"Paired" test for 3 variables

[Hugh Sandoz]

Hi,

I've conducted a research in which each sample/respondent is tested for 3 attributes (how much he/she likes A, B and C). A, B and C are not repeated measurements of the same thing, but measurements of different things from the same person. I need to compare the means of A, B and C, and see which is larger/smaller than which (basically multiple pairwise comparisons).

Because the responses are matched by respondents, should I first test for independence of A, B and C before deciding if I should use ANOVA (independent variables) or something like a paired t-test (but for 3 variables)? Or should I assume that they are dependent and directly go for the latter?

And how would I do a test for 3 dependent/matched variables? If I'm not wrong, repeated-measures one-way ANOVA only works for repeated measures of the same thing. Should I use multivariate analysis of variance (MANOVA)?

I've tried running MANOVA on SPSS with 3 dependent variables and no fixed factors - it's given me significant results for Pillai's Trace, Wilk's Lambda, Hotelling's Trace and Roy's Largest  Root. However, I have no idea what these results mean despite looking up the SPSS manual, nor do I know which mean is bigger than which. As far as I can tell, no pairwise comparisons have been conducted. Can anyone please advise?

Thanks in advance.

[Swank, Paul R]

While they may not be precisely what you would call repeated measures, if you wish to compare related observations then a repeated measures ANOVA or mixed model would be the best approach for the omnibus test. Follow-up up to a significant omnibus test would be by protected paired t-tests or either pairwise differences or contrasts.

Dr. Paul R. Swank,
Professor and Director of Research
Children's Learning Institute
University of Texas Health Science Center-Houston


-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Hugh Sandoz
Sent: Sunday, June 07, 2009 10:22 AM
To: [hidden email]
Subject: "Paired" test for 3 variables

Hi,

I've conducted a research in which each sample/respondent is tested for 3
attributes (how much he/she likes A, B and C). A, B and C are not repeated
measurements of the same thing, but measurements of different things from
the same person. I need to compare the means of A, B and C, and see which is
larger/smaller than which (basically multiple pairwise comparisons).

Because the responses are matched by respondents, should I first test for
independence of A, B and C before deciding if I should use ANOVA
(independent variables) or something like a paired t-test (but for 3
variables)? Or should I assume that they are dependent and directly go for
the latter?

And how would I do a test for 3 dependent/matched variables? If I'm not
wrong, repeated-measures one-way ANOVA only works for repeated measures of
the same thing. Should I use multivariate analysis of variance (MANOVA)?

I've tried running MANOVA on SPSS with 3 dependent variables and no fixed
factors - it's given me significant results for Pillai's Trace, Wilk's
Lambda, Hotelling's Trace and Roy's Largest  Root. However, I have no idea
what these results mean despite looking up the SPSS manual, nor do I know
which mean is bigger than which. As far as I can tell, no pairwise
comparisons have been conducted. Can anyone please advise?

Thanks in advance.
--
View this message in context: http://www.nabble.com/%22Paired%22-test-for-3-variables-tp23912030p23912030.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

[Art Kendall]

if "liking" for A, B, and C are measured on the same response scale,  then this would be a kind of repeated measure where the levels are A, B, and C.

If "liking" is not the same construct with regard to the 3 stimulus conditions, then comparing means would not make sense.
 
Art Kendall
Social Research Consultants

Hugh Sandoz wrote:

Hi,

I've conducted a research in which each sample/respondent is tested for 3
attributes (how much he/she likes A, B and C). A, B and C are not repeated
measurements of the same thing, but measurements of different things from
the same person. I need to compare the means of A, B and C, and see which is
larger/smaller than which (basically multiple pairwise comparisons).

Because the responses are matched by respondents, should I first test for
independence of A, B and C before deciding if I should use ANOVA
(independent variables) or something like a paired t-test (but for 3
variables)? Or should I assume that they are dependent and directly go for
the latter?

And how would I do a test for 3 dependent/matched variables? If I'm not
wrong, repeated-measures one-way ANOVA only works for repeated measures of
the same thing. Should I use multivariate analysis of variance (MANOVA)?

I've tried running MANOVA on SPSS with 3 dependent variables and no fixed
factors - it's given me significant results for Pillai's Trace, Wilk's
Lambda, Hotelling's Trace and Roy's Largest  Root. However, I have no idea
what these results mean despite looking up the SPSS manual, nor do I know
which mean is bigger than which. As far as I can tell, no pairwise
comparisons have been conducted. Can anyone please advise?

Thanks in advance.
--
View this message in context: http://www.nabble.com/%22Paired%22-test-for-3-variables-tp23912030p23912030.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

[mpirritano]

I know that when I enter categorical variables into a logistic
regression with pasw I do not need to enter in all dummy or effect coded
versions of the variable. Correct me if I'm wrong. In OLS regression,
yes, you do need to enter all dummy variables.

The question is: In genlin do you need to enter all dummies, or if you
enter a categorical as a factor then is all taken care of? I notice that
I do get one fewer effect than the total number of categories, as I
would with dummies. If genlin takes care of the coding when I enter the
categorical variable what type of coding does it use (effect, dummy)?

Actually I've just noticed that when I request contrast coefficient (L)
matrices the matrices look like effect coding. Kind of annoying that
it's so inconsistent. Sometimes you need to do the coding yourself
(regression), sometimes you don't (logistic regression, genlin)? Or does
it hurt if you do enter them manually for logistic or genlin?

Please confirm.

Thanks for all the help folks!

matt

Matthew Pirritano, Ph.D.
Research Analyst IV
Medical Services Initiative (MSI)
Orange County Health Care Agency
[phone]

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

[Ergul, Emel A.]

Hi to all:

As a newbie SPSS syntact-er, I apologize if my message is annoyingly simple.
Here is the thing I need help:

I have 7 digit string variable (7 digit number) with random "#"  sign in it. In
variable there are few cases where space has been used instead of "#". I want to
get rid of both.

Q1. Does the position of sign and space matter?
Q2. Do you have any code that I can try to adjust my case?

Many thanks to all!!

emel


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.

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

[Oliver, Richard]

compute varname=replace(varname, "#", "").
compute varname=replace(varname, " ", "").

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ergul, Emel A.
Sent: Monday, June 08, 2009 10:50 AM
To: [hidden email]
Subject: How to get rid of a character/sign in string

Hi to all:

As a newbie SPSS syntact-er, I apologize if my message is annoyingly simple.
Here is the thing I need help:

I have 7 digit string variable (7 digit number) with random "#"  sign in it. In
variable there are few cases where space has been used instead of "#". I want to
get rid of both.

Q1. Does the position of sign and space matter?
Q2. Do you have any code that I can try to adjust my case?

Many thanks to all!!

emel


The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.

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

[Ruben Geert van den Berg]

Dear Emel,
 
You could do it with replace. Here's some SPSS test data with the characters # and " " (space) in it and how to remove them. Obviously, you don't need to create new variables (you could replace Var2 and Var3 by Var) but perhaps this makes it clearer. Just mind you: all spaces and instances of # will be removed.
 
Kind regards,
 
Ruben van den Berg
 
***Create data.
 
DATA LIST FREE/Var (A5).
BEGIN DATA
"#bla" "b#la" "bl#a" "bla#" " bla" "b la" "bl a" "bla "
END DATA.
 
***Create new variables (empty).
 
STR Var2 Var3 (A5).
 
***Create Var2 (=Var with # replaced by nothing).
 
COMP Var2=REP(Var,"#","")./*REP = REPlace.
 
***Create Var3 (=Var2 with space replaced by nothing).
 
COMP Var3=REP(Var2," ","").
EXE.




 



 

> Date: Mon, 8 Jun 2009 11:49:52 -0400
> From: [hidden email]
> Subject: How to get rid of a character/sign in string
> To: [hidden email]
>
> Hi to all:
>
> As a newbie SPSS syntact-er, I apologize if my message is annoyingly simple.
> Here is the thing I need help:
>
> I have 7 digit string variable (7 digit number) with random "#" sign in it. In
> variable there are few cases where space has been used instead of "#". I want to
> get rid of both.
>
> Q1. Does the position of sign and space matter?
> Q2. Do you have any code that I can try to adjust my case?
>
> Many thanks to all!!
>
> emel
>
>
> The information in this e-mail is intended only for the person to whom it is
> addressed. If you believe this e-mail was sent to you in error and the e-mail
> contains patient information, please contact the Partners Compliance HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in error
> but does not contain patient information, please contact the sender and properly
> dispose of the e-mail.
>
> =====================
> 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

---

Express yourself instantly with MSN Messenger! MSN Messenger

[Ergul, Emel A.]

Richard and Ruben:

 

Thank you so much!

 

emel

The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail.

[aayush25]

I DOING A RESEARCH PAPER AS A SUMMER INTERN IN MUTUAL FUND COMPANY. I NEED HELP REGARDING WHICH STATISTICAL TEST TO RUN IN SPSS ON DATA COLLECTED BY ME.

PROJECT OBJECTIVE IS TO FIND OUT THE FACTORS AFFECTING THE PERCEPTION OF MUTUAL FUND COMPANY IN ITS DISTRIBUTION CHANNEL. AND COMPARE THE PERCEPTION OF TOP 3 COMPANY ON SAME FACTORS

I DID A LITERATURE STUDY FIRST, THEN IDENTIFIED THE FACTORS WHICH MAY AFFECT TH PERCEPTION OF DISTRIBUTORS, I HAVE PREPARED A QUESTIONNAIRE FOR SAME, WHICH HAS BEEN APPROVED BY MY INDUSTRY GUIDE, I HAVE STARTED COLLECTING DATA ALSO. BUT I FACE ONE PROBLEM : WHICH STATISTICAL TEST TO BE USED?

IF YOU SEE MY QUESTIONNAIRE, ITS MAINLY DIVIDED INTO 3 PARTS, FIRSTLY THERE ARE BASIC QUESTIONS, SECOND PART CONTAINS QUESTION IN WHICH RESPONDENT ( DISTRIBUTOR ) HAS TO RATE FACTOR( 23 DIFFERENT FACTORS ) ON SEMANTIC DIFFERENTIAL SCALE (1 TO 5, 1 BEING LEAST IMPORTANT AND 5 BEING MOST IMPORTANT). THEN THE THIRD PART CONTAINS QUESTION IN WHICH RESPONDENT HAS TO RATE THE PERFORMANCE OF DIFFERENT MUTUAL FUND COMPANY ON ABOVE SAME FACTORS. STAPEL SCALE HAS BEEN USED.

NOW I HAVE THOUGHT THAT I'LL DO FACTOR ANALYSIS (DATA REDUCTION) ON QUESTION 5, THIS WILL GIVE ME UNCORRELATED FACTORS WHICH AFFECT THE PERCEPTION OF DISTRIBUTORS. ON QUESTION 6, DESCRIPTIVE ANALYSIS CAN BE DONE, BUT WHAT ELSE ( STATISTICAL TEST) CAN BE DONE.

WHAT SHOULD I DO TO COMPARE MEAN OF 3 VARIABLES? SOMETHING LIKE PAIRED  T-TEST BUT WITH 3 VARIABLES. THEN MY HYPOTHESIS WILL PERFORMANCE OF COMPANY X ON SPECIFIC VARIABLES IS BETTER THAN OTHER TWO COMPANYS.

PLEASE HELP ME. ITS URGENT!

THANK YOU

IF YOU WANT QUESTIONNAIRE PLEASE DO CONTACT ME, I'LL MAIL IT TO YOU.

[Bruce Weaver]

PLEASE DO NOT SHOUT!  Look at how much easier it is to read your post when it is in sentence case!

--- original post in sentence case ---

I doing a research paper as a summer intern in mutual fund company. I need help regarding which statistical test to run in spss on data collected by me.

Project objective is to find out the factors affecting the perception of mutual fund company in its distribution channel. And compare the perception of top 3 company on same factors

i did a literature study first, then identified the factors which may affect th perception of distributors, i have prepared a questionnaire for same, which has been approved by my industry guide, i have started collecting data also. But i face one problem : which statistical test to be used?

If you see my questionnaire, its mainly divided into 3 parts, firstly there are basic questions, second part contains question in which respondent ( distributor ) has to rate factor( 23 different factors ) on semantic differential scale (1 to 5, 1 being least important and 5 being most important). Then the third part contains question in which respondent has to rate the performance of different mutual fund company on above same factors. Stapel scale has been used.

Now i have thought that i'll do factor analysis (data reduction) on question 5, this will give me uncorrelated factors which affect the perception of distributors. On question 6, descriptive analysis can be done, but what else ( statistical test) can be done.

What should i do to compare mean of 3 variables? Something like paired  t-test but with 3 variables. Then my hypothesis will performance of company x on specific variables is better than other two companys.

Please help me. Its urgent!

Thank you

if you want questionnaire please do contact me, i'll mail it to you.

aayush25 wrote

I DOING A RESEARCH PAPER AS A SUMMER INTERN IN MUTUAL FUND COMPANY. I NEED HELP REGARDING WHICH STATISTICAL TEST TO RUN IN SPSS ON DATA COLLECTED BY ME.

PROJECT OBJECTIVE IS TO FIND OUT THE FACTORS AFFECTING THE PERCEPTION OF MUTUAL FUND COMPANY IN ITS DISTRIBUTION CHANNEL. AND COMPARE THE PERCEPTION OF TOP 3 COMPANY ON SAME FACTORS

I DID A LITERATURE STUDY FIRST, THEN IDENTIFIED THE FACTORS WHICH MAY AFFECT TH PERCEPTION OF DISTRIBUTORS, I HAVE PREPARED A QUESTIONNAIRE FOR SAME, WHICH HAS BEEN APPROVED BY MY INDUSTRY GUIDE, I HAVE STARTED COLLECTING DATA ALSO. BUT I FACE ONE PROBLEM : WHICH STATISTICAL TEST TO BE USED?

IF YOU SEE MY QUESTIONNAIRE, ITS MAINLY DIVIDED INTO 3 PARTS, FIRSTLY THERE ARE BASIC QUESTIONS, SECOND PART CONTAINS QUESTION IN WHICH RESPONDENT ( DISTRIBUTOR ) HAS TO RATE FACTOR( 23 DIFFERENT FACTORS ) ON SEMANTIC DIFFERENTIAL SCALE (1 TO 5, 1 BEING LEAST IMPORTANT AND 5 BEING MOST IMPORTANT). THEN THE THIRD PART CONTAINS QUESTION IN WHICH RESPONDENT HAS TO RATE THE PERFORMANCE OF DIFFERENT MUTUAL FUND COMPANY ON ABOVE SAME FACTORS. STAPEL SCALE HAS BEEN USED.

NOW I HAVE THOUGHT THAT I'LL DO FACTOR ANALYSIS (DATA REDUCTION) ON QUESTION 5, THIS WILL GIVE ME UNCORRELATED FACTORS WHICH AFFECT THE PERCEPTION OF DISTRIBUTORS. ON QUESTION 6, DESCRIPTIVE ANALYSIS CAN BE DONE, BUT WHAT ELSE ( STATISTICAL TEST) CAN BE DONE.

WHAT SHOULD I DO TO COMPARE MEAN OF 3 VARIABLES? SOMETHING LIKE PAIRED  T-TEST BUT WITH 3 VARIABLES. THEN MY HYPOTHESIS WILL PERFORMANCE OF COMPANY X ON SPECIFIC VARIABLES IS BETTER THAN OTHER TWO COMPANYS.

PLEASE HELP ME. ITS URGENT!

THANK YOU

IF YOU WANT QUESTIONNAIRE PLEASE DO CONTACT ME, I'LL MAIL IT TO YOU.