2^k factorial DOE analysis estimates main effects and interactions from a two-level designed experiment, ranks them on a Pareto chart of standardized effects, and reports main-effect means.
In a 2^k factorial experiment every factor is run at two levels and all combinations are covered, so each effect is estimated from all runs simultaneously — vastly more efficient than one-factor-at-a-time, and the only way to see interactions. Factors are coded −1/+1; the effect of a term is 2× its regression coefficient: the average change in response from the low to the high level.
With replication, effects get t-tests from the residual variance. In an unreplicated saturated design there are no residual degrees of freedom, so Lenth's pseudo standard error (a robust estimate from the smaller effects themselves) provides the yardstick — the standard approach for screening designs.
The Pareto chart of |standardized effects| with a significance cutoff is the classic readout: active factors stick out past the line.
A numeric response + ≥ 2 factor columns each with exactly 2 levels + the interaction order to fit.
Effect / SE / t / p per term, R², a Pareto chart of standardized effects with the α = 0.05 cutoff, and low/high means per factor.
A large interaction means those factors must be set jointly — read the interaction before interpreting its main effects. Level ordering is inferred (numeric content, low/high vocabulary, else alphabetical), so check the low → high direction in the means table.