Strip-Plot Design

Strip-Plot Design is a restricted randomization experimental design where two factors are applied in perpendicular strips — one factor is applied in horizontal strips and another in vertical strips, creating a grid where each cell receives a unique combination of the two strip factors.

How Strip-Plot Design Works

• Row Strips: Factor A is applied to entire horizontal strips (e.g., temperature across a batch of wafers).
• Column Strips: Factor B is applied to entire vertical strips (e.g., etch time for a group of wafers).
• Intersections: Each row-column intersection gets a unique (A, B) combination.
• Error Structure: Three error terms — row strip, column strip, and intersection — reflecting the randomization restrictions.

Why It Matters

• Practical Constraints: Reflects real fab operations where some factors cannot be independently randomized for each run.
• Efficiency: When hardness of factor levels varies, strip-plot designs are more practical than fully randomized designs.
• Semiconductor: Common when batch factors (furnace temperature) are crossed with per-wafer factors.

Strip-Plot Design is experimenting with perpendicular constraints — a practical design for when two factors must each be applied to groups of experimental units.