More often then not, captured panels crack because of their inability to move rather than due to the actual movement itself.

If a species panel is glued on all four edges to the door stiles and rails, when the large panel tries to contract as its environmental conditions change, the edges are held tight so the panel may crack along the grain.

Do you know if the panels in your doors were glued or floating?

Even if floating, poly or paint can seal a panel to the rail and restrict movement. So even with the best of intentions, Mother Nature can ruin our plans.

For your door, What I'd do it finish the panels before the door is assembled, sealing the panels on all sides. Then assemble it with the panels floating.

The other reason I prefer to finish the panels before assembly is so when the panel shrinks I can't see unfinished slivers of wood where the panel slots into the stiles and rails.

With the large-sized panels required in a 2-panel shaker door, have you considered using species plywood for the flat panels? Plywood, being more stable than species lumber, will not move as much.

A properly sized bathroom fan will prevent excessive humidity in the bath as well, reducing the uneven swings in humidity the door might face.

Mongo

The door is already assembled and sitting in my basement, so unfortunately, sealing the panels before assembly is not an option. Our old doors had floating solid pine panels, and I restricted the floating when I put the poly on, so that was likely the problem.

Any suggestions for applying the poly on the new door to avoid restricting panel movement, since the door is already assembled?

In our new door, the panels are plywood with an oak veneer, so it sounds like that might help. Also, we've put in a Panasonic fan with a timer switch (and there's a window in the room), so hopefully humidity will be less of a problem.