What springs to mind is this:
If you know
the impedances that the O/P ports want to 'see'
- i.e. what will match them, then a suitable resistive
pad between the O/P under test and the 50-ohm input
of the VNA will keep everything matched, thus enabling
you to make the measurements. The unused port will
need to be terminated in its appropriate R, for
which you may - or may not - need a pad, depending
on just what the O/P Z of that port actually is.
Of course, you then need to correct the measurements
(return loss, insertion loss, etc.) to take into
account the attenuation of the pad(s). As for the
pads themselves, you might have to make these yourself,
and if the freqs. invloved are many tens of MHz,
that could be difficult without introducing stray
L and C, thus putting errors into your measurement.
If, however, you have a 75-ohm to 50-ohm requirement,
suitable coaxial pads are available commercially.
Quantum Mechanics - the
dreams that Stuff is made of.