Why are some pollsters producing results among "likely voters" that vary, either over time or compared to other pollsters? A week ago, I offered a primer on how pollsters define and choose likely voters. Today, I want to consider whether the variation we are seeing is unusual, and if so, what may explain it.