone has an idea for good
resources (online or print) for Likert analysis I'd appreciate it. Thanks.

- Steve

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

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

John Fiedler

Re: Likert Scales Analysis - CATPCA

In reply to this post by Steve-41

CATPCA, like Correspondence Analysis, was developed out of Leiden University
by, I think, the same group. Since correspondence analysis was available
long before its inclusion into SPSS, perhaps CATPCA is similarly available.

I really like it. I think SPSS offers a free 30 day trial for its add on
modules. Give it a try!

JOHN

----- Original Message -----
From: "Steve" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, August 14, 2008 9:56 AM
Subject: Re: Likert Scales Analysis - CATPCA

> On Thu, 14 Aug 2008 09:23:05 -0600, John Fiedler <[hidden email]>
> wrote:
>
>>Hi Steve.
>>
>>Yes, it is part of Categories. Not sure of the version history, maybe only
>>13 and later. However it may be available outside SPSS.
>>
>
> Thanks John. I have version 16 and I think it is part of "Categories".
> What did you mean by "it may be available outside SPSS"?? As an addon? Or
> you mean as a stand alone application?
>
>
>>Yes, it can be done mostly in point and click though not every option is
>>available that way.
>>
> I couldn't see any point and click options under Analyze so am assuming
> that
> it would be visible only if I had Categories - which I'm afraid I do not.
> Going to check up with the IT people but I think that wasn't included in
> our
> license. :(
>
>
>>Yes, it is probably as good or better than any other approach.
>>
> Any opinions about OLS regression?
>
>
>>Lastly, one neat thing is that you can see the scale values calculated for
>>each point of ordinal Likert scale which will confirm that it is not an
>>interval scale.
>>
> That's good. I did not want to analyze it as an interval scale although
> the
> person this data belongs to was quite confident that the spread would be
> uniform between each Likert value. I know there's a lot of stuff out
> there
> as to when one can and cannot assume that a Likert is an interval scale
> but
> I wanted to stick to an ordinal scale.
>
> Thanks again for your help. Again, if someone has an idea for good
> resources (online or print) for Likert analysis I'd appreciate it. Thanks.
>
> - Steve
>
>
>
>
>>Good luck!
>>
>>JOHN
>>
>>
>>
>>----- Original Message -----
>>From: "Steve" <[hidden email]>
>>To: <[hidden email]>
>>Sent: Wednesday, August 13, 2008 6:22 PM
>>Subject: Likert Scales Analysis
>>
>>
>>> Hi All,
>>>
>>> A simplified explanation of the project I'm working on is that I'm
>>> evaluating the association between the level of stress and depression in
>>> individuals and the severity of and handicap due to their terminal
>>> illness.
>>> Both the independent (disease characteristics) and dependent
>>> (psychological
>>> status) variables are assessed via questionnaires involving Likert
>>> scales
>>> (1-5 scaling). There are separate set of questions for stress and for
>>> depression.
>>>
>>> I've been going over all the posts related to analysis of Likert scales
>>> and
>>> am even more confused than I started. :) Suggestions for analysis has
>>> been
>>> to use CATPCA. Is this performed using syntax only or is there a
>>> point-and-click method in the menubar? Also, can this be performed
>>> without
>>> the SPSS Categories Add-on??
>>>
>>> Would CATPCA be the best method to analyze this dataset? Should I try
>>> to
>>> get some overall "psychological status score" and "disease severity
>>> score"
>>> and try to perform a regression? I worry about doing the latter due to
>>> the
>>> ordinal nature of the each of the scales.
>>>
>>> Also, if someone knows of some good resource for Likert Scale Analysis,
>>> please let me know. I've been browsing different websites for a week
>>> and
>>> am
>>> no closer to an answer than before. Thanks to all in advance.
>>>
>>> - Steve
>>>
>>> =====================
>>> 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
>>>
>>>
>>
>>=====================
>>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
>
> =====================
> 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
>
>

Daniel Robertson

Re: Likert Scales Analysis - CATPCA

FWIW... if I'm remembering the Gifi System details correctly, I believe
that CATPCA is in theory a special case of HOMALS (Homogeneity Analysis),
e.g., HOMALS with certain constraints. And, there is a HOMALS package
available for R.

Dan R.

> CATPCA, like Correspondence Analysis, was developed out of Leiden
> University
> by, I think, the same group. Since correspondence analysis was available
> long before its inclusion into SPSS, perhaps CATPCA is similarly
> available.
>
> I really like it. I think SPSS offers a free 30 day trial for its add on
> modules. Give it a try!
>
> JOHN
>
>
>
>
> ----- Original Message -----
> From: "Steve" <[hidden email]>
> To: <[hidden email]>
> Sent: Thursday, August 14, 2008 9:56 AM
> Subject: Re: Likert Scales Analysis - CATPCA
>
>
>> On Thu, 14 Aug 2008 09:23:05 -0600, John Fiedler <[hidden email]>
>> wrote:
>>
>>>Hi Steve.
>>>
>>>Yes, it is part of Categories. Not sure of the version history, maybe
>>> only
>>>13 and later. However it may be available outside SPSS.
>>>
>>
>> Thanks John. I have version 16 and I think it is part of "Categories".
>> What did you mean by "it may be available outside SPSS"?? As an addon?
>> Or
>> you mean as a stand alone application?
>>
>>

Reutter, Alex

Re: Likert Scales Analysis - CATPCA

It's the reverse. The HOMALS "model" is essentially a special case of what can be fit by PRINCALS, which in turn is a special case of OVERALS. For that matter, the CATREG model is essentially a special case of OVERALS and CORRESPONDENCE a special case of HOMALS.

In Categories, PRINCALS has since been replaced by CATPCA, and HOMALS has been replaced by MULTIPLE CORRESPONDENCE. These new procedures offer considerably more functionality than their predecessors, but fit the same "models".

Alex

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Daniel Robertson
Sent: Thursday, August 14, 2008 8:38 PM
To: [hidden email]
Subject: Re: Likert Scales Analysis - CATPCA

FWIW... if I'm remembering the Gifi System details correctly, I believe
that CATPCA is in theory a special case of HOMALS (Homogeneity Analysis),
e.g., HOMALS with certain constraints. And, there is a HOMALS package
available for R.

Dan R.

--
Daniel Robertson
Senior Research and Planning Associate
Institutional Research and Planning
Cornell University / irp.cornell.edu

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

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

Reutter, Alex

Re: Likert Scales Analysis - CATPCA

In reply to this post by John Fiedler

Clarification/correction:

* The HOMALS /model/ is less "restricted" than the CATPCA model in the sense that no restrictions are placed upon the quantifications, while quantifications in the CATPCA model may be restricted to be of rank-one and to be ordinal, numeric, etc.

* The CATPCA /procedure/ is less "restricted" than the HOMALS procedure in the sense that it gives you this ability to place rank-one and scaling level restrictions on the quantifications

Dan, you're absolutely correct that the CATPCA model is the HOMALS model with constraints; I was thinking in terms of the CATPCA procedure's ability to fit a HOMALS solution, but not the reverse, and made unfortunate use of the word "model".

Cheers,
Alex

-----Original Message-----
From: Reutter, Alex
Sent: Friday, August 15, 2008 9:27 AM
To: [hidden email]
Subject: RE: Likert Scales Analysis - CATPCA

It's the reverse. The HOMALS "model" is essentially a special case of what can be fit by PRINCALS, which in turn is a special case of OVERALS. For that matter, the CATREG model is essentially a special case of OVERALS and CORRESPONDENCE a special case of HOMALS.

In Categories, PRINCALS has since been replaced by CATPCA, and HOMALS has been replaced by MULTIPLE CORRESPONDENCE. These new procedures offer considerably more functionality than their predecessors, but fit the same "models".

Alex

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Daniel Robertson
Sent: Thursday, August 14, 2008 8:38 PM
To: [hidden email]
Subject: Re: Likert Scales Analysis - CATPCA

FWIW... if I'm remembering the Gifi System details correctly, I believe
that CATPCA is in theory a special case of HOMALS (Homogeneity Analysis),
e.g., HOMALS with certain constraints. And, there is a HOMALS package
available for R.

Dan R.