Quantcast

Tukey HSD following lme

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Tukey HSD following lme

C.Schaedel
Hi everyone

I'm using Tukey HSD as post-hoc test following a lme analysis. I'm
measuring hemicelluloses in different species treated with three
different CO2 concentrations (l=low, m=medium, h=high). The whole
experiment is a split-plot design and the Tukey-function from the
package multcomp is suitable for lme-analysis with random factors.

The analysis works fine but I get a non significant lme-result and if I
do a Tukey afterwards (I know that one usually does a Tukey following a
significant anova-result) I get highly significant p-values for two
multiple comparisons.

How can this happen? How can the p-values from the Tukey become
significant when the lme-model wasn't?


the data are: d

  Species  Block  CO2   hemicell
  Ps         a     l    9.027363
  Ps         b     l    9.647537
  Ps         a     m   10.051916
  Ps         b     m   10.112294
  Ps         a     h   10.342162
  Ps         b     h   10.303091


my lme model:
anova(hc<-lme(asin(sqrt(0.01*hemicell))~CO2,random=~1|Block/CO2,data=d))

             numDF denDF   F-value p-value
(Intercept)     1     2 30403.248  <.0001
CO2             2     2     8.051  0.1105


Tukey with the lme-object:

summary(glht(hc, linfct=mcp(CO2="Tukey")))

yielding:

Linear Hypotheses:
            Estimate Std. Error z value p value
m - l == 0 0.012616   0.004317   2.922 0.00963 **
h - l == 0 0.016590   0.004317   3.843 < 0.001 ***
h - m == 0 0.003973   0.004317   0.920 0.62738
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
(Adjusted p values reported -- single-step method)


Thank you very much for your help

Christina



--
Christina Schädel
Institute of Botany
University of Basel
Schönbeinstrasse 6
CH-4056 Basel
ph. +41 61 267 35 06
fax +41 61 267 29 80
E-Mail [hidden email]

______________________________________________
[hidden email] 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.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Tukey HSD following lme

Mark Difford
Hi Christina,

>> How can this happen? How can the p-values from the Tukey become
>> significant when the lme-model wasn't?

The link below, with an explanation by Prof. Fox is relevant to your question:

http://www.nabble.com/Strange-results-with-anova.glm()-td13471998.html#a13475563

Another way to see the error of your ways is to fit your model using the default treatment contrasts and then do a Dunnett post hoc using glht:

##
summary(glht(hc, linfct=mcp(CO2="Dunnett")))

Regards, Mark.

C.Schaedel wrote
Hi everyone

I'm using Tukey HSD as post-hoc test following a lme analysis. I'm
measuring hemicelluloses in different species treated with three
different CO2 concentrations (l=low, m=medium, h=high). The whole
experiment is a split-plot design and the Tukey-function from the
package multcomp is suitable for lme-analysis with random factors.

The analysis works fine but I get a non significant lme-result and if I
do a Tukey afterwards (I know that one usually does a Tukey following a
significant anova-result) I get highly significant p-values for two
multiple comparisons.

How can this happen? How can the p-values from the Tukey become
significant when the lme-model wasn't?


the data are: d

  Species  Block  CO2   hemicell
  Ps         a     l    9.027363
  Ps         b     l    9.647537
  Ps         a     m   10.051916
  Ps         b     m   10.112294
  Ps         a     h   10.342162
  Ps         b     h   10.303091


my lme model:
anova(hc<-lme(asin(sqrt(0.01*hemicell))~CO2,random=~1|Block/CO2,data=d))

             numDF denDF   F-value p-value
(Intercept)     1     2 30403.248  <.0001
CO2             2     2     8.051  0.1105


Tukey with the lme-object:

summary(glht(hc, linfct=mcp(CO2="Tukey")))

yielding:

Linear Hypotheses:
            Estimate Std. Error z value p value
m - l == 0 0.012616   0.004317   2.922 0.00963 **
h - l == 0 0.016590   0.004317   3.843 < 0.001 ***
h - m == 0 0.003973   0.004317   0.920 0.62738
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
(Adjusted p values reported -- single-step method)


Thank you very much for your help

Christina



--
Christina Schädel
Institute of Botany
University of Basel
Schönbeinstrasse 6
CH-4056 Basel
ph. +41 61 267 35 06
fax +41 61 267 29 80
E-Mail C.Schaedel@unibas.ch

______________________________________________
R-help@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.
Mark Difford (Ph.D.)
Research Associate
Botany Department
Nelson Mandela Metropolitan University
Port Elizabeth, South Africa
Loading...