fixed effect with ancova

[David Marso]

IMNSHO,
Using Length as a "Covariate" in this case is *WRONG*.  It is correlated with the treatments (diet).
I should think of this perhaps as a Bivariate response (use both weight and length as dependent variables).
OTOH, I don't know WTF I am talking about...Sounds like the OP is fishing!!!
HTH, David

ro wrote

Hi all,
Gene, Longcms is the lenght, this was the variate that have good correlation with weight, in a GLM analysis i can add a covariate...for me this is an ancova analysis, i am right?...

my data was checked twice for 2 different people, but iam gonna review again...

Art, i plot data and lines are not parallel, interaction exist!, but i am not agree with a point (mean), i will check again.

anyway, can i add interaction to a lineal model (remember that i am try to find the best lineal model that adjust to my data), i only thought that fixed effects and covariates are main effects to my model...

Thanks Gene, Rich and Art for you support and you spend time, i will check all data (previously was checked but i am gonna do again with original data...).
any news i ll stay in touch...

[Maguin, Eugene]

IMNSHO=In My Not So Humble Opinion??

Question is when was longcms measured  Before the fish started on their
diet, as a proxy for intial weight, or after when they were weighed.

Gene Maguin

-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
David Marso
Sent: Tuesday, August 23, 2011 10:39 AM
To: [hidden email]
Subject: Re: fixed effect with ancova

IMNSHO,
Using Length as a "Covariate" in this case is *WRONG*.  It is correlated
with the treatments (diet).
I should think of this perhaps as a Bivariate response (use both weight and
length as dependent variables).
OTOH, I don't know WTF I am talking about...Sounds like the OP is fishing!!!
HTH, David

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

[ro]

longcms was measured when i finish my experiment, all data in this analysis was in the same time, when i finish my experiment, others measures (weight and lenght each 3 months) are in other datasheet..

[Maguin, Eugene]

Ro,

David's right (IMVCO). Longcms has to be highly correlated with weight. You
have two DVs. So, two choices: 1) analyze each DV separately as an ANOVA; 2)
analyze both DVs together as a multivariate ANOVA, switching over to GLM for
this.

Gene Maguin



-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of ro
Sent: Tuesday, August 23, 2011 11:14 AM
To: [hidden email]
Subject: Re: fixed effect with ancova

longcms was measured when i finish my experiment, all data in this analysis
was in the same time, when i finish my experiment, others measures (weight
and lenght each 3 months) are in other datasheet..

--
View this message in context:
http://spssx-discussion.1045642.n5.nabble.com/fixed-effect-with-ancova-tp471
3723p4727131.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

[ro]

Hi

i finish to check all data, then my tank numbers was reduced to 21 and 7 diets (the design was the same that before: 3 tanks by each diet), the number of fishes in each tank is variable, the minimum number of fishes in each tank is 6 but the most of tanks have 12 fishes (unbalanced?)...

i am not finish to analysis my data (all data normalized), but it appear have the same problem, i can not find tank effect and interaction and diet effect disappear, other problem is that degree of freedom (ds) in the analysis only compute 14 tanks From 21!! and 0 in diet!!...why? i did the same analysis with genstat software and compute 20 df and 6 diet (exactly the same data) and recognized diet and tank effect but no interaction... when i plot means for tank (x) and diet (y) i see interaction very clear, i thought that the problem is in diet effect, because spss dont compute diet effect...

i attach pdf with results

[David Marso]

I believe we have clearly established that ANCOVA is *NOT* an appropriate model for this data.
Length and Weight are highly correlated and both are affected by diet and tank.  
How are your data organized.  At one point is seemed as though you had aggregated data by tank, but that does not seem to be the case.  IIRC you also have multiple measures over time points (but that might be a different thread).
What do you mean by "all data normalized"?  Do you mean they have been verified and possibly rectified or have you used some function to "normalize" the data?
What happens if you use GLM to jointly model both weight and length as dependent variables with tank and diet as fixed factors.  Are you sure you want to treat these as fixed factors?

ro wrote

Hi

i finish to check all data, then my tank numbers was reduced to 21 and 7 diets (the design was the same that before: 3 tanks by each diet), the number of fishes in each tank is variable, the minimum number of fishes in each tank is 6 but the most of tanks have 12 fishes (unbalanced?)...

i am not finish to analysis my data (all data normalized), but it appear have the same problem, i can not find tank effect and interaction and diet effect disappear, other problem is that degree of freedom (ds) in the analysis only compute 14 tanks From 21!! and 0 in diet!!...why? i did the same analysis with genstat software and compute 20 df and 6 diet (exactly the same data) and recognized diet and tank effect but no interaction... when i plot means for tank (x) and diet (y) i see interaction very clear, i thought that the problem is in diet effect, because spss dont compute diet effect...

i attach pdf with results

[Jon K Peck]

You can't any anything between the GGRAPH command and the BEGIN GPL line.  Delete the comment and blank lines, and you should be fine.

Jon Peck (no "h")
Senior Software Engineer, IBM
[hidden email]
new phone: [phone]




From:        Eins Bernardo <[hidden email]>
To:        [hidden email]
Date:        08/23/2011 03:54 AM
Subject:        [SPSSX-L] What is wrong in this syntax?
Sent by:        "SPSSX(r) Discussion" <[hidden email]>

---

I tried to run the syntax below using version 18 to generate Forest Plot.  Unfortunately, it does not work.  A warning appeared on the screen as follows:
 

Warnings

An inline graph specification was expected but not found.

Execution of this command stops.

 
I appreciate any help.
 Eins

* Sample dataset (processed data )*.
DATA LIST LIST/ trial(F4) year(A5) study(A10) measure cilow ciup percwi(4 F8.3).
BEGIN DATA
1 "1989" "Hodnett  " .502 .262  .962   4.940
2 "1991" "Kennell  " .352 .216  .575   8.694
3 "1992" "Bréart-Fr" .785 .483 1.276   8.878
4 "1992" "Bréart-Bg" .811 .653 1.007  44.598
5 "1997" "Gagnon   " .867 .573 1.311  12.237
6 "1998" "Langer   " .280 .203  .384  20.654
7 "    " "Total    " .594 .514  .687 100.000
END DATA.

*/ Assuming you have SPSS 14:

SORT CASES BY trial(D).
STRING YearAndStudy(A30).
COMPUTE YearAndStudy=CONCAT(RTRIM(year)," ",study).
COMPUTE RefLine=1.

GGRAPH
 /GRAPHDATASET NAME="graphdataset" VARIABLES=YearAndStudy ciup cilow measure percwi RefLine
 MISSING=LISTWISE REPORTMISSING=NO
/GRAPHSPEC SOURCE=INLINE.

* If the effect you are measuring is OR or RR, then this syntax can
do the task *.

BEGIN GPL
SOURCE: s=userSource(id("graphdataset"))
DATA:   YearAndStudy=col(source(s),   name("YearAndStudy"), unit.category())
DATA:   ciup=col(source(s),    name("ciup"))
DATA:   cilow=col(source(s),   name("cilow"))
DATA:   measure=col(source(s), name("measure"))
DATA:   percwi=col(source(s),  name("percwi"))
DATA:   RefLine=col(source(s), name("RefLine"))
COORD:  transpose(rect(dim(1,2), transpose()))
GUIDE:  axis(dim(2), label(" Favours treatment                 Favours Control"))
SCALE:  cat(dim(1))
SCALE:  log(dim(2))
ELEMENT: interval(position(region.spread.range(YearAndStudy*(cilow+ciup))),
 shape(shape.line), color(color.black))
ELEMENT: point(position(YearAndStudy*measure), shape(shape.square),
 size(percwi), color.interior(color.black))
ELEMENT: line(position(YearAndStudy*RefLine), shape(shape.line))
END GPL.

[ro]

Ok David thanks for your comments, yes Length and Weight are highly correlated and both are affected by diet and tank, i supposed that this was one ancova analysis (some papers said the same...), i dont know if a bivariate analysis is anyway one ancova analysis, but anyway i will do one bivariate analysis like you suggest...

all data normalized mean all data have been verified and possibly rectified for THIS analysis...(in other analysis i have used tipical functions: log10, sqrt...)

When i run a GLm that you suggest i can not find significatn differences, and have always the same problem that i told you earlier, SPSS can not compute diet effect and df in tank are very low 14 (are 21 tanks...) its normal?

my datasheet looks like:
fish         diet    tank  weight  lenght
1             1       1       #          #
2             1       1       #          #
...15        1       1       #          #
1             1       2       #          #
2             1       2       #          #
...12        1       2      ...          ...
1             1       3
2             1       3
...9        1       3
1             2       1
2             2       1
...6        2       1
1             2       2
2             2       2
...12        2       2
1             2       3
2             2       3
...1        2       3
...until diet 7 ...

[ro]

sorry i mistake the results table...here is the correct result...

[David Marso]

The way you have your data coded for tank is confusing. You have codes 1,2,3 for each of diets 1,2.
Are these different brands of tank? Same variety of tank?  If you indeed have 21 tanks which are *distinct* but homogenous in attributes  they should be coded 1,2,...21.  Also note that you CANNOT estimate the interaction of diet and tank since they are not CROSSED (if I'm not mistaken diet is nested in tank but so far the description of your experiment is muddled).  I suspect the only way you are likely to receive any definitive resolution from posting here is to Attach a copy of your actual data.  Something fishy is going on here as if you have 21 tanks you should end up with 21-1=20 df for tank.  
Take a look at the manual for Nested effects.  I suspect rather than
/DESIGN diet tank diet*tank
you might want
/DESIGN diet tank diet(tank)
So,  something is fishy!
Revisit your Linear Statistical Models, Regression, ANOVA and MANOVA texts and please clarify the exact nature of your experimental design.
HTH, David
--

ro wrote

Ok David thanks for your comments, yes Length and Weight are highly correlated and both are affected by diet and tank, i supposed that this was one ancova analysis (some papers said the same...), i dont know if a bivariate analysis is anyway one ancova analysis, but anyway i will do one bivariate analysis like you suggest...

all data normalized mean all data have been verified and possibly rectified for THIS analysis...(in other analysis i have used tipical functions: log10, sqrt...)

When i run a GLm that you suggest i can not find significatn differences, and have always the same problem that i told you earlier, SPSS can not compute diet effect and df in tank are very low 14 (are 21 tanks...) its normal?

my datasheet looks like:
fish         diet    tank  weight  lenght
1             1       1       #          #
2             1       1       #          #
...15        1       1       #          #
1             1       2       #          #
2             1       2       #          #
...12        1       2      ...          ...
1             1       3
2             1       3
...9        1       3
1             2       1
2             2       1
...6        2       1
1             2       2
2             2       2
...12        2       2
1             2       3
2             2       3
...1        2       3
...until diet 7 ...

[ro]

sorry i mistake, you are right this table was confusing...
yep i agree with you, if you look my first ask in the forum i ask for one nested design i did suspect that maybe this could be a good solution... so i wil do: /DESIGN diet tank diet(tank)

all tanks are the same brand and specifications they received the same flux of water and oxigen, only differ from diet treatment...

my datasheet REALLY looks like:
fish         diet    tank  weight  lenght
1             1       1       #          #
2             1       1       #          #
...15        1       1       #          #
1             1       2       #          #
2             1       2       #          #
...12        1       2      ...          ...
1             1       3
2             1       3
...9          1       3
1             2       4
2             2       4
...6          2       4
1             2       5
2             2       5
...12        2       5
1             2       6
2             2       6
...1          2       6

[ro]

Hi all,
I just write to give thanks to all who give me comments about my post, now i am more clear about my model and i have good result in my analysis...

Regards
Rodrigo