Unit 4 Comparing more than 2 groups: ANOVA, Posthoc analysis

Question 1

Whats an ANOVA for

Answer 1

ANOVA is a single test to determine the significance of the difference between the means of 3 or more groups/if the variance of the populations are the same.

Question 2

What does ANOVA describe

Answer 2

ANOVA is a general linear model that describes the
relation between a dependent variable and a nominal (categorical) independent variable.

Question 3

What are the checklist of requirements for an ANOVA one way analysis of variance

Answer 3

In any one analysis, there must be only one independent variable (factor), e.g. species
There should be more than two levels for that independent variable or factor, e.g. mussel, oyster.
There must be only one dependent variable, e.g. condition index

Question 4

Assumptions of ANOVA one way analysis of variance

Answer 4

The populations from which the samples were taken are normally distributed (Test for normality: Kolmogorov-Smirnof (conservative) or Shapiro Wilk’s tests)
Homocedasticity (homogeneity of variance): Barlett’s
(parametric) or Levene’s Test (non-parametric)
The observations are all independent of one another (i.e. unrelated, no pseudoreplication)

Question 5

What do you do after an ANOVA

Answer 5

ANOVA informs us there are differences among groups, but not which groups are different if > 2 groups.
Post hoc tests are conducted ‘after the fact’ (the statistically significant result)

Question 6

When do you do a post hoc test

Answer 6

is necessary when you reject the null and there are three or more groups.
We can use both Tukey HSD or Dunnett
(Eg. If we are more interested in whether the supplement has an increase in fish weight compared to the controls we use Dunnett)