This Element presents a computational theory of syntactic variation that brings together (i) models of individual differences across distinct speakers, (ii) models of dialectal differences across distinct populations, and (iii) models of register differences across distinct contexts. This computational theory is based in Construction Grammar (CxG) because its usage-based representations can capture differences in productivity across multiple levels of abstraction. Drawing on corpora representing over 300 local dialects across 14 countries, this Element undertakes three data-driven case-studies to show how variation unfolds across the entire grammar. These case-studies are reproducible given supplementary material that accompanies the Element. Rather than focus on discrete variables in isolation, we view the grammar as a complex system. The essential advantage of this computational approach is scale: we can observe an entire grammar across many thousands of speakers representing dozens of local populations.