Experimental Design Flashcards
What is a factor?
Independent variable of interest.
What is a response?
Dependent variable.
What is meant by a factorial DoE?
In each trial, all possible random combinations of all level factors are considered.
What is meant by the level of a factor?
The conditions used for a factor in an experiment. This is usually high or low.
What is meant by treatments?
A specific level of a factor of interest. eg. when A is high and B is low the treatment is a..
How do you calculate the main effect of A for a 2^2 design?
= A high level - A low level
= (a+ab)/2n - (b+(1))/2n
How do you calculate the main effect of B for a 2^2 design?
= B high level - B low level
= (b+ab)/2n - (a+(1))/2n
How do you calculate the AB interaction for a 2^2 design?
= B(high+low) - A(high+low)
= (ab+(1))/2n - (a+b)/2n
For a 2^k design what is the number of independent variables and what is the number of combinations?
Indep. variables = k
Combinations = 2^k
What does the sparcity of effects principle state?
A single replicate should be used at each point with no repeat and 3 variable interactions and above are negligible.
What is the difference between a replicate and a repeat?
Repeats take multiple measurements at a given time and condition. Whereas, replicates take multiple measurements at a given condition at different times.
What is the purpose of centre points?
- Gets a pure error estimate
- Assumes no interactions
- Tests for curvature/npn-linearity
For a fractional design how can a 2^3 experiment be split up & what does it form?
Let I = ABC
Therefore, a, b, c & abc will have +ABC and ab, ac, bc & (1) will have -ABC.
Here, a = bc, b = ac & c = ab which is known as an alias.
Prove the alias for A for a 2^3 fractional design.
Where I = ABC
A = A.I = A. +ABC = A^2 BC = BC
