Compound Factors
When the effect of one factor is different across levels of another factor.
What is an interaction effect?
In addition to randomization and factorial crossing, this design principal is present in the split plot/repeated measures design.
What is blocking?
This is a table with a design's universal and structural factors listed in the rows and the SS, df, MS, and F-ratio quantities in the columns.
What is an ANOVA table?
The probability that we'd calculate an F-ratio as big or bigger than we did if our treatment effect variability were actually zero.
What is a p-value?
This Fisher assumption holds if the average of my residuals is roughly zero.
What is the Z (zero) assumption?
A BF[3] design can test up to this number of two-way interactions.
What is three?
This new design principal, present only in SP/RM designs, necessitates the use of the MS blocks for the denominator of the between-block factor's F-ratio.
What is nesting?
If I assume an additive model for a design that has compound within-blocks factors, I use this as my denominator for my F-ratio's.
What is the MS residual (factorA X factorB X blocks)?
This design allows you to count for two sources of nuisance variance.
What is the Latin Square design?
We consider this assumption violated if the SD of the biggest group is more than three times as big as the SD of the smallest group.
What is the S (same sd) assumption.
This is the number of research questions that are tested in a SP/RM[1,1] design.
What is five?
One way of blocking is by creating similar groups of experimental materials, this is another way.
What is by reusing or sub-dividing experimental material?
In a split plot/repeated measures design, I use this as the denominator for any F-ratio's testing fully between-block factors (main or compound).
What is the MS blocks?
To help me find the equations for my treatment effects I can label these on my factor diagram.
What are the inside and outside factors?
We're looking for the points to fall along this line when making a scatterplot of observations for two levels of a within-block factor.
What is the y = x line?
When the pattern of a two-way interaction is different across levels of a third factor.
What is a three-way interaction?
This is a design with blocking present, but only a within blocks treatment factor and no between blocks treatment factor.
What is complete block, or CB[1], design?
This is the sum of the effects of all outside factors.
What is the partial fit?
In the Latin Square design, the number of rows is equal to the number of columns which is equal to this.
What is the number of treatment levels?
This Fisher assumption holds that observations within the same treatment condition are measurements of the same true value.
What is (C) constant treatment effects?
If I suspect that a within-block treatment effect differs across blocks, I should use this model for my residuals.
What is the non-additive model?
Why might it be helpful to block observations according to a variable correlated with the response variable?
What is to remove nuisance variance?
This table, which is especially helpful for designs with compound factors, organizes the design factors with columns labeled unit, grouping, main, 2-way, and 3-way.
What is a summary table?
If I assume a non-additive model for my residuals in a design with compound within-block factors, I use this as my denominator for the F-ratio for Factor A.
What is MS Factor A x Blocks?
This data visualization will help me check for independence in my residuals if I have a within block treatment factor that has two levels.
What is a scatterplot?