Quantcast

Check All That Apply

classic Classic list List threaded Threaded
10 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Check All That Apply

Paige Ransford

Dear Colleagues,

 

  I am currently in the process of constructing a SPSS file for a survey that has several questions that are in a “check all that apply” format. Several of these questions have 10+options, so it would be quite inconvenient to create dummy variables for each option. Is there any way to format a “check all that apply” code in SPSS? Any suggestions on other ways to format other than dummy coding?

 

Thanks,

 

Paige Ransford

Senior Researcher

The Center for Women in Politics and Public Policy

University of Massachusetts Boston

Boston, MA  02125

617-287-7407

www.mccormack.umb.edu/centers/cwppp

 

 

 

 

 

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

Hector Maletta

The usual way is “multiple dichotomies” i.e. dummies. Another alternative is “multiple responses”. In this case, you establish first what is the maximum number of simultaneous choices per person (perhaps the individual choosing the largest number chose 7 out of the 10 options presented). If the maximum number of choices per person is k, you create k variables with values 1 to 10 each, each coding a different choice for that particular question. A third alternative is to combine the numerous response patters into a single figure; suppose you have 5 such questions; a person choosing the first alternative in all questions would have a value 11111. Better still, you may put a first figure (say, another 1) without any meaning, just to enable the introduction of zeroes if any question is left unanswered: a person choosing 1 at questions 2 and 5, and leaving the rest unanswered, would be represented by 101001.

These alternatives are viable only if the maximum number of choices per person is not too high, or if the number of response patterns is not extremely large. In one case I handled some time ago, I used the third alternative, found that a few patterns accounted for about 90% of the cases, and grouped the remaining (many) patterns into a couple of categories, to avoid having an excessive number of patterns in my tables.

 

Hector

 


From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Paige Ransford
Sent: 26 January 2010 17:12
To: [hidden email]
Subject: Check All That Apply

 

Dear Colleagues,

 

  I am currently in the process of constructing a SPSS file for a survey that has several questions that are in a “check all that apply” format. Several of these questions have 10+options, so it would be quite inconvenient to create dummy variables for each option. Is there any way to format a “check all that apply” code in SPSS? Any suggestions on other ways to format other than dummy coding?

 

Thanks,

 

Paige Ransford

Senior Researcher

The Center for Women in Politics and Public Policy

University of Massachusetts Boston

Boston, MA  02125

617-287-7407

www.mccormack.umb.edu/centers/cwppp

 

 

 

 

 

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

Bruce Weaver
Administrator
In reply to this post by Paige Ransford
Paige Ransford wrote
Dear Colleagues,

  I am currently in the process of constructing a SPSS file for a survey
that has several questions that are in a "check all that apply" format.
Several of these questions have 10+options, so it would be quite
inconvenient to create dummy variables for each option. Is there any way
to format a "check all that apply" code in SPSS? Any suggestions on
other ways to format other than dummy coding?

Thanks,
I fail to see what's so inconvenient about it, unless you mean that you are setting up the variables by hand in the variable view.  If that's what you mean, try doing it with syntax instead.  E.g., suppose item 7 is a check-all-that-apply with 10 options.

numeric q7.1 to q7.10 (f1.0).
exe.
var lab
 q7.1 'Variable label for 7.1'
 q7.2 'Variable label for 7.2'
 etc
 q7.10 'Variable label for 7.0'
.
val lab q7.1 to q7.10
 0 'No'
 1 'Yes'
.

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

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

Bruce Weaver
Administrator
In reply to this post by Hector Maletta
Hector Maletta wrote
The usual way is "multiple dichotomies" i.e. dummies. Another alternative is
"multiple responses". In this case, you establish first what is the maximum
number of simultaneous choices per person (perhaps the individual choosing
the largest number chose 7 out of the 10 options presented). If the maximum
number of choices per person is k, you create k variables with values 1 to
10 each, each coding a different choice for that particular question. A
third alternative is to combine the numerous response patters into a single
figure; suppose you have 5 such questions; a person choosing the first
alternative in all questions would have a value 11111. Better still, you may
put a first figure (say, another 1) without any meaning, just to enable the
introduction of zeroes if any question is left unanswered: a person choosing
1 at questions 2 and 5, and leaving the rest unanswered, would be
represented by 101001.

These alternatives are viable only if the maximum number of choices per
person is not too high, or if the number of response patterns is not
extremely large. In one case I handled some time ago, I used the third
alternative, found that a few patterns accounted for about 90% of the cases,
and grouped the remaining (many) patterns into a couple of categories, to
avoid having an excessive number of patterns in my tables.


Hector
I was thinking about Hector's "binary coding" method too--someone suggested it for a similar problem posted to comp.soft-sys.stat.spss not too long ago.  But rather than including the meaningless first digit, I think one could use an N-format.  E.g., for a 5-option item:

format item (n5.0).  

If the 2nd and 5th items are checked, it should display as:  01001

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

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Print Label and Variable name in Factor

Granaas, Michael
In reply to this post by Bruce Weaver
I have a large batch of questions, cleverly named q1, q2, q3, etc.  The actual question appears in the data set as the question label. 
 
When I run principal components analysis and print out the rotated component matrix I get all the question labels.   What I want is the label and the variable name.  Preferably with a format like:  Label (q3).
 
Any suggestions?
 
Thanks,
 
Michael

****************************************************
Michael Granaas             [hidden email]
Assoc. Prof.                Phone: 605 677 5295
Dept. of Psychology         FAX:  605 677 3195
University of South Dakota
414 E. Clark St.
Vermillion, SD 57069
*****************************************************

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

John F Hall
In reply to this post by Paige Ransford
Paige
 
I used to do this sort of thing for a living, so if you send me an example of your data or an extract from your questionnaire I can advise you on the best way to proceed. 
 
What format are your raw data?
 
You have a choice of n of variables = max n of replies or of using say 3 out of n.  Latter is more efficient, but former catches everything.  Using binary is one way eg 0010010011 etc, but tabulation then sometimes has to be modified.  There's a short introduction to the SPSS command mult response in Analysing Multiple Response with SPSS: an introducti­on

This explains the history of how questions capable of more than one response (eg "Tick as many as apply") are handled in computer analysis. Describes various ways this is done in SPSS and discusses best practice with examples of frequency counts and crosstabulations from real surveys, in particular the fifth form survey. [NB The output examples in this section are from older versions of SPSS. Later versions now have full pivot tables for the output and examples of the same tables are given in the appendix.]

There's a worked example from the 2002 European Social Survey in my Old Dog, Old Tricks paper to ASSESS (SPSS Users) York Univ. (UK) Dec 2006

Old Dog, Old Tricks: Using SPSS Syntax to Avoid the Mouse Trap (pp27 - 35)

and the 3rd slide show

Part 3 of slide presentation to accompany main paper

Critical discussion of problems posed for secondary research by the SPSS saved file for the European Social Survey 2002, covering use of CAPI questionnaire and awkward conventions for defining variable names and labels. Offers suggestions for improvements (some now implemented by the ESS team for later waves of the survey) with fully worked examples. Also step-by-step analysis (using SPSS procedure MULT RESPONSE) of a complex question on experience of discrimination, with suggestions for improved data coding and presentation of tables.

3.1:  Variable names and labels
3.2:  An example of awkward labelling
3.3:  Multiple response

The above only use SPSS in syntax mode which is far easier and quicker than using the menus.
 
John Hall
 
PS  I did all the processing and analysis of a survey for Ruth Hall (no relation) of Women Against Rape when she wrote her book Ask Any Woman (Falling Wall Press) which was based on endless mult resp tables.  It got the law changed so that rape within marriage became a criminal offence in the UK.
 
 
----- Original Message -----
Sent: Tuesday, January 26, 2010 9:11 PM
Subject: Check All That Apply

Dear Colleagues,

 

  I am currently in the process of constructing a SPSS file for a survey that has several questions that are in a “check all that apply” format. Several of these questions have 10+options, so it would be quite inconvenient to create dummy variables for each option. Is there any way to format a “check all that apply” code in SPSS? Any suggestions on other ways to format other than dummy coding?

 

Thanks,

 

Paige Ransford

Senior Researcher

The Center for Women in Politics and Public Policy

University of Massachusetts Boston

Boston, MA  02125

617-287-7407

www.mccormack.umb.edu/centers/cwppp

 

 

 

 

 

Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

ViAnn Beadle
In reply to this post by Bruce Weaver
IMHO, this is an inconvenient way to map multiple responses if you are
primarily interested in tabulating them to find counts and percents in
either MULT RESPONSE or CTABLES. I'd set up the variables as a multiple
response set rather than a categorical dichotomies set. with as many
variables as expected selections (assuming that the respondent can choose
all of the choices).

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Bruce Weaver
Sent: Tuesday, January 26, 2010 3:18 PM
To: [hidden email]
Subject: Re: Check All That Apply

Hector Maletta wrote:

>
> The usual way is "multiple dichotomies" i.e. dummies. Another
> alternative is "multiple responses". In this case, you establish first
> what is the maximum number of simultaneous choices per person (perhaps
> the individual choosing the largest number chose 7 out of the 10
> options presented). If the maximum number of choices per person is k,
> you create k variables with values 1 to
> 10 each, each coding a different choice for that particular question.
> A third alternative is to combine the numerous response patters into a
> single figure; suppose you have 5 such questions; a person choosing
> the first alternative in all questions would have a value 11111.
> Better still, you may put a first figure (say, another 1) without any
> meaning, just to enable the introduction of zeroes if any question is
> left unanswered: a person choosing
> 1 at questions 2 and 5, and leaving the rest unanswered, would be
> represented by 101001.
>
> These alternatives are viable only if the maximum number of choices
> per person is not too high, or if the number of response patterns is
> not extremely large. In one case I handled some time ago, I used the
> third alternative, found that a few patterns accounted for about 90%
> of the cases, and grouped the remaining (many) patterns into a couple
> of categories, to avoid having an excessive number of patterns in my
> tables.
>
>
> Hector
>
>

I was thinking about Hector's "binary coding" method too--someone suggested
it for a similar problem posted to comp.soft-sys.stat.spss not too long ago.
But rather than including the meaningless first digit, I think one could use
an N-format.  E.g., for a 5-option item:

format item (n5.0).

If the 2nd and 5th items are checked, it should display as:  01001



-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE:  My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context:
http://old.nabble.com/Check-All-That-Apply-tp27329139p27330799.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

=====================
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
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

Hector Maletta
ViAnn,
I agree, but in some instances researchers are interested in response
patterns, and this involves looking at the combinations of responses. For
other purposes, such as those you mention, multiple dichotomies are the best
option.
Hector
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
ViAnn Beadle
Sent: 26 January 2010 21:03
To: [hidden email]
Subject: Re: Check All That Apply

IMHO, this is an inconvenient way to map multiple responses if you are
primarily interested in tabulating them to find counts and percents in
either MULT RESPONSE or CTABLES. I'd set up the variables as a multiple
response set rather than a categorical dichotomies set. with as many
variables as expected selections (assuming that the respondent can choose
all of the choices).

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Bruce Weaver
Sent: Tuesday, January 26, 2010 3:18 PM
To: [hidden email]
Subject: Re: Check All That Apply

Hector Maletta wrote:

>
> The usual way is "multiple dichotomies" i.e. dummies. Another
> alternative is "multiple responses". In this case, you establish first
> what is the maximum number of simultaneous choices per person (perhaps
> the individual choosing the largest number chose 7 out of the 10
> options presented). If the maximum number of choices per person is k,
> you create k variables with values 1 to
> 10 each, each coding a different choice for that particular question.
> A third alternative is to combine the numerous response patters into a
> single figure; suppose you have 5 such questions; a person choosing
> the first alternative in all questions would have a value 11111.
> Better still, you may put a first figure (say, another 1) without any
> meaning, just to enable the introduction of zeroes if any question is
> left unanswered: a person choosing
> 1 at questions 2 and 5, and leaving the rest unanswered, would be
> represented by 101001.
>
> These alternatives are viable only if the maximum number of choices
> per person is not too high, or if the number of response patterns is
> not extremely large. In one case I handled some time ago, I used the
> third alternative, found that a few patterns accounted for about 90%
> of the cases, and grouped the remaining (many) patterns into a couple
> of categories, to avoid having an excessive number of patterns in my
> tables.
>
>
> Hector
>
>

I was thinking about Hector's "binary coding" method too--someone suggested
it for a similar problem posted to comp.soft-sys.stat.spss not too long ago.
But rather than including the meaningless first digit, I think one could use
an N-format.  E.g., for a 5-option item:

format item (n5.0).

If the 2nd and 5th items are checked, it should display as:  01001



-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE:  My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context:
http://old.nabble.com/Check-All-That-Apply-tp27329139p27330799.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

=====================
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
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Check All That Apply

ViAnn Beadle
There is nothing that prevents the user from creating multiple mappings when
the user has multiple methods of analysis and multiple reporting goals.

By focusing on just one approach, you really limit your options or make
analysis difficult. Data management choice is dependent upon what the user
wants to do to  the data  and what SPSS expects the data to look like for
that particular goal.

Unfortunately posters rarely tell us what their end goal and frequently fail
to tell us what the data look like now.

-----Original Message-----
From: Hector Maletta [mailto:[hidden email]]
Sent: Tuesday, January 26, 2010 5:45 PM
To: 'ViAnn Beadle'; [hidden email]
Subject: RE: Check All That Apply

ViAnn,
I agree, but in some instances researchers are interested in response
patterns, and this involves looking at the combinations of responses. For
other purposes, such as those you mention, multiple dichotomies are the best
option.
Hector
-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
ViAnn Beadle
Sent: 26 January 2010 21:03
To: [hidden email]
Subject: Re: Check All That Apply

IMHO, this is an inconvenient way to map multiple responses if you are
primarily interested in tabulating them to find counts and percents in
either MULT RESPONSE or CTABLES. I'd set up the variables as a multiple
response set rather than a categorical dichotomies set. with as many
variables as expected selections (assuming that the respondent can choose
all of the choices).

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Bruce Weaver
Sent: Tuesday, January 26, 2010 3:18 PM
To: [hidden email]
Subject: Re: Check All That Apply

Hector Maletta wrote:

>
> The usual way is "multiple dichotomies" i.e. dummies. Another
> alternative is "multiple responses". In this case, you establish first
> what is the maximum number of simultaneous choices per person (perhaps
> the individual choosing the largest number chose 7 out of the 10
> options presented). If the maximum number of choices per person is k,
> you create k variables with values 1 to
> 10 each, each coding a different choice for that particular question.
> A third alternative is to combine the numerous response patters into a
> single figure; suppose you have 5 such questions; a person choosing
> the first alternative in all questions would have a value 11111.
> Better still, you may put a first figure (say, another 1) without any
> meaning, just to enable the introduction of zeroes if any question is
> left unanswered: a person choosing
> 1 at questions 2 and 5, and leaving the rest unanswered, would be
> represented by 101001.
>
> These alternatives are viable only if the maximum number of choices
> per person is not too high, or if the number of response patterns is
> not extremely large. In one case I handled some time ago, I used the
> third alternative, found that a few patterns accounted for about 90%
> of the cases, and grouped the remaining (many) patterns into a couple
> of categories, to avoid having an excessive number of patterns in my
> tables.
>
>
> Hector
>
>

I was thinking about Hector's "binary coding" method too--someone suggested
it for a similar problem posted to comp.soft-sys.stat.spss not too long ago.
But rather than including the meaningless first digit, I think one could use
an N-format.  E.g., for a 5-option item:

format item (n5.0).

If the 2nd and 5th items are checked, it should display as:  01001



-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE:  My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context:
http://old.nabble.com/Check-All-That-Apply-tp27329139p27330799.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

=====================
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
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Print Label and Variable name in Factor

Jon K Peck
In reply to this post by Granaas, Michael

Go to Edit>Options>Output Labels and choose Names and Labels in the Pivot Table Labeling group.  You won't get quite the format you are asking for, but you will get both variable names and variable labels.

You can do the same sort of thing with values and value labels.

If you really need the parenthesized form, you would need to create a script or use SPSSINC MODIFY TABLES with a little custom function.

HTH,

Jon Peck
SPSS, an IBM Company
[hidden email]
312-651-3435



From: "Granaas, Michael" <[hidden email]>
To: [hidden email]
Date: 01/26/2010 08:12 PM
Subject: [SPSSX-L] Print Label and Variable name in Factor
Sent by: "SPSSX(r) Discussion" <[hidden email]>





I have a large batch of questions, cleverly named q1, q2, q3, etc.  The actual question appears in the data set as the question label.
 
When I run principal components analysis and print out the rotated component matrix I get all the question labels.   What I want is the label and the variable name.  Preferably with a format like:  Label (q3).
 
Any suggestions?
 
Thanks,
 
Michael

****************************************************
Michael Granaas             [hidden email]
Assoc. Prof.                Phone: 605 677 5295
Dept. of Psychology         FAX:  605 677 3195
University of South Dakota
414 E. Clark St.
Vermillion, SD 57069
*****************************************************




Loading...