Data coding for constant sum scale

[shoaib ul-haq]

Dear All

I am a new user of SPSS software and is still in the
exploration stage. I have subscribed to this list
because I want to learn from the SPSS gurus out there.


So here comes the first one. How should I code the
question with a constant sum scale in SPSS.

Thanks for your help.

Regards
Shoaib




____________________________________________________________________________________Luggage? GPS? Comic books?
Check out fitting gifts for grads at Yahoo! Search
http://search.yahoo.com/search?fr=oni_on_mail&p=graduation+gifts&cs=bz

[Art Kendall]

Please give us an example or two of what you mean by a constant sum scale.

Art Kendall
Social Research Consultants

shoaib ul-haq wrote:

> Dear All
>
> I am a new user of SPSS software and is still in the
> exploration stage. I have subscribed to this list
> because I want to learn from the SPSS gurus out there.
>
>
> So here comes the first one. How should I code the
> question with a constant sum scale in SPSS.
>
> Thanks for your help.
>
> Regards
> Shoaib
>
>
>
>
> ____________________________________________________________________________________Luggage? GPS? Comic books?
> Check out fitting gifts for grads at Yahoo! Search
> http://search.yahoo.com/search?fr=oni_on_mail&p=graduation+gifts&cs=bz
>
>
>

[Bob Walker-2]

Shoaib,

Constant sum questions are common in marketing research. I'm assuming that your constant sum question is something akin to...

"Of the last/next 100 cans of diet soda you buy, how many were/will be for each of the brands listed below?  [Please type in the number of cans for each brand using whole numbers only. If you have not/would not buy one of the brands listed, type in a “0”. Note that the total must add to 100.]
 
Brand A
Brand B...
Brand Z

Each of the brands should be recorded as a separate variable in your data file. Your variables should be Brand_A, Brand_B, and then continuing all the way to Brand_Z. Assuming that each respondent enters a value for every brand, the sum of the means across these variables should add up to 100 (or whatever constant sum value total you have pre-defined, 10, 20, etc). You can also run brand interaction, cannibalization, and proximities analyses based on this data.

Does this help, or is your question more specific?

Bob Walker
Surveys & Forecasts, LLC
www.safllc.com


----- Original Message -----
From: shoaib ul-haq
Date: Thursday, May 24, 2007 3:18 am
Subject: Data coding for constant sum scaleTo: [hidden email]

> Dear All
>
> I am a new user of SPSS software and is still in the
> exploration stage. I have subscribed to this list
> because I want to learn from the SPSS gurus out there.
>
>
> So here comes the first one. How should I code the
> question with a constant sum scale in SPSS.
>
> Thanks for your help.
>
> Regards
> Shoaib
>
>
>
>
> ____________________________________________________________________________________Luggage? GPS? Comic books?
> Check out fitting gifts for grads at Yahoo! Search
> http://search.yahoo.com/search?fr=oni_on_mail&p=graduation+gifts&cs=bz
>

 Bob Walker
Surveys & Forecasts, LLC
www.safllc.com

[Pirritano, Matthew]

Hello all,

I'm doing analyses of couple's responses in an infertility study. The questionnaires of interest assess things such as how they cope with infertility and the extent to which they experience different forms of stress. I have this data at baseline and a one year follow up.

I was thinking that I could do a repeated-measures MANCOVA (with a variety of covariates such as how long they've been trying to get pregnant, age, etc.). I was planning on using the sex of the individual (husband or wife) as one of the repeated measures factors. There would be three repeated measures factors in the case of the analysis of coping styles: sex, time, and coping style.

First question: Does it make sense to treat the data from each member of the couple (the sex factor) as a repeated-measure? Being that they will likely be correlated it seems to make sense. I wanted to run this by folks first, however, because I believe it will require restructuring the data. The data currently has each individual (husband and wife) on their own line. I believe that I would need to put each couple on one line? Am I correct? Is there a more logical way to do this.

I know that I could probable do this in a multilevel model but I wanted to start out with a simple analysis (for me that is, not being too familiar with multilevel models, especially not in SPSS).

Any advice'd be much appreciated.

Thanks,
Matt



Matthew Pirritano, Ph.D.
Assistant Professor of Psychology
Smith Hall 116C
Chapman University
Department of Psychology
One University Drive
Orange, CA 92866
Telephone [phone]
FAX [phone]

[Kathy McKnight]

You might want to have a look at Dave Kenny's (at University of Connecticut)
website in which he discusses dyadic analysis, which sounds like what you
want to do. He provides a brief tutorial. The link to his website:
http://davidakenny.net/dyad.htm

Here's an example of the topics he covers (taken from this website):

Topic 1. <http://davidakenny.net/dyad.htm#Top1>  What is a standard dyadic
design?

Topic 2. <http://davidakenny.net/dyad.htm#Top2>  What is the level of
measurement of the outcome measure?

Topic 3. <http://davidakenny.net/dyad.htm#Top3>  Are the dyad members
distinguishable or not?

Topic 4. <http://davidakenny.net/dyad.htm#Top4>  Determine the types of
variables in the analysis.

Topic 5. <http://davidakenny.net/dyad.htm#Top5>  The assessment of
nonindependence

Topic 6. <http://davidakenny.net/dyad.htm#Top6> Consequences of
nonindependence on significance testing

Topic 7. <http://davidakenny.net/dyad.htm#Top7>  How are the effects of a
between-dyads <http://davidakenny.net/dyad.htm#BET> predictor variable
estimated?

Topic 8. <http://davidakenny.net/dyad.htm#Top8>  How are the effects
measured when the predictor variable is
within<http://davidakenny.net/dyad.htm#WITH>dyads?

Topic 9. <http://davidakenny.net/dyad.htm#Top9>  How is nonindependence
controlled if there are several predictor variables, some of which are
between and others of which are within dyad?

Topic 10. <http://davidakenny.net/dyad.htm#Top10>  How can effects be
estimated if the predictor variable is a
mixed<http://davidakenny.net/dyad.htm#MIXED>variable?

Bottom line: I think you'll find his tutorial would recommend against the
analysis you propose for this type of study design.

Kathy McKnight




On 5/24/07, Pirritano, Matthew <[hidden email]> wrote:

>
> Hello all,
>
> I'm doing analyses of couple's responses in an infertility study. The
> questionnaires of interest assess things such as how they cope with
> infertility and the extent to which they experience different forms of
> stress. I have this data at baseline and a one year follow up.
>
> I was thinking that I could do a repeated-measures MANCOVA (with a variety
> of covariates such as how long they've been trying to get pregnant, age,
> etc.). I was planning on using the sex of the individual (husband or wife)
> as one of the repeated measures factors. There would be three repeated
> measures factors in the case of the analysis of coping styles: sex, time,
> and coping style.
>
> First question: Does it make sense to treat the data from each member of
> the couple (the sex factor) as a repeated-measure? Being that they will
> likely be correlated it seems to make sense. I wanted to run this by folks
> first, however, because I believe it will require restructuring the data.
> The data currently has each individual (husband and wife) on their own line.
> I believe that I would need to put each couple on one line? Am I correct? Is
> there a more logical way to do this.
>
> I know that I could probable do this in a multilevel model but I wanted to
> start out with a simple analysis (for me that is, not being too familiar
> with multilevel models, especially not in SPSS).
>
> Any advice'd be much appreciated.
>
> Thanks,
> Matt
>
>
>
> Matthew Pirritano, Ph.D.
> Assistant Professor of Psychology
> Smith Hall 116C
> Chapman University
> Department of Psychology
> One University Drive
> Orange, CA 92866
> Telephone [phone]
> FAX [phone]
>