Check Experiment Variance
This script performs an analysis of variance in a observation among the experiments. All the replicas are considered. The result will be a contrast between the experiments on the response variable. The output will be two files: one containing the plain text analysis output, and another with the graphical representation of the Tukey's test. The argument of this scripts is only one : Input data file. This file must contain 4 columns: Experiment Replica Module Obs. The ANOVA is performed between the first column (Experiment) and the fourth column (Obs). All the Observations from the same experiment will be considered.
The Output are the following:
Df Sum Sq Mean Sq F value Pr(>F)
data.experiment 2 21.4 10.7 1.4062 0.2453
Residuals 2187 16634.1 7.6
Tukey multiple comparisons of means
95% family-wise confidence level
Fit: aov(formula = data.obs ~ data.experiment)
$data.experiment
diff lwr upr p adj
Model2-Model1 -0.1121622 -0.4484200 0.2240956 0.7139638
Model3-Model1 0.1306814 -0.2091099 0.4704726 0.6390999
Model3-Model2 0.2428435 -0.0969477 0.5826348 0.2145961
Analysis: From the graphical Tukey's test we can deduce that there is no significant difference between the three compared models (due to the zero is included on all the 3 intervals). When the case of the zero is not included on an interval, we think that there are a difference between the contrasted models. Also, the ANOVA table is giving a
p value of 0.2453. This means that the null hypothesis of the ANOVA (that the means are equal) can not be rejected. By consequence, the means may be equal. The Tukey's test detailed contrasts shows also the probability of reject the null hypothesis on each contrast. For more information how to interpret an ANOVA, please check the Douglas C. Montgomery book's
Design and Analysis of Experiments.
Download the Script