Having equal or proportional cell sizes. Cell sizes are proportional if for every pair of factors, each cell size equals the product of the row and column marginal totals divided by the grand total. See Winer pg. 420.
Between-Subjects The test on a grouping factor is a test on the between-subjects sums of squares.
Cell Cells are formed by the crossing of 2 or more factors. Each cell is defined by a particular combination of levels; 1 level from each of the factors being crossed.
Covariate A variate associated with another variate, usually of secondary interest, is termed a covariate. This term carries with it the notion of statistical control. For example, we might obtain some measurement (variate) following an experiment. In order to control for random variation in the population, we obtain the same measurement (covariate) prior to administering the experiment.
Crossed Two factors are crossed when all possible pairwise combinations of factor levels have been obtained.
Dependent Related to.
Dependent Variable The measured response. A variable which is explained by the independent variables.
A dummy or artificial variable represents the presence or absence of a particular level of a nominal/ordinal variable. A set of dummy variables (one less than the number of levels) may be used in place of the nominal/ordinal variable when carrying out a regression analysis.
Effect The result of a factor (upon a dependent variable).
Factor A variable consisting of one or more levels (treatments) thought to be a cause of variation in a dependent variable.
Factorial Design An experiment which tests the effect of one or more factors.
Grouping Factor (Between-Subjects Factor) A factor whose levels separate the sampling units into groups whereby no subject is exposed to more than one level of the factor.
Independent Not related to.
Independent Variable The explanatory or predictor variables in the model, either factors or covariates.
Interaction effect The effect (upon a dependent variable) of crossing 2 or more factors which has not been predicted by the sum of the main effects. This interaction is a measure of the effect upon the dependent variable by one factor as a second factor is varied from level to level.
Latin Square Design An experiment which uses a Latin square to allocate treatments. The Latin square is defined by crossing 2 factors each with k levels. Furthermore, the k levels of a third factor are allocated such that each level occurs once in each row and column.
Level A factor treatment. For example, a drug dosage (the factor) may be administered at three different levels.
Main effect The effect of a single factor (upon a dependent variable).
Nested If all levels of one factor do not appear with all levels of another factor then the first factor is said to be nested.
Nested Design A variant on the Factorial Design in which at least one factor is nested.
Orthogonality The characteristic of independence among factors.
Repeated Measure Any measurement which is taken more than once on an individual sampling unit. A person's reaction time to each of three different doses of inoculant and the result of a test administered to a subject at four time periods are examples of repeated measures.
Repeated Measures Design A variant on the Factorial Design in which at least one factor is a repeated measure.
Replication An increase in the number of sampling units which decreases the error variance (mean square error).
Within-Subjects The test on a within factor is a test on the within-subjects sums of squares.
Within Factor (Within-Subjects Factor) A factor whose levels are crossed with the "sampling-unit factor" (typically the subject). That is, a subject is exposed to more than one level of the factor.