Hi Kay,
I also get significant values (both from adonis and anova) when
running your code. I'm currently running my own adonis analysis which
most have significant results (this has been worrying me for some
time).
I do recall some R discussion about general R setting/value which
might set all the values analysed to be logarithmic. If there is this
kind of setting, does anyone know what it is and how to set it off?
I'm using Revolution R Enterprise 4.3.0 (14.4.2011) Academic version,
R version 2.12.2 (2011-02-25)
-Kari
My results:
for (i in 1:4) {print(summary(aov(spdf[, i] ~ treat * imp)))}
Df Sum Sq Mean Sq F value Pr(>F)
treat 2 166.017 83.008 1316.5029 <2e-16 ***
imp 1 15.978 15.978 253.4022 <2e-16 ***
treat:imp 2 0.021 0.011 0.1666 0.847
Residuals 54 3.405 0.063
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Df Sum Sq Mean Sq F value Pr(>F)
treat 2 155.836 77.918 1323.0303 <2e-16 ***
imp 1 13.185 13.185 223.8864 <2e-16 ***
treat:imp 2 0.237 0.119 2.0159 0.1431
Residuals 54 3.180 0.059
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Df Sum Sq Mean Sq F value Pr(>F)
treat 2 158.672 79.336 1091.8367 < 2e-16 ***
imp 1 16.069 16.069 221.1473 < 2e-16 ***
treat:imp 2 0.363 0.182 2.4987 0.09165 .
Residuals 54 3.924 0.073
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Df Sum Sq Mean Sq F value Pr(>F)
treat 2 159.194 79.597 943.528 <2e-16 ***
imp 1 14.289 14.289 169.383 <2e-16 ***
treat:imp 2 0.093 0.047 0.554 0.5779
Residuals 54 4.555 0.084
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
RevoMods:::revoFix(x)adonis(spdf ~ treat * imp, dist = "euclidean")
Call:
adonis(formula = spdf ~ treat * imp, dist = "euclidean")
Df SumsOfSqs MeanSqs F.Model R2 Pr(>F)
treat 2 3.4133 1.70666 640.19 0.80850 0.001 ***
imp 1 0.3150 0.31504 118.18 0.07462 0.001 ***
treat:imp 2 0.3495 0.17473 65.54 0.08278 0.001 ***
Residuals 54 0.1440 0.00267 0.03410
Total 59 4.2218 1.00000
On Sun, Jun 26, 2011 at 8:00 PM,
<r-sig-ecology-request at r-project.org> wrote:
Send R-sig-ecology mailing list submissions to
? ? ? ?r-sig-ecology at r-project.org
To subscribe or unsubscribe via the World Wide Web, visit
? ? ? ?https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
or, via email, send a message with subject or body 'help' to
? ? ? ?r-sig-ecology-request at r-project.org
You can reach the person managing the list at
? ? ? ?r-sig-ecology-owner at r-project.org
When replying, please edit your Subject line so it is more specific
than "Re: Contents of R-sig-ecology digest..."
Today's Topics:
? 1. Re: adonis question (Kay Cecil Cichini)
----------------------------------------------------------------------
Message: 1
Date: Sat, 25 Jun 2011 12:20:44 +0200
From: Kay Cecil Cichini <Kay.Cichini at uibk.ac.at>
To: Etienne Laliberte <etiennelaliberte at gmail.com>
Cc: r-sig-ecology at r-project.org
Subject: Re: [R-sig-eco] adonis question
Message-ID: <20110625122044.68992dwaacyxu6ps at web-mail.uibk.ac.at>
Content-Type: text/plain; charset=ISO-8859-1; DelSp="Yes";
? ? ? ?format="flowed"
hello etienne,
i tried many times and it's alway significant.
i also added:
par(mfrow = c(2, 2))
for (i in 1:4) {print(interaction.plot(treat, imp, spdf[, i]))}
...that did not show an interaction for any set of random numbers i tried.
yours,
kay
Zitat von Etienne Laliberte <etiennelaliberte at gmail.com>:
Dear Kay,
I ran your code and found no significant interactions. You just got
"unlucky" with rnorm(), as can happen. Try it again.
Cheers,
Etienne
-----Original Message-----
From: r-sig-ecology-bounces at r-project.org
[mailto:r-sig-ecology-bounces at r-project.org] On Behalf Of Kay Cecil Cichini
Sent: Friday, 24 June 2011 9:31 PM
To: r-sig-ecology
Subject: [R-sig-eco] adonis question
hi all,
i played around with adonis and came across a result that i can't
explain - more precisely i get a sign interaction-effect where i
suppose there should be none:
library(vegan)
# species matrix/df
spdf <- matrix(NA, 60, 4, dimnames = list(1:60, c("sp1", "sp2", "sp3",
"sp4")))
spdf <- as.data.frame(spdf)
# 1st factor = treatment:
treat <- gl(3, 20, labels = paste(1:3, "t", sep=""))
# 2nd factor = impact:
imp <- rep(gl(2, 10, labels = c("yes", "no")), 3)
# simulating effect -
# simulation will add similar effect
# across level combinations, no interactions:
eff <- sort(rep(1:6, 10))
# add random noise:
spdf$sp1 = eff + rnorm(60, 0, 0.25)
spdf$sp2 = eff + rnorm(60, 0, 0.25)
spdf$sp3 = eff + rnorm(60, 0, 0.25)
spdf$sp4 = eff + rnorm(60, 0, 0.25)
# difference imp:yes vs imp:no is the same across all levels of treatment
aggregate(sp1 ~ imp * treat, data = spdf, FUN = mean)
# univariate ANOVAs with no interactions, as expected:
for (i in 1:4) {print(summary(aov(spdf[, i] ~ treat * imp)))}
# PERMANOVA - strangly gives sign. interaction (??)
adonis(spdf ~ treat * imp, dist = "euclidean")
thanks in advance,
kay