The commonest confusion about agnosticism is to picture belief as a single line running from theist through agnostic to atheist, with agnosticism as the moderate middle. That gets the geometry wrong, and almost every argument about the two words traces back to it.
There are two questions, not one, and they are independent. The first is about belief: do you believe that a god exists? Answering yes makes you a theist, and anything short of yes makes you an atheist in the broad sense. The second is about knowledge: do you claim to know? Claiming to know makes you gnostic on the point, declining to claim it makes you agnostic. Because the questions are separate, all four combinations exist and are held by real people.
An agnostic theist believes without claiming certainty, which describes a very large number of religious people and is close to what the apophatic strands of Christianity, Judaism and Islam positively insist on: that God exceeds what human knowledge can capture. A gnostic theist claims certain knowledge that God exists. An agnostic atheist does not claim to know and does not believe, which is probably the most populous position among people who use either label. A gnostic atheist claims to know that no god exists, and is rarer than either side's rhetoric suggests.
Bertrand Russell put the practical upshot better than anyone. Before a philosophical audience he would call himself an agnostic, because he could not disprove the God of the philosophers any more than he could disprove the gods of Homer; before a general audience he would say atheist, because in ordinary speech that conveyed his actual expectations. The two labels were describing different things about the same man.
A consequence follows for anyone reading survey data: questionnaires that force a single choice between agnostic and atheist are collapsing two axes into one, and their figures cannot simply be added together or compared across countries.