I have a collection of fairly accurate models - thinking about Hot's comments about center of mass, I can see his point for all but low speed switching locomotives. When getting to the center of mass becomes a long stretch, as with 4-10-2s and 4-12-2s, the crosshead is guided further behind the cylinder to reduce main rod length.
A quick glance at my display shows the main driver almost always somewhere close to the middle of the boiler. Only not true for 0-4-0s and Harriman 0-6-0s.
it is harder to see on articulateds - in the case of the Challenger, it is probably driven more by main rod max angle, and interchange of parts between engines.