ANOVA Flashcards
What is Experimental Design?
Process of designing an experiment to ensure data quality
(right type and enough) and to ensure the validity/precision of analysis
What is an Experimental Unit?
The physical entity/object/thing to which a treatment is
applied or which receives the treatment. The experimental unit may differ from the
observational unit, which is the entity from which a measurement is taken. Example:
Wheat fields, each receiving a different fertilizer
What are the 3 principles to ensure an experiment is valid?
Randomization : Treatments are randomly assigned to experimental units such that each unit is
equally likely to receive any given treatment.
Replication : When each treatment has been applied independently to more than one experimental unit to ensure valid comparisons between treatments - the observed differences should be related to treatment and due to difference between experimental units receiving the same treatment
Reduce noise (blocking) : Arranging experimental units that are similar to each other in groups (blocks) to try and minimize factors that transmit variability to experiment. Try to control conditions of experiment as much as possible. Reduce variation between experimental units to control factors that may cause systematic changes over time. To ensure that any variation in response is due to treatment and not due to any variation in experimental units.
What are factors, factor levels and treatments in an experiment?
Factors : Independent variables being actively manipulated and become the explanatory
variables (e.g. Type of fertilizer
Factor Levels : The different levels of the thing being manipulated. So the different types of fertilizers (Fertilizer A, B and C).
Treatments : The individual factor level or combination of factor levels (when there is more than one factor being manipulated). The factor level being applied to the experimental unit.
What is Completely Randomized Design?
One independent variable (one factor) and treatments are allocated to experimental units completely at random and without any restriction (i.e. blocking).
To analyze the results of a CRD, we can use a one-way ANOVA.
