PSO-X incorporates dozens of algorithm components that have been proposed to solve single-objective continuous optimization problems using particle swarm optimization (PSO). While component-based frameworks allow for flexible algorithm configuration and enable designers to automatically generate implementations tailored to specific optimization problems, understanding which components matter most and how they interact remains an open question. In this paper, we used performance data from 1,424 PSO algorithms instantiated from the PSO-X framework and apply functional ANOVA to quantify the impact that algorithm components, and their combinations, have on algorithm performance across different problem landscapes.