mixed ANOVA

[Ratul]

Hi Listers,

Hi all,

I have data on the microbial species from two different body parts of the
same individual and I also have an outcome variable (which is 0/1). I want
to see if there are significant differences in species body parts and the
outcome. Would a repeated measures ANOVA with species as within subject
factor and outcome & bodypart as be suitable test for this type of
experimental design. My dataset somemwhat looks like this

PatientID   Outcome  Part       Sp1   Sp2   Sp3
1             1       skin
              1      tongue
2             0       skin
              0      tongue

Is my data arrangement correct? Is it right to represent the disease
(outcome) of the same individual twice? Any suggestion would be of great
help. I am a student and trying to learn.

Thank you

Nabaneeta

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

[Maguin, Eugene]

Nabaneeta,

I don't understand your data. Did you examine the tongue and the skin of a
number of patients, your subjects, and at each examination location
determine some quantity about species 1 (Sp1), species 2 (Sp2) and species
(Sp3). Is the quantity determined dichotomous category like present/not
present or a quantity like number of cells? And then did you also measure
'outcome' at each examination location? Can the value of 'outcome' be
different for location=tongue and location=skin on the same individual?

Also, please state the question you want your analysis to answer.

Gene Maguin




-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Ratul
Sent: Friday, November 05, 2010 9:09 AM
To: [hidden email]
Subject: mixed ANOVA

Hi Listers,

Hi all,

I have data on the microbial species from two different body parts of the
same individual and I also have an outcome variable (which is 0/1). I want
to see if there are significant differences in species body parts and the
outcome. Would a repeated measures ANOVA with species as within subject
factor and outcome & bodypart as be suitable test for this type of
experimental design. My dataset somemwhat looks like this

PatientID   Outcome  Part       Sp1   Sp2   Sp3
1             1       skin
              1      tongue
2             0       skin
              0      tongue

Is my data arrangement correct? Is it right to represent the disease
(outcome) of the same individual twice? Any suggestion would be of great
help. I am a student and trying to learn.

Thank you

Nabaneeta

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

[Nabaneeta Saha]

Thanks Gene,
I examined the tongue and skin (Part) of patients and quantified the amount of different species (sp1, sp2..sp9). The amount is the number of each species present and is a continous variable. I also have another variable about these patients - which is a disease (outcome) and this is a binary variable (ie. whether they had the disease (1) or not (0). Each patient has one outcome - either disease present (1) or absent (0).
I want to ask the question whether
1)the quantity of a single species varies with location (tongue/skin) and disease outcome (0/1)
2) effect of location (tongue/skin) and oucome (0/1) on all the species as a whole.
 
Nabaneeta

On Fri, Nov 5, 2010 at 9:09 AM, Ratul <[hidden email]> wrote:

Hi Listers,

Hi all,

I have data on the microbial species from two different body parts of the
same individual and I also have an outcome variable (which is 0/1). I want
to see if there are significant differences in species body parts and the
outcome. Would a repeated measures ANOVA with species as within subject
factor and outcome & bodypart as be suitable test for this type of
experimental design. My dataset somemwhat looks like this

PatientID   Outcome  Part       Sp1   Sp2   Sp3
1             1       skin
             1      tongue
2             0       skin
             0      tongue

Is my data arrangement correct? Is it right to represent the disease
(outcome) of the same individual twice? Any suggestion would be of great
help. I am a student and trying to learn.

Thank you

Nabaneeta

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

[parisec]

Hi All,

 

I have archived hundreds of posts to this list beginning in 1999. I search these before posting to see if the topic has posted before.  In several posts, David Nichols referred to SPSS technical papers but these links on the SPSS site no longer work. Is there a new link that has these papers?

 

Thanks

Carol

 

[Maguin, Eugene]

Nabaneeta,

I wanted to post this back to the list so that others can see and, if
appropriate, criticize my advice to you.

I think you actually have a one between (outcome)-two within (body part and
species) sort of design. However, I'd like to suggest that you work up to a
complex analysis that considers all factors. One less complex analysis is to
test for species effects by averaging across body location. I think you'd
need to use a multivariate ANOVA because I don't know whether the species
variables will meet the assumptions  assumed in GLM-repeated measures. The
other less complex analysis is to either average over species or conduct one
analysis for each species. The analysis would be a repeated measures
analysis. I suspect that outcome by location interactions would be
meaningful.

To put all three factors together, I think that a mixed model (Mixed, not
GLM) because I think you can specify a doubly repeated measures design and
allow the residual variance-covariance matrix to have a non standard
structure. You'll have to restructure your data--assuming it is now in a
wide (multivariate) format--using varstocases.

I'm guessing that the command would be (but I'd welcome correction from
folks that know more). Count is the count of a species at a location.

Mixed count by outcome species location/fixed=outcome species location
   outcome*species outcome*location outcome*species*location/
   repeated species location species*location | subj(yourIDvariable)
covtype=(UN).


Gene Maguin






________________________________

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Nabaneeta Saha
Sent: Friday, November 05, 2010 5:52 PM
To: [hidden email]
Subject: Re: mixed ANOVA


Thanks Gene,
I examined the tongue and skin (Part) of patients and quantified the amount
of different species (sp1, sp2..sp9). The amount is the number of each
species present and is a continous variable. I also have another variable
about these patients - which is a disease (outcome) and this is a binary
variable (ie. whether they had the disease (1) or not (0). Each patient has
one outcome - either disease present (1) or absent (0).
I want to ask the question whether
1)the quantity of a single species varies with location (tongue/skin) and
disease outcome (0/1)
2) effect of location (tongue/skin) and oucome (0/1) on all the species as a
whole.

Nabaneeta




>>Nabaneeta,

I don't understand your data. Did you examine the tongue and the skin of a
number of patients, your subjects, and at each examination location
determine some quantity about species 1 (Sp1), species 2 (Sp2) and species
(Sp3). Is the quantity determined dichotomous category like present/not
present or a quantity like number of cells? And then did you also measure
'outcome' at each examination location? Can the value of 'outcome' be
different for location=tongue and location=skin on the same individual?

Also, please state the question you want your analysis to answer.

Gene Maguin



On Fri, Nov 5, 2010 at 9:09 AM, Ratul <[hidden email]> wrote:

        Hi Listers,
        Hi all,

        I have data on the microbial species from two different body parts
of the
        same individual and I also have an outcome variable (which is 0/1).
I want
        to see if there are significant differences in species body parts
and the
        outcome. Would a repeated measures ANOVA with species as within
subject
        factor and outcome & bodypart as be suitable test for this type of
        experimental design. My dataset somemwhat looks like this

        PatientID   Outcome  Part       Sp1   Sp2   Sp3
        1             1       skin
                     1      tongue
        2             0       skin
                     0      tongue

        Is my data arrangement correct? Is it right to represent the disease
        (outcome) of the same individual twice? Any suggestion would be of
great
        help. I am a student and trying to learn.

        Thank you

        Nabaneeta

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