Summary Ecological Theory and Data Analysis (ETDA) notes

Ecological theory and data analysis

Some R code
- t.test(extra~group, data = sleep, paired = TRUE)
- cor.test(c(1:10), c(11:20))
Correlation is not slope but if one goes up the other goes up (x and y are correlated)
c = ‘combine’
- Anova(m1,m2): compares model 1 and 2
- Drop1: drop explanatory variables one by one

Distributions
Types of distributions that best represent data of variables




Normal (Gaussian) distribution in R




Histogram in R

,Poisson distribution in R (only positive discrete values, thus no fractions nor negatives)
Lambda parameter (only 1)
 glm with family/argument set to poisson
 means log link is used




Binomial distribution in R (only 2 possibilities)




Beta distribution in R (limited by 0 and 1, no negative values, but useful for continuous data (fractions))
Look at Wikipedia for the different shape parameters (shape1=1, shape2=4)

, Gamma distribution (continuous, positive data)
Can for example go down very low to almost zero; Poisson often go to that extreme




Negative binomial distribution (large, positive integers)
When the variance is much larger than the mean (variance>mean) (Poisson distribution cannot do this)




Multinomial distribution in R (limited outcomes, but more than 2)