One-way ANOVA

Mice experiment: Short version of data set to illustrate ANOVA

See this link for a description of the mice data.

To illustrate the method of one-way ANOVA, we use a subset of the mice.df data called mice.short.df. The data are given below. Put this into a data frame called miceshort.df by cutting and pasting with this R code:


miceshort.df <- read.table(file='clipboard',header=T)

The vectors data, grand, effects, and resid are maximum likelihood estimates of model parameters discussed in class and in the textbook. Look at the results of the following R commands. They illustrate sums of squares calculations we make in the one-way ANOVA discussion.

Here, now, is R code for do an analysis:


# First description of the data via plot and summary statistics:


boxplot(Lifetime ~ Treatment)
nn <- tapply(Lifetime, Treatment, length) 
mm <- tapply(Lifetime, Treatment, mean) 
ss <- tapply(Lifetime, Treatment, sd) 
round(cbind(nn,mm,ss),1)

#Next, calculate an ANOVA table (noting agreement with the 
# partitioning of sums of squares:

attach(mice.short.df)
mice.aov <- aov(Lifetime ~ Treatment))
summary(mice.aov)

Finally we do some Follow-up analysis:
pairwise.t.test(Lifetime, Treatment, p.adj="bonferroni")


Mice data set: short version

Here is the data:


     Treatment Lifetime    grand    Effects Resid
1           NP     35.5 46.00667 -11.126667  0.62
2           NP     35.4 46.00667 -11.126667  0.52
3           NP     34.9 46.00667 -11.126667  0.02
4           NP     34.8 46.00667 -11.126667 -0.08
5           NP     33.8 46.00667 -11.126667 -1.08
6        N/N85     42.3 46.00667  -6.226667  2.52
7        N/N85     40.1 46.00667  -6.226667  0.32
8        N/N85     39.5 46.00667  -6.226667 -0.28
9        N/N85     38.6 46.00667  -6.226667 -1.18
10       N/N85     38.4 46.00667  -6.226667 -1.38
11       N/R50     49.7 46.00667   2.773333  0.92
12       N/R50     49.3 46.00667   2.773333  0.52
13       N/R50     48.6 46.00667   2.773333 -0.18
14       N/R50     48.3 46.00667   2.773333 -0.48
15       N/R50     48.0 46.00667   2.773333 -0.78
16       R/R50     49.1 46.00667   2.433333  0.66
17       R/R50     48.7 46.00667   2.433333  0.26
18       R/R50     48.3 46.00667   2.433333 -0.14
19       R/R50     48.1 46.00667   2.433333 -0.34
20       R/R50     48.0 46.00667   2.433333 -0.44
21 N/R50_lopro     50.7 46.00667   4.433333  0.26
22 N/R50_lopro     50.6 46.00667   4.433333  0.16
23 N/R50_lopro     50.5 46.00667   4.433333  0.06
24 N/R50_lopro     50.3 46.00667   4.433333 -0.14
25 N/R50_lopro     50.1 46.00667   4.433333 -0.34
26       N/R40     54.6 46.00667   7.713333  0.88
27       N/R40     54.0 46.00667   7.713333  0.28
28       N/R40     53.8 46.00667   7.713333  0.08
29       N/R40     53.3 46.00667   7.713333 -0.42
30       N/R40     52.9 46.00667   7.713333 -0.82



sum(Effects*Resid)
[1] 3.596845e-13
sum(Resid^2)
16.416
sum(Effects^2)
1276.683
sum((Lifetime - grand)^2)
1293.099