Interactive concepts produce nonlinear outcomes that regression models systematically misrepresent.

Evidence

The model is very nonlinear since the outcome is not the sum of the individual effects, as is the case in statistical models like regression and event history. This is due to its highly interactive character. — Gary Goertz, Social Science Concepts and Measurement, 1568

Categorization

• XREF: Connects to the reader’s familiarity with causal-inference debates about interaction effects and the limits of additive regression.
• SEED: Angle: a short essay on why additive regression flattens the interactive, necessary/sufficient structures that qualitative researchers actually theorize.