Okay, now let's take it a bit further, pre-empting the next question, why have two nodes, "transform" and "shape" when only one would make more sense. Well, what is a transform node? Ask the question, what is a group node. Create a group (Ctrl+g) and check its attributes. It looks exactly like a transform node. So a group node is basically a transform node, and so a piece of geometry is basically a shape node just grouped with itself so it has transform information.
Also note that you can delete only the shape node, and you'll have an empty node in your scene that you can move around, rotate scale. So... I guess having both shape and transform nodes is important because transform nodes exist independently (like in the form of group nodes) and shape nodes simply give content for the group/transform nodes.
Just FYI. Hope I'm not being confusing.