2x2 Latin square design analysis help

[Sur1605]

Hi,

I am doing an analysis on my data with a 2x2 Latin square design. So basically I have four groups, diet intervention group,exercise intervention group, Diet and exercise combination intervention group and a control group. I am trying to perform a 2 way anova and can't figure out how to set up the raw data on a spss spreadsheet. Also, Is a 2way repeated measure anova a better test for this design?

Thanks for the help in advance.
SB

[Bruce Weaver]

A 2x2 Latin square would look like this:

A B
B A

I.e., it suggests you have two conditions (A and B) with repeated measures and the order counterbalanced between subjects.  (For the 2x2 case, this is equivalent to complete counterbalancing.)

But what you describe is NOT a 2x2 Latin square.  Rather, you describe a plain ordinary 2x2 design with both factors manipulated between subjects.  You might try a Google search on <factorial anova spss 2x2>.  The hits even include some video tutorials.

HTH.

Sur1605 wrote

Hi,

I am doing an analysis on my data with a 2x2 Latin square design. So basically I have four groups, diet intervention group,exercise intervention group, Diet and exercise combination intervention group and a control group. I am trying to perform a 2 way anova and can't figure out how to set up the raw data on a spss spreadsheet. Also, Is a 2way repeated measure anova a better test for this design?

Thanks for the help in advance.
SB

[David Marso]

I don't see how this is a Repeated Measures design nor a Latin Square unless you are leaving out essential facts of your design.

Sounds like a simple regression problem (of which ANOVA is a special case).

Y=B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise + Resid.
                 Diet  Exercise
Control          0      0
Diet              1      0
Exercise        0      1
Diet&Exer      1      1
-----------------
The following illustrates this.
If you examine the data file after running this exercise you will see how your data should be structured.
**SIMULATED DATA **.
new file.
input program.
loop caseid=1 to 100.
compute exercise=trunc(uniform(2)).
compute diet=trunc(uniform(2)).
compute exerdiet=exercise*diet.
compute Y=10 + 5*exercise + 8 * diet + 10* exercise*diet + normal(1).
end case.
end loop.
end file.
end input program.

crosstabs table exercise by diet.

regression dep Y
 / method enter exercise diet exerdiet/save pred.

RECODE DIET EXERCISE (0=1)(1=0).
UNIANOVA
  y  BY exercise diet
  /PRINT = PARAMETER
  /DESIGN = exercise diet exercise*diet /SAVE PRED.

Notice that the parameter estimates, associated statistics and predicted values match precisely!!!
HTH, David

Bruce Weaver wrote

A 2x2 Latin square would look like this:

A B
B A

I.e., it suggests you have two conditions (A and B) with repeated measures and the order counterbalanced between subjects.  (For the 2x2 case, this is equivalent to complete counterbalancing.)

But what you describe is NOT a 2x2 Latin square.  Rather, you describe a plain ordinary 2x2 design with both factors manipulated between subjects.  You might try a Google search on <factorial anova spss 2x2>.  The hits even include some video tutorials.

HTH.

Sur1605 wrote

Hi,

I am doing an analysis on my data with a 2x2 Latin square design. So basically I have four groups, diet intervention group,exercise intervention group, Diet and exercise combination intervention group and a control group. I am trying to perform a 2 way anova and can't figure out how to set up the raw data on a spss spreadsheet. Also, Is a 2way repeated measure anova a better test for this design?

Thanks for the help in advance.
SB

[Rich Ulrich]

When you have repeated measures, you certainly cannot
legitimately ignore that part of the design.

Is it indeed true that you have repeated measures, such that
every subject was tested under each of the 4 conditions?
(Depending on what the criterion is, you might have serious
concerns about taking into account the effect of Order.)

 - If so:  You do have to encode the Ordering. 
A simple way to list the data would be to encode as
1-2-3-4  the order that each treatment-combination was received,
by each subject.  If each is on a separate row, you have
  ID, Tr1, Tr2, Order, Outcome.

--
Rich Ulrich

> Date: Tue, 31 Jan 2012 22:59:32 -0800

> From: [hidden email]
> Subject: 2x2 Latin square design analysis help
> To: [hidden email]
>
> Hi,
>
> I am doing an analysis on my data with a 2x2 Latin square design. So
> basically I have four groups, diet intervention group,exercise intervention
> group, Diet and exercise combination intervention group and a control group.
> I am trying to perform a 2 way anova and can't figure out how to set up the
> raw data on a spss spreadsheet. Also, Is a 2way repeated measure anova a
> better test for this design?
>

[...]

[David Marso]

OTOH:  *IF* you do indeed have repeated measures then please describe your research design in some detail (and IGNORE my previous post).  You speak of 4 groups.  When I think of 'groups' I think of *INDEPENDENT* groups of subjects.
What is your response variable?  Is this some sort of before/after inquiry?    ESP and partial information regarding details sometimes will lead to misinformation and bad advise!

David Marso wrote

I don't see how this is a Repeated Measures design nor a Latin Square unless you are leaving out essential facts of your design.

Sounds like a simple regression problem (of which ANOVA is a special case).

Y=B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise + Resid.
                 Diet  Exercise
Control          0      0
Diet              1      0
Exercise        0      1
Diet&Exer      1      1
-----------------
The following illustrates this.
If you examine the data file after running this exercise you will see how your data should be structured.
**SIMULATED DATA **.
new file.
input program.
loop caseid=1 to 100.
compute exercise=trunc(uniform(2)).
compute diet=trunc(uniform(2)).
compute exerdiet=exercise*diet.
compute Y=10 + 5*exercise + 8 * diet + 10* exercise*diet + normal(1).
end case.
end loop.
end file.
end input program.

crosstabs table exercise by diet.

regression dep Y
 / method enter exercise diet exerdiet/save pred.

RECODE DIET EXERCISE (0=1)(1=0).
UNIANOVA
  y  BY exercise diet
  /PRINT = PARAMETER
  /DESIGN = exercise diet exercise*diet /SAVE PRED.

Notice that the parameter estimates, associated statistics and predicted values match precisely!!!
HTH, David

Bruce Weaver wrote

A 2x2 Latin square would look like this:

A B
B A

I.e., it suggests you have two conditions (A and B) with repeated measures and the order counterbalanced between subjects.  (For the 2x2 case, this is equivalent to complete counterbalancing.)

But what you describe is NOT a 2x2 Latin square.  Rather, you describe a plain ordinary 2x2 design with both factors manipulated between subjects.  You might try a Google search on <factorial anova spss 2x2>.  The hits even include some video tutorials.

HTH.

Sur1605 wrote

Hi,

I am doing an analysis on my data with a 2x2 Latin square design. So basically I have four groups, diet intervention group,exercise intervention group, Diet and exercise combination intervention group and a control group. I am trying to perform a 2 way anova and can't figure out how to set up the raw data on a spss spreadsheet. Also, Is a 2way repeated measure anova a better test for this design?

Thanks for the help in advance.
SB

[Sur1605]

So, to be clear about the research design, i had four independent groups.

Group 1: Diet grp
Group 2: Exercise grp
Group 3: Diet+Exercise grp
Group 4: Control grp

 All these are independent groups and i wanted to see the effect of these treatments on continuous variables like body weight, body fat etc etc after 12 week of intervention. I have the data, and i was planning on performing  a 2way anova. But i got stuck at the point where i am not clear on how to put in the data in spss spreadsheet for this particular test.  I want to see if the effect on continuous variable is due to exercise, deit or their is an interaction effect.

Hope i explained it well this time.
Sb

[David Marso]

Well, can you wrap your brain around the following?
---
Diet  YES    Exer  YES     GROUP=3
Diet  YES    Exer  NO      GROUP=1
Diet   NO    Exer  YES     GROUP=2
Diet   NO    Exer   NO     GROUP=4

SPSS data structure
Diet    Exercise    Dependent variables.....
1           1         .......
1           0
0           1
0           0

Questions to ask yourself?
Were subjects RANDOMLY ASSIGNED to the four groups?
If yes then carry on...
If NO then you have a big problem.
----
Were there any baseline measures of interest taken BEFORE the intervention?
Are they the same variables you measured AFTER the intervention?
Affirmative answers to these would suggest somewhat more complicated analyses than either a simple 2x2 ANOVA or regreesion I suggested in my previous reply (Possibly a MIXED model or an (M)ANCOVA).

If so then the question becomes a bit more interesting and more tricky to properly analyze.

Sur1605 wrote

So, to be clear about the research design, i had four independent groups.

Group 1: Diet grp
Group 2: Exercise grp
Group 3: Diet+Exercise grp
Group 4: Control grp

 All these are independent groups and i wanted to see the effect of these treatments on continuous variables like body weight, body fat etc etc after 12 week of intervention. I have the data, and i was planning on performing  a 2way anova. But i got stuck at the point where i am not clear on how to put in the data in spss spreadsheet for this particular test.  I want to see if the effect on continuous variable is due to exercise, deit or their is an interaction effect.

Hope i explained it well this time.
Sb

[Sur1605]

Thanks David for such an amazing description. This is exactly how i tried doing it before i posted my question on this forum. But everytime i analysed my data somehow it gave me this message:
"Post hoc tests are not performed for Diet because there are fewer than three groups.
Post hoc tests are not performed for Ex because there are fewer than three groups."

Basically its not able to perform the posthoc analysis. I thought i was doing something wrong in feeding in the data in the spreadsheet.

And yes, i measured the same variables before and after the trial. Looks like i will have to reconsider the choice of test being used here.

Thanks a lot for your help.

Sb

[Ryan]

From what I can tell, you have one between-subjects variable (Condition) which has four levels (Diet, Exercise, Diet+Exercise, Control) and one within-subjects variable (Time) which has two levels (baseline, post-intervention). This is commonly referred to as a mixed ANOVA (not to be confused with the MIXED procedure). You could analyze your data by fitting a general linear model or linear mixed model in SPSS. I don't have time to write code for you at the moment, but quite frankly, there must be numerous SPSS examples online. Bottom line is that you need to take into account correlation among residuals obtained by repeated observations of each subject. You can write specific contrasts of interest using the appropriate subcommand. I am always in favor of using the subcommand which requires that you understand the coefficient (L) matrix.

Ryan

On Thu, Feb 2, 2012 at 12:08 PM, Sur1605 <[hidden email]> wrote:

Thanks David for such an amazing description. This is exactly how i tried
doing it before i posted my question on this forum. But everytime i analysed
my data somehow it gave me this message:
"Post hoc tests are not performed for Diet because there are fewer than
three groups.
Post hoc tests are not performed for Ex because there are fewer than three
groups."

Basically its not able to perform the posthoc analysis. I thought i was
doing something wrong in feeding in the data in the spreadsheet.

And yes, i measured the same variables before and after the trial. Looks
like i will have to reconsider the choice of test being used here.

Thanks a lot for your help.

Sb



--
View this message in context: http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5451327.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

[Peña, Hugo]

Estaré fuera de la oficina hasta el lunes 6 de febrero al martes 28 de febrero del 2012.

[Bruce Weaver]

But as David noted, the 4 conditions are formed by the 2x2 combination of two binary variables.  With the usual coding (i.e., 1 = Yes, 0 = No):

D E   Description
---------------------
0 0   Control
1 0   Diet, No Exercise
0 1   Exercise, No Diet
1 1   Diet and Exercise

If the outcome variable was measured at both baseline and followup, then the preferred approach, arguably, would be a 2x2 ANCOVA, which is equivalent to an ordinary least squares regression model as follows:

   Y1 = b0 + b1*Y0 + b2*D + b3*E + b4*D*E + error

where
Y1 = outcome variable at followup
Y0 = outcome variable at baseline
D and E are as shown above, and D*E = their product.

Of course one could run this model using MIXED, but for users not familiar with MIXED (and I get the impression the OP is not), it will be much easier to run it via UNIANOVA (Analyze - GLM - Univariate, with D and E entered as "fixed factors", and Y0 as a "covariate", and the D*E interaction included).  

Another important question David raised (and I don't recall seeing an answer) is whether there was random assignment to the 4 cells.  

There's my two cents!

R B wrote

From what I can tell, you have one between-subjects variable (Condition)
which has four levels (Diet, Exercise, Diet+Exercise, Control) and one
within-subjects variable (Time) which has two levels (baseline,
post-intervention). This is commonly referred to as a mixed ANOVA (not to
be confused with the MIXED procedure). You could analyze your data by
fitting a general linear model or linear mixed model in SPSS. I don't have
time to write code for you at the moment, but quite frankly, there must be
numerous SPSS examples online. Bottom line is that you need to take into
account correlation among residuals obtained by repeated observations of
each subject. You can write specific contrasts of interest using the
appropriate subcommand. I am always in favor of using the subcommand which
requires that you understand the coefficient (L) matrix.

Ryan

On Thu, Feb 2, 2012 at 12:08 PM, Sur1605 <[hidden email]> wrote:

> Thanks David for such an amazing description. This is exactly how i tried
> doing it before i posted my question on this forum. But everytime i
> analysed
> my data somehow it gave me this message:
> "Post hoc tests are not performed for Diet because there are fewer than
> three groups.
> Post hoc tests are not performed for Ex because there are fewer than three
> groups."
>
> Basically its not able to perform the posthoc analysis. I thought i was
> doing something wrong in feeding in the data in the spreadsheet.
>
> And yes, i measured the same variables before and after the trial. Looks
> like i will have to reconsider the choice of test being used here.
>
> Thanks a lot for your help.
>
> Sb
>
>
>
> --
> View this message in context:
> http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5451327.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
>

[David Marso]

*AND* there are multiple outcome variables measured both pre/post and it seems that there might be interesting functions which could be explored.  
i.e
ANCOVA using simple Pre/Post weight might be overly simplistic.
One might be inclined to look at (Post-Pre)/Pre
Proportionate weight loss.  Just my intuition.
It is easier for very heavy person to lose 15 pounds than a somewhat heavy person?
OTOH, I don't know diddly squat about the physiological dynamics but it just seems that theory might toss some interesting dice into the game.

Also :  I would postulate directional hypotheses a-priori .

Weight loss (Weight_Pre -Weight_Post)
Control < Diet ?=? Exercise < Diet+Exercise

Finally,  We can consider the following models:

PostWeight = B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise + Residual
vs
PostWeight = B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise + B4*PreWeight+Residual2

If Random Assignment then one would expect that Diet,Exercise,Preweight would be uncorrelated
hence Residual = B4*PreWeight+Residual2

*SO*  ANCOVA reduces the Error SS with a sacrifice of 1 df (assuming Pre and post are correlated).
----
Enough!?
OTOH: It becomes more interesting when we have a multivariate context.
SEM?
---------------------------------------------------------------------------

Bruce Weaver wrote

But as David noted, the 4 conditions are formed by the 2x2 combination of two binary variables.  With the usual coding (i.e., 1 = Yes, 0 = No):

D E   Description
---------------------
0 0   Control
1 0   Diet, No Exercise
0 1   Exercise, No Diet
1 1   Diet and Exercise

If the outcome variable was measured at both baseline and followup, then the preferred approach, arguably, would be a 2x2 ANCOVA, which is equivalent to an ordinary least squares regression model as follows:

   Y1 = b0 + b1*Y0 + b2*D + b3*E + b4*D*E + error

where
Y1 = outcome variable at followup
Y0 = outcome variable at baseline
D and E are as shown above, and D*E = their product.

Of course one could run this model using MIXED, but for users not familiar with MIXED (and I get the impression the OP is not), it will be much easier to run it via UNIANOVA (Analyze - GLM - Univariate, with D and E entered as "fixed factors", and Y0 as a "covariate", and the D*E interaction included).  

Another important question David raised (and I don't recall seeing an answer) is whether there was random assignment to the 4 cells.  

There's my two cents!

R B wrote

From what I can tell, you have one between-subjects variable (Condition)
which has four levels (Diet, Exercise, Diet+Exercise, Control) and one
within-subjects variable (Time) which has two levels (baseline,
post-intervention). This is commonly referred to as a mixed ANOVA (not to
be confused with the MIXED procedure). You could analyze your data by
fitting a general linear model or linear mixed model in SPSS. I don't have
time to write code for you at the moment, but quite frankly, there must be
numerous SPSS examples online. Bottom line is that you need to take into
account correlation among residuals obtained by repeated observations of
each subject. You can write specific contrasts of interest using the
appropriate subcommand. I am always in favor of using the subcommand which
requires that you understand the coefficient (L) matrix.

Ryan

On Thu, Feb 2, 2012 at 12:08 PM, Sur1605 <[hidden email]> wrote:

> Thanks David for such an amazing description. This is exactly how i tried
> doing it before i posted my question on this forum. But everytime i
> analysed
> my data somehow it gave me this message:
> "Post hoc tests are not performed for Diet because there are fewer than
> three groups.
> Post hoc tests are not performed for Ex because there are fewer than three
> groups."
>
> Basically its not able to perform the posthoc analysis. I thought i was
> doing something wrong in feeding in the data in the spreadsheet.
>
> And yes, i measured the same variables before and after the trial. Looks
> like i will have to reconsider the choice of test being used here.
>
> Thanks a lot for your help.
>
> Sb
>
>
>
> --
> View this message in context:
> http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5451327.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
>

[Buhi, Eric]

CONTENTS DELETED

The author has deleted this message.

[Ryan]

DAVID: A multivariate response along with repeated measures does make for an intriguing model. The good news is that such a model can be employed via the MIXED procedure. I have provided code previously on how to fit such a model here:

http://listserv.uga.edu/cgi-bin/wa?A2=ind1104&L=spssx-l&P=R23020 

One could then use the TEST subcommand to answer very interesting research questions! 

BRUCE: I see the rationale for employing an ANCOVA model as you have parameterized it. Admittedly, because I'm often thinking about the optimizing residual and random effects covariance matrices, I tend to gravitate towards certain types of designs. For this particular situation, I thought (although I never stated it) that the residual variance of post-intervention observations could be considerably smaller than baseline, especially for some of the conditions. The MIXED procedure would allow for heterogeneous group*time residual variances if one were to parameterize the model in a particular way. 

Ryan

On Thu, Feb 2, 2012 at 11:22 PM, David Marso <[hidden email]> wrote:

*AND* there are multiple outcome variables measured both pre/post and it
seems that there might be interesting functions which could be explored.
i.e
ANCOVA using simple Pre/Post weight might be overly simplistic.
One might be inclined to look at (Post-Pre)/Pre
Proportionate weight loss.  Just my intuition.
It is easier for very heavy person to lose 15 pounds than a somewhat heavy
person?
OTOH, I don't know diddly squat about the physiological dynamics but it just
seems that theory might toss some interesting dice into the game.

Also :  I would postulate directional hypotheses a-priori .

Weight loss (Weight_Pre -Weight_Post)
Control < Diet ?=? Exercise < Diet+Exercise

Finally,  We can consider the following models:

PostWeight = B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise + Residual
vs
PostWeight = B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise +
B4*PreWeight+Residual2

If Random Assignment then one would expect that Diet,Exercise,Preweight
would be uncorrelated
hence Residual = B4*PreWeight+Residual2

*SO*  ANCOVA reduces the Error SS with a sacrifice of 1 df (assuming Pre and
post are correlated).
----
Enough!?
OTOH: It becomes more interesting when we have a multivariate context.
SEM?
---------------------------------------------------------------------------


Bruce Weaver wrote

>
> But as David noted, the 4 conditions are formed by the 2x2 combination of
> two binary variables.  With the usual coding (i.e., 1 = Yes, 0 = No):
>
> D E   Description
> ---------------------
> 0 0   Control
> 1 0   Diet, No Exercise
> 0 1   Exercise, No Diet
> 1 1   Diet and Exercise
>
> If the outcome variable was measured at both baseline and followup, then
> the preferred approach, arguably, would be a 2x2 ANCOVA, which is
> equivalent to an ordinary least squares regression model as follows:
>
>    Y1 = b0 + b1*Y0 + b2*D + b3*E + b4*D*E + error
>
> where
> Y1 = outcome variable at followup
> Y0 = outcome variable at baseline
> D and E are as shown above, and D*E = their product.
>
> Of course one /could/ run this model using MIXED, but for users not
> familiar with MIXED (and I get the impression the OP is not), it will be
> much easier to run it via UNIANOVA (Analyze - GLM - Univariate, with D and
> E entered as "fixed factors", and Y0 as a "covariate", and the D*E
> interaction included).
>
> Another important question David raised (and I don't recall seeing an
> answer) is whether there was random assignment to the 4 cells.
>
> There's my two cents!
>
>
>
> R B wrote
>>
>> From what I can tell, you have one between-subjects variable (Condition)
>> which has four levels (Diet, Exercise, Diet+Exercise, Control) and one
>> within-subjects variable (Time) which has two levels (baseline,
>> post-intervention). This is commonly referred to as a mixed ANOVA (not to
>> be confused with the MIXED procedure). You could analyze your data by
>> fitting a general linear model or linear mixed model in SPSS. I don't
>> have
>> time to write code for you at the moment, but quite frankly, there must
>> be
>> numerous SPSS examples online. Bottom line is that you need to take into
>> account correlation among residuals obtained by repeated observations of
>> each subject. You can write specific contrasts of interest using the
>> appropriate subcommand. I am always in favor of using the subcommand
>> which
>> requires that you understand the coefficient (L) matrix.
>>
>> Ryan
>>

>> On Thu, Feb 2, 2012 at 12:08 PM, Sur1605 &lt;surabhi1605@&gt; wrote:
>>
>>> Thanks David for such an amazing description. This is exactly how i
>>> tried
>>> doing it before i posted my question on this forum. But everytime i
>>> analysed
>>> my data somehow it gave me this message:
>>> "Post hoc tests are not performed for Diet because there are fewer than
>>> three groups.
>>> Post hoc tests are not performed for Ex because there are fewer than
>>> three
>>> groups."
>>>
>>> Basically its not able to perform the posthoc analysis. I thought i was
>>> doing something wrong in feeding in the data in the spreadsheet.
>>>
>>> And yes, i measured the same variables before and after the trial. Looks
>>> like i will have to reconsider the choice of test being used here.
>>>
>>> Thanks a lot for your help.
>>>
>>> Sb
>>>
>>>
>>>
>>> --
>>> View this message in context:
>>> http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5451327.html
>>> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>>>
>>> =====================
>>> To manage your subscription to SPSSX-L, send a message to
>>> LISTSERV@.UGA (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
>>>
>>
>


--

View this message in context: http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5452678.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

[MICHAEL J TONER]

 

Hello,

 

I will be out of the office starting Monday, February 6 through Tuesday, February 14, 2012.

 

If you are a graduate applicant and have a question about the status of your submitted application, please contact the department to which you applied. They have your application file and know the status of your application.

 

If you are having an issue with the online application system, please contact our Help Desk ([hidden email]).

 

Best,

 

Michael Toner

 

[Sur1605]

Thanks Ryan, David, Bruce for such detailed input. I am not a pro at either stats or SPSS, so i am learning a lot here

Finally, i met a biostatistician today and discussed the issue. She who also suggested a very similar approach as you all are suggesting. According to her mixed model is a fairly strong procedure and will fit perfectly with my study design. But she also said that if in some of the dependent variables i have no missing data, i should go for a repeated measure ANOVA. 

Now since i know what i will be doing with my data and if i also want to check if there was an interaction between diet and exercise, is it right if i set up the SPSS spreadsheet as  

Diet  YES    Exer  YES     GROUP=3 
Diet  YES    Exer  NO      GROUP=1 
Diet   NO    Exer  YES     GROUP=2 
Diet   NO    Exer   NO     GROUP=4 

SPSS data structure

Subject  Diet  Ex    Preweight    Postweight
    1         0       0        
    2         1       0   
    3         0       1   
    4         1       1   

Bruce: Yeah, subjects were divided into the groups using stratified randomization.

Sb

On Fri, Feb 3, 2012 at 7:06 PM, R B [via SPSSX Discussion] <[hidden email]> wrote:

DAVID: A multivariate response along with repeated measures does make for an intriguing model. The good news is that such a model can be employed via the MIXED procedure. I have provided code previously on how to fit such a model here:

http://listserv.uga.edu/cgi-bin/wa?A2=ind1104&L=spssx-l&P=R23020 

One could then use the TEST subcommand to answer very interesting research questions! 

BRUCE: I see the rationale for employing an ANCOVA model as you have parameterized it. Admittedly, because I'm often thinking about the optimizing residual and random effects covariance matrices, I tend to gravitate towards certain types of designs. For this particular situation, I thought (although I never stated it) that the residual variance of post-intervention observations could be considerably smaller than baseline, especially for some of the conditions. The MIXED procedure would allow for heterogeneous group*time residual variances if one were to parameterize the model in a particular way. 

Ryan

On Thu, Feb 2, 2012 at 11:22 PM, David Marso <[hidden email]> wrote:

*AND* there are multiple outcome variables measured both pre/post and it
seems that there might be interesting functions which could be explored.
i.e
ANCOVA using simple Pre/Post weight might be overly simplistic.
One might be inclined to look at (Post-Pre)/Pre
Proportionate weight loss.  Just my intuition.
It is easier for very heavy person to lose 15 pounds than a somewhat heavy
person?
OTOH, I don't know diddly squat about the physiological dynamics but it just
seems that theory might toss some interesting dice into the game.

Also :  I would postulate directional hypotheses a-priori .

Weight loss (Weight_Pre -Weight_Post)
Control < Diet ?=? Exercise < Diet+Exercise

Finally,  We can consider the following models:

PostWeight = B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise + Residual
vs
PostWeight = B0 + B1*Diet + B2*Exercise + B3*Diet*Exercise +
B4*PreWeight+Residual2

If Random Assignment then one would expect that Diet,Exercise,Preweight
would be uncorrelated
hence Residual = B4*PreWeight+Residual2

*SO*  ANCOVA reduces the Error SS with a sacrifice of 1 df (assuming Pre and
post are correlated).
----
Enough!?
OTOH: It becomes more interesting when we have a multivariate context.
SEM?
---------------------------------------------------------------------------


Bruce Weaver wrote

>
> But as David noted, the 4 conditions are formed by the 2x2 combination of
> two binary variables.  With the usual coding (i.e., 1 = Yes, 0 = No):
>
> D E   Description
> ---------------------
> 0 0   Control
> 1 0   Diet, No Exercise
> 0 1   Exercise, No Diet
> 1 1   Diet and Exercise
>
> If the outcome variable was measured at both baseline and followup, then
> the preferred approach, arguably, would be a 2x2 ANCOVA, which is
> equivalent to an ordinary least squares regression model as follows:
>
>    Y1 = b0 + b1*Y0 + b2*D + b3*E + b4*D*E + error
>
> where
> Y1 = outcome variable at followup
> Y0 = outcome variable at baseline
> D and E are as shown above, and D*E = their product.
>

> Of course one /could/ run this model using MIXED, but for users not

> familiar with MIXED (and I get the impression the OP is not), it will be
> much easier to run it via UNIANOVA (Analyze - GLM - Univariate, with D and
> E entered as "fixed factors", and Y0 as a "covariate", and the D*E
> interaction included).
>
> Another important question David raised (and I don't recall seeing an
> answer) is whether there was random assignment to the 4 cells.
>
> There's my two cents!
>
>
>
> R B wrote
>>
>> From what I can tell, you have one between-subjects variable (Condition)
>> which has four levels (Diet, Exercise, Diet+Exercise, Control) and one
>> within-subjects variable (Time) which has two levels (baseline,
>> post-intervention). This is commonly referred to as a mixed ANOVA (not to
>> be confused with the MIXED procedure). You could analyze your data by
>> fitting a general linear model or linear mixed model in SPSS. I don't
>> have
>> time to write code for you at the moment, but quite frankly, there must
>> be
>> numerous SPSS examples online. Bottom line is that you need to take into
>> account correlation among residuals obtained by repeated observations of
>> each subject. You can write specific contrasts of interest using the
>> appropriate subcommand. I am always in favor of using the subcommand
>> which
>> requires that you understand the coefficient (L) matrix.
>>
>> Ryan
>>

>> On Thu, Feb 2, 2012 at 12:08 PM, Sur1605 &lt;surabhi1605@&gt; wrote:
>>
>>> Thanks David for such an amazing description. This is exactly how i
>>> tried
>>> doing it before i posted my question on this forum. But everytime i
>>> analysed
>>> my data somehow it gave me this message:
>>> "Post hoc tests are not performed for Diet because there are fewer than
>>> three groups.
>>> Post hoc tests are not performed for Ex because there are fewer than
>>> three
>>> groups."
>>>
>>> Basically its not able to perform the posthoc analysis. I thought i was
>>> doing something wrong in feeding in the data in the spreadsheet.
>>>
>>> And yes, i measured the same variables before and after the trial. Looks
>>> like i will have to reconsider the choice of test being used here.
>>>
>>> Thanks a lot for your help.
>>>
>>> Sb
>>>
>>>
>>>
>>> --
>>> View this message in context:
>>> http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5451327.html
>>> Sent from the SPSSX Discussion mailing list archive at Nabble.com.
>>>
>>> =====================
>>> To manage your subscription to SPSSX-L, send a message to

>>> LISTSERV@.UGA (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
>>>
>>
>

--

View this message in context: http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5452678.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

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/2x2-Latin-square-design-analysis-help-tp5446710p5455538.html

To unsubscribe from 2x2 Latin square design analysis help, click here.
NAML