I can't say a smaller speaker is going to image better. The only distortion a smaller speaker intrinsically cures is less diffraction. Cabinet resonances also mess up the sound (the 801's and 802's have ceramic heads with non-parallel sidewalls whereas the cheaper models are still wood "boxes.") B&W was using concrete for some of the heads in the original 801 Matrix series. Not to mention when the designer knows they have the luxury of the LF driver they can set the crossover point a little higher than a usual sub setup and take some work of the midrange driver, reducing intermodulation distortion in the midrange and opening up new design options-like the "surroundless" midrange (which obviously have very low excursions and are physically limited from going very deep at all--hence I've never seen one in a 2 way). The bigger better models are typically using better drivers in the mid to upper frequencies which helps soundstaging, imaging, and everything else. Personnally I don't think B&W are that great--but its an example to kick around a few generalizations showing what one manufacturer did. (There's always the issue of how well they did it, the specifics may be the next model up does kind of suck, but....)
Imaging also becomes a function of the speakers off-axis response and room interactions. These two imaging properties tend to exist separate of baffle size and even driver quality. 3/4" domes have better off-axis response than 1" domes in the highs, cones have better off-axis response than domes, for better or worse [Colloms just says "there's some debate on issue" and drops it there in his book. However, I did see an AES paper by another individual that specifically addressed broad off-axis responses v. beaming, using the word loosely]. Not to mention a big baffle can be covered in foam to deal with the diffraction (Dunlavy). I'm not sure on the room standing wave issues. A sub can be placed optimally for the standing waves. But, I've also heard two subs (or two full-range speakers) will tend to smooth out the room nodes as opposed to a single lf unit. I don't think there are any generalizations that can be made on the issue, it comes down to specific designs. I won't deny some sub/sat systems can do a better job than some full-rangers--usually if you're willing to jump product lines and mix-n-match. But if you're talking within a manufacturer's model line the bigger unit *should* be better across the board. I won't deny adding a dedicated sub can probably get you deeper bass though than many "full-rangers".
Imaging also becomes a function of the speakers off-axis response and room interactions. These two imaging properties tend to exist separate of baffle size and even driver quality. 3/4" domes have better off-axis response than 1" domes in the highs, cones have better off-axis response than domes, for better or worse [Colloms just says "there's some debate on issue" and drops it there in his book. However, I did see an AES paper by another individual that specifically addressed broad off-axis responses v. beaming, using the word loosely]. Not to mention a big baffle can be covered in foam to deal with the diffraction (Dunlavy). I'm not sure on the room standing wave issues. A sub can be placed optimally for the standing waves. But, I've also heard two subs (or two full-range speakers) will tend to smooth out the room nodes as opposed to a single lf unit. I don't think there are any generalizations that can be made on the issue, it comes down to specific designs. I won't deny some sub/sat systems can do a better job than some full-rangers--usually if you're willing to jump product lines and mix-n-match. But if you're talking within a manufacturer's model line the bigger unit *should* be better across the board. I won't deny adding a dedicated sub can probably get you deeper bass though than many "full-rangers".