I know there are several questions (here and on other websites) regarding the comparison of values, but I am still lost with the data I have and the analyses I should conduct.
I have just 18 values (home range areas) and I want to compare them twice: first between study sites and then by gender (male/female), to see if there are significant differences in home range areas within these groups. I would also like to consider the interaction between Study.site*Gender. Here is the data:
ID Area Study.site Gender
v1 6.558201 1 0
v2 3.274141 2 0
v3 4.841487 2 0
v4 6.419502 2 0
v5 2.351595 3 0
v6 1.104536 3 0
v7 3.687818 3 0
v8 3.708099 3 0
v9 4.198809 3 0
v10 3.831449 4 0
v11 5.547973 5 0
v12 7.731753 1 1
v13 1.772005 2 1
v14 2.433105 2 1
v15 2.142429 3 1
v16 1.918333 4 1
v17 1.951922 4 1
v18 3.740321 5 1
I fitted a linear model, after a "sqrt" transformation of the "area" values (since they were not normally distributed):
test.1 <- lm(Area ~ Study.site*Type, data = data1)
Then I believe an ANOVA would be good to test these differences, but I am having trouble with the assumptions; I tested first for the residuals and it looks that they are not normally distributed.
> test1.res <- resid(test.1)
> shapiro.test(test1.res)
Shapiro-Wilk normality test
data: test1.res
W = 0.88702, p-value = 0.03432
My main questions are: 1) is the ANOVA actually a good way to obtain the results I want, considering the (few) data? If so, 2) how should I solve the normality issue?
