We perform an experiment to answer a research question regarding some response variable of interest in relation to a treatment defined by factors and/or covariates.
The main objective is:
- To establish differences in response between treatments (using t-test / ANOVA)
- To establish relationship between response y and quantitative explanatory variable x (using regression)
- We aim to establish a cause-effect relationship between the dependent response variable and the explanatory factors and/or covariables.
We (aim to) design an experiment that is appropriate given the research question and the practical possibilities. Designs that will be described in this and further modules are:
- Completely randomized designs (CRD).
- Randomized complete block designs (RCBD),
- Split plot designs,
- (Balanced) incomplete block design (BIB, lattice, a design)
In all designs, a basic principle is that treatments are randomly allocated to the experimental units, and the randomization should always respect the structural features of the design (we will discuss this in more detail later).