An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20121005/68fcbf16/attachment.pl>
Dúvida função Anova pacote car - Medidas repetidas
5 messages · Diego Pujoni, John Fox, Rui Barradas
An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20121005/4fc021af/attachment.pl>
Dear Diego,
This is close enough to Spanish for me to understand it (I think).
Using Anova() in the car package for repeated-measures designs requires a
multivariate linear model for all of the responses, which in turn requires
that the data set be in "wide" format, with each response as a variable. In
your case, there are two crossed within-subjects factors and no
between-subjects factors. If this understanding is correct (but see below),
then you could proceed as follows, where the crucial step is reshaping the
data from "long" to "wide":
------------- snip --------------
Pa2$type.day <- with(Pa2, paste(Type, Day, sep="."))
(Wide <- reshape(Pa2, direction="wide", v.names="logbiovolume",
idvar="Replicate", timevar="type.day", drop=c("Type", "Day")))
day <- ordered(rep(c(0, 2, 4), each=2))
type <- factor(rep(c("c", "t"), 3))
(idata <- data.frame(day, type))
mod <- lm(cbind(logbiovolume.c.0, logbiovolume.t.0, logbiovolume.c.2,
logbiovolume.t.2, logbiovolume.c.4, logbiovolume.t.4) ~ 1, data=Wide)
Anova(mod, idata=idata, idesign=~day*type)
------------- snip --------------
This serves to analyze the data that you showed; you'll have to adapt it for
the full data set.
I'm assuming that the "replicates" are independent units, and that the
design is therefore entirely within replicate. If that's wrong, then the
analysis I've suggested is also incorrect.
I hope this helps,
John
-----------------------------------------------
John Fox
Senator McMaster Professor of Social Statistics
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On Behalf Of Diego Pujoni
Sent: Friday, October 05, 2012 9:57 AM
To: r-help at r-project.org
Subject: [R] D?vida fun??o Anova pacote car - Medidas repetidas
Ola pessoal, estou realizando uma ANOVA com medidas repetidas e estou
utilizando a fungco "Anova" do pacote "car".
Medi o biovolume de algas a cada dois dias durante 10 dias (no banco de
dados abaixo ss coloquei ati o 40 dia). Tenho 2 tratamentos ("c","t") e
o
experimento foi realizado em triplicas ("A","B","C").
Pa2
Day Type Replicate logbiovolume
1 0 c A 19.34
2 0 c B 18.27
3 0 c C 18.56
4 0 t A 18.41
5 0 t B 18.68
6 0 t C 18.86
7 2 c A 18.81
8 2 c B 18.84
9 2 c C 18.52
10 2 t A 18.29
11 2 t B 17.91
12 2 t C 17.67
13 4 c A 19.16
14 4 c B 18.85
15 4 c C 19.36
16 4 t A 19.05
17 4 t B 19.09
18 4 t C 18.26
.
.
.
Pa2.teste = within(Pa2,{group = factor(Type)
time = factor(Day)
id = factor(Replicate)})
matrix =
with(Pa2.teste,cbind(Pa2[,VAR][group=="c"],Pa2[,VAR][group=="t"]))
matrix
[,1] [,2]
[1,] 19.34 18.41
[2,] 18.27 18.68
[3,] 18.56 18.86
[4,] 18.81 18.29
[5,] 18.84 17.91
[6,] 18.52 17.67
[7,] 19.16 19.05
[8,] 18.85 19.09
[9,] 19.36 18.26
[10,] 19.63 18.96
[11,] 19.94 18.06
[12,] 19.54 18.37
[13,] 19.98 17.96
[14,] 20.99 17.93
[15,] 20.45 17.74
[16,] 21.12 17.60
[17,] 21.66 17.33
[18,] 21.51 18.12
model <- lm(matrix ~ 1)
design <- factor(c("c","t"))
options(contrasts=c("contr.sum", "contr.poly"))
aov <- Anova(model, idata=data.frame(design), idesign=~design,
type="III")
summary(aov, multivariate=F)
Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
SS num Df Error SS den Df F Pr(>F)
(Intercept) 12951.2 1 6.3312 17 34775.336 < 2.2e-16 ***
design 19.1 1 17.3901 17 18.697 0.0004606 ***
---
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1
O problema i que eu acho que esta fungco nco esta levando em
consideragco
os dias, nem as riplicas. Como fago para introduzir isto na analise.
Vocjs
conhecem alguma fungco correspondente nco paramitrica para este teste?
Tipo
um teste de Friedman com dois grupos (tratamento e riplica) e um bloco
(tempo)?
Muito Obrigado
Diego PJ
[[alternative HTML version deleted]]
Hello, Yes, your Spanish is close enough to Portuguese for you to understand it. I thought it was homework and didn't read untill the end. Apologies to Diego, and thanks to John. Rui Barradas Em 05-10-2012 22:48, John Fox escreveu:
Dear Diego,
This is close enough to Spanish for me to understand it (I think).
Using Anova() in the car package for repeated-measures designs requires a
multivariate linear model for all of the responses, which in turn requires
that the data set be in "wide" format, with each response as a variable. In
your case, there are two crossed within-subjects factors and no
between-subjects factors. If this understanding is correct (but see below),
then you could proceed as follows, where the crucial step is reshaping the
data from "long" to "wide":
------------- snip --------------
Pa2$type.day <- with(Pa2, paste(Type, Day, sep="."))
(Wide <- reshape(Pa2, direction="wide", v.names="logbiovolume",
idvar="Replicate", timevar="type.day", drop=c("Type", "Day")))
day <- ordered(rep(c(0, 2, 4), each=2))
type <- factor(rep(c("c", "t"), 3))
(idata <- data.frame(day, type))
mod <- lm(cbind(logbiovolume.c.0, logbiovolume.t.0, logbiovolume.c.2,
logbiovolume.t.2, logbiovolume.c.4, logbiovolume.t.4) ~ 1, data=Wide)
Anova(mod, idata=idata, idesign=~day*type)
------------- snip --------------
This serves to analyze the data that you showed; you'll have to adapt it for
the full data set.
I'm assuming that the "replicates" are independent units, and that the
design is therefore entirely within replicate. If that's wrong, then the
analysis I've suggested is also incorrect.
I hope this helps,
John
-----------------------------------------------
John Fox
Senator McMaster Professor of Social Statistics
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On Behalf Of Diego Pujoni
Sent: Friday, October 05, 2012 9:57 AM
To: r-help at r-project.org
Subject: [R] D?vida fun??o Anova pacote car - Medidas repetidas
Ola pessoal, estou realizando uma ANOVA com medidas repetidas e estou
utilizando a fungco "Anova" do pacote "car".
Medi o biovolume de algas a cada dois dias durante 10 dias (no banco de
dados abaixo ss coloquei ati o 40 dia). Tenho 2 tratamentos ("c","t") e
o
experimento foi realizado em triplicas ("A","B","C").
Pa2
Day Type Replicate logbiovolume
1 0 c A 19.34
2 0 c B 18.27
3 0 c C 18.56
4 0 t A 18.41
5 0 t B 18.68
6 0 t C 18.86
7 2 c A 18.81
8 2 c B 18.84
9 2 c C 18.52
10 2 t A 18.29
11 2 t B 17.91
12 2 t C 17.67
13 4 c A 19.16
14 4 c B 18.85
15 4 c C 19.36
16 4 t A 19.05
17 4 t B 19.09
18 4 t C 18.26
.
.
.
Pa2.teste = within(Pa2,{group = factor(Type)
time = factor(Day)
id = factor(Replicate)})
matrix =
with(Pa2.teste,cbind(Pa2[,VAR][group=="c"],Pa2[,VAR][group=="t"]))
matrix
[,1] [,2]
[1,] 19.34 18.41
[2,] 18.27 18.68
[3,] 18.56 18.86
[4,] 18.81 18.29
[5,] 18.84 17.91
[6,] 18.52 17.67
[7,] 19.16 19.05
[8,] 18.85 19.09
[9,] 19.36 18.26
[10,] 19.63 18.96
[11,] 19.94 18.06
[12,] 19.54 18.37
[13,] 19.98 17.96
[14,] 20.99 17.93
[15,] 20.45 17.74
[16,] 21.12 17.60
[17,] 21.66 17.33
[18,] 21.51 18.12
model <- lm(matrix ~ 1)
design <- factor(c("c","t"))
options(contrasts=c("contr.sum", "contr.poly"))
aov <- Anova(model, idata=data.frame(design), idesign=~design,
type="III")
summary(aov, multivariate=F)
Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
SS num Df Error SS den Df F Pr(>F)
(Intercept) 12951.2 1 6.3312 17 34775.336 < 2.2e-16 ***
design 19.1 1 17.3901 17 18.697 0.0004606 ***
---
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1
O problema i que eu acho que esta fungco nco esta levando em
consideragco
os dias, nem as riplicas. Como fago para introduzir isto na analise.
Vocjs
conhecem alguma fungco correspondente nco paramitrica para este teste?
Tipo
um teste de Friedman com dois grupos (tratamento e riplica) e um bloco
(tempo)?
Muito Obrigado
Diego PJ
[[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Sorry, Phone, daughter, forgot to sign. Rui Barradas Em 06-10-2012 12:28, Rui Barradas escreveu:
Hello, Yes, your Spanish is close enough to Portuguese for you to understand it. I thought it was homework and didn't read untill the end. Apologies to Diego, and thanks to John. Rui Barradas Em 05-10-2012 22:48, John Fox escreveu:
Dear Diego,
This is close enough to Spanish for me to understand it (I think).
Using Anova() in the car package for repeated-measures designs
requires a
multivariate linear model for all of the responses, which in turn
requires
that the data set be in "wide" format, with each response as a
variable. In
your case, there are two crossed within-subjects factors and no
between-subjects factors. If this understanding is correct (but see
below),
then you could proceed as follows, where the crucial step is
reshaping the
data from "long" to "wide":
------------- snip --------------
Pa2$type.day <- with(Pa2, paste(Type, Day, sep="."))
(Wide <- reshape(Pa2, direction="wide", v.names="logbiovolume",
idvar="Replicate", timevar="type.day", drop=c("Type", "Day")))
day <- ordered(rep(c(0, 2, 4), each=2))
type <- factor(rep(c("c", "t"), 3))
(idata <- data.frame(day, type))
mod <- lm(cbind(logbiovolume.c.0, logbiovolume.t.0, logbiovolume.c.2,
logbiovolume.t.2, logbiovolume.c.4, logbiovolume.t.4) ~ 1, data=Wide)
Anova(mod, idata=idata, idesign=~day*type)
------------- snip --------------
This serves to analyze the data that you showed; you'll have to adapt
it for
the full data set.
I'm assuming that the "replicates" are independent units, and that the
design is therefore entirely within replicate. If that's wrong, then the
analysis I've suggested is also incorrect.
I hope this helps,
John
-----------------------------------------------
John Fox
Senator McMaster Professor of Social Statistics
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
-----Original Message-----
From: r-help-bounces at r-project.org
[mailto:r-help-bounces at r-project.org]
On Behalf Of Diego Pujoni
Sent: Friday, October 05, 2012 9:57 AM
To: r-help at r-project.org
Subject: [R] D?vida fun??o Anova pacote car - Medidas repetidas
Ola pessoal, estou realizando uma ANOVA com medidas repetidas e estou
utilizando a fungco "Anova" do pacote "car".
Medi o biovolume de algas a cada dois dias durante 10 dias (no banco de
dados abaixo ss coloquei ati o 40 dia). Tenho 2 tratamentos ("c","t") e
o
experimento foi realizado em triplicas ("A","B","C").
Pa2
Day Type Replicate logbiovolume
1 0 c A 19.34
2 0 c B 18.27
3 0 c C 18.56
4 0 t A 18.41
5 0 t B 18.68
6 0 t C 18.86
7 2 c A 18.81
8 2 c B 18.84
9 2 c C 18.52
10 2 t A 18.29
11 2 t B 17.91
12 2 t C 17.67
13 4 c A 19.16
14 4 c B 18.85
15 4 c C 19.36
16 4 t A 19.05
17 4 t B 19.09
18 4 t C 18.26
.
.
.
Pa2.teste = within(Pa2,{group = factor(Type)
time = factor(Day)
id = factor(Replicate)})
matrix =
with(Pa2.teste,cbind(Pa2[,VAR][group=="c"],Pa2[,VAR][group=="t"]))
matrix
[,1] [,2]
[1,] 19.34 18.41
[2,] 18.27 18.68
[3,] 18.56 18.86
[4,] 18.81 18.29
[5,] 18.84 17.91
[6,] 18.52 17.67
[7,] 19.16 19.05
[8,] 18.85 19.09
[9,] 19.36 18.26
[10,] 19.63 18.96
[11,] 19.94 18.06
[12,] 19.54 18.37
[13,] 19.98 17.96
[14,] 20.99 17.93
[15,] 20.45 17.74
[16,] 21.12 17.60
[17,] 21.66 17.33
[18,] 21.51 18.12
model <- lm(matrix ~ 1)
design <- factor(c("c","t"))
options(contrasts=c("contr.sum", "contr.poly"))
aov <- Anova(model, idata=data.frame(design), idesign=~design,
type="III")
summary(aov, multivariate=F)
Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
SS num Df Error SS den Df F Pr(>F)
(Intercept) 12951.2 1 6.3312 17 34775.336 < 2.2e-16 ***
design 19.1 1 17.3901 17 18.697 0.0004606 ***
---
Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1
O problema i que eu acho que esta fungco nco esta levando em
consideragco
os dias, nem as riplicas. Como fago para introduzir isto na analise.
Vocjs
conhecem alguma fungco correspondente nco paramitrica para este teste?
Tipo
um teste de Friedman com dois grupos (tratamento e riplica) e um bloco
(tempo)?
Muito Obrigado
Diego PJ
[[alternative HTML version deleted]]
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
______________________________________________ R-help at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.