Chapter 1 – Introduction To Designed Experiments
- p 5 à see graphical representation of factorial experiment
- randomization, replication, blocking
- important distinction between replication and repeated measurements.
o example -> four wafers are processed simultaneously in an oxidation furnace and then a measurement taken on oxide thickness of each wafer
§ this is repeated measurements
§ replication reflects sources of variation both between runs and (potentially) within runs.
- Blocking
o set of relatively homogenous experimental conditions
o each level of nuisance factor is a block
- recognition of and statement of problem
o characterization or factor screening
o optimization
o confirmation
o discovery
o stability / robustness
- design factors
o design factors – factors selected for study
o held constant factors
o allowed to vary factors à variations in experimental material à rely on randomization to balance out effects
- nuisance factors
o controllable à blockable
o uncontrollable à analysis of covariance
o noise
- cause and effect diagrams – useful technique for organizing some of the information generated in pre-experimental planning.
- Industrial era à development of response surface methodology
o immediacy
o sequentiality
- robust parameter design
o Taguchi
o Wu
o Kackar
Chapter 4: Experiments with Blocking Factors
- nuisance factor
o unknown and uncontrolled à randomization
- known but uncontrolled
o analysis of covariance
- known and controllable
o blocking
- paired comparison problem
o improve precision by making comparisons within matched pairs of experimental material
§ example – testing two tips
· test each tip on same material
- problem with completely random
o experimental error will reflect both random error and variability between test beds
o block – test each tip once on each test bed
o randomized complete block
- effects model
o response = overall mean + treatment effect + block effect + error
o if experiment were just completely random, variability for blocking would move into error
o RCBD à noise reducing technique
- model adequacy checking
o normality assumption
o unequal error variance by treatment or block
o block – treatment interaction
- Some other aspects of randomized complete block design
o additivity of randomized block model
o where interactions are of interest à factorial design
o fixed effects / random effects
Latin Square Design
- is used to eliniate two nuisance sources of variability
- blocking in two directions
- two restrictions on randomisation
No comments:
Post a Comment