Ferrites on a cable is essentially a sign of a failure to do proper EMI design in the actual PCB. The fact that you see them so often should tell you something..
Without going to gruesome details of EMI design issues, it can be said that a ferrite on a cable is a) ineffective and b) expensive band-aid to a design which didn't pass the CISPR test requirements for consumer electronics. Ferrites do not work very well on cables because the cable already typically has high impedance and you're essentially doing energy division..
Capacitors are perfectly good solution for EMI, when they're on PCB itself. Typically you would use three terminal flow-through caps to eliminate differential noise. Capacitors may also eliminate common mode noise when connected from both signals to reference plane (usually GND) but this tends to kill your differential signal as well. Or single-ended signal. In short, capacitors are bad for your signal.
Trumping all that is a common mode choke. These are available on bifilar configurations that work even on high-speed data lines such as gigabit ethernet. Even better is having proper shielding and grounding strategy which will short the common mode noise back to the source instead of passing it to the cable.
Incidentally ferrites can work very well if you put them INSIDE the case and use a cap to form a RC (well, LC) circuit.
Clip-on Ferrites have their place but they should be a last resort/stopgap solution or used for difficult cases such as high-speed V-by-one or LVDS signal cables.
Here's a primer from Murata on the subject:
edit Corrected some nonsense about capacitors across differential pairs.