BIO Stat WK 6

Occur if the act of applying an experimental treatment has the unintended effect of changing the value of another variable for one or more of the experimental subjects. (Placebo effect)

A manipulative study on human subjects in which neither the subjects nor the persons administering treatments to the subjects know which experimental group any subject belongs to.

Between two factors in an ANOVA design if the additive effect of one factor on the response variable depends on the value of the other factor

Between three factors in an ANOVA design if the pattern of two-way interaction between two of the factors depends on the value of the third factor.

Two factors in an ANOVA design are said to be crossed if the design is factorial with respect to those two factors when the other factors are ignored.

An ANOVA design in which there is more than one factor and the values used for all factors are represented in all possible combinations among the groups. Two factors that are in a factorial design with respect to one another are said to be crossed.

Two factors in an ANOVA are orthogonal if they are statistically independent of each other within the actual data set that is to be analyzed. All factors will be orthogonal if the design is proportional

If the relative proportion of data points obtained for any particular value of one factor is the same for all values of any other factor. A balanced design is a special case. If the design is proportional then all factors are mutually orthogonal.

If the values used were NOT randomly sub-sampled from a naturally occurring distribution of values. This could be because the factor values were chosen deliberately, were sampled in a biased manner, or were sampled exhaustively. Inferences from the ANOVA are only valid for the factor values actually used in the experiment.

A factor in an ANOVA is a random-effects factor if the values used were randomly sub-sampled from a naturally occurring distribution of values. Inferences about  the effects of the factor pertain to the entire distribution of values from which the sample was taken.