Interaction effects in statistical models constitute a core mechanism within multivariate analysis where the influence of one independent variable on a dependent variable is contingent upon the level of another factor. This concept formalizes the deviation from additivity, defined mathematically as non-zero coefficients for product terms (e.g., $\beta_{xy}X_i X_j$) in linear or generalized linear model equations. As a subfield of regression analysis and experimental design theory, it addresses the structural complexity of systems where joint variable configurations produce outcomes distinct from their isolated marginal effects.
Prerequisite chain context: requires Two-Level Factor Structures for Variables.