Factorial Designs

Two+ independent variables and interactions

Factorial designs are experimental research designs that simultaneously manipulate two or more independent variables (factors). This approach estimates each factor's main effect on a dependent variable while also revealing interactions between factors — situations where the effect of one factor depends on the level of another. Because they are both efficient and informationally rich, factorial designs are among the most widely used frameworks in experimental research.

Core Concept

In a traditional one-factor experiment, a researcher varies only one independent variable at a time. Factorial designs remove this constraint by addressing two or more factors simultaneously. For example, a 2×2 design has two factors each at two levels, producing four experimental conditions. The average effect of each factor on the dependent variable is called a main effect, while the combined, joint effect of two factors is called an interaction. Statistically, an interaction means that the effect of one factor differs depending on the level of the other factor — a relationship invisible in single-factor studies.

How It Works and Main Types

Factorial design notation expresses the number of levels for each factor using multiplication: a 2×3 design has two factors — the first at two levels, the second at three — yielding six conditions. If participants are assigned to all conditions, the design is a full factorial. Participants may be assigned to only one condition (between-subjects), to every condition (within-subjects/repeated measures), or to some of each (mixed factorial). When many factors are included, the number of conditions grows rapidly (a 2×2×2 yields eight conditions), making fractional factorial designs — where only a subset of conditions is run — practical for large experiments.

A Concrete Example

An educational researcher wants to examine two factors affecting students' test performance: teaching method (traditional lecture vs. problem-based learning) and class size (small vs. large). This 2×2 design produces four conditions. The researcher can test both the main effect of each factor (does teaching method matter overall? does class size matter overall?) and the interaction. If the benefit of problem-based learning emerges only in small classes, that is a meaningful interaction — a finding that two separate single-factor studies could never reveal, illustrating precisely why factorial designs are so powerful.

Common Pitfalls and Best Practice

The most common mistake is focusing on main effects while ignoring a significant interaction; a meaningful interaction changes or can nullify the interpretation of main effects. Another frequent problem is underestimating the sample size required — as the number of conditions grows, adequate statistical power demands more observations per cell. Researchers should also resist inflating the number of factors without theoretical justification and should prioritize testing interactions that are substantively meaningful. When reporting results, both main effects and interactions should be presented alongside appropriate effect size estimates to convey practical significance.

Key terms

Main Effect: The average effect of a single factor on the dependent variable, ignoring all other factors.
Interaction: The condition where the effect of one factor differs depending on the level of another factor.
Full Factorial Design: A design that includes every possible combination of all factor levels.
Mixed Factorial: A factorial design containing both between-subjects and within-subjects factors simultaneously.
Cell: A single experimental condition representing one specific combination of factor levels in a factorial design.