You could quite easily test that, by comparing bit-flipped domains to keyboard-neighbour domains.

Or alternatively, just by looking at the distribution of hits to bit-flipped domains (which should be broadly uniform; typos will be non-uniform).