Lz? I don't see Lz on the datasheet...
They give real and imaginary components, which is nice when you want to know these things. For example, it makes a terrible inductor, because although the impedance is useful at most frequencies, the Q factor is not (X is reactance). So it's only a passable inductor below 8MHz or so, and not a good one at that. Which is good, because you generally want to use these sorts of parts to dampen resonances, not make them worse.
You don't always see differential mode characteristics; in power line chokes, it usually amounts to a few uH, with dips and swells similar to the common mode case (resonances with the winding geometry and such), but at lower impedances / attenuation.
On this part, the (C, N, O) plot shows impedance for three modes: open means one winding only (the other open), which you should expect has about the same result as common mode does, but with differences mostly at higher frequencies (perhaps higher because the core looks relatively larger, or lower because the core is less efficiently utilized). Common mode is both windings in parallel, driven as a single component. Normal is either with one shorted across, or with the two wired series-opposing (shorted at the far end), which gives low impedance at low frequencies (transformer action), peaking to a high impedance (apparently much higher in this case) where the windings act in parallel resonance (between the wire-core-wire capacitance and the wire-to-wire leakage).
Chokes for data lines, boasting 100s of MHz to GHz / Gbits of bandwidth, are bifilar wound and, although they'll often provide differential mode impedance curves, you shouldn't read into these too much, because they exhibit very large phase shifts at high frequencies -- you see not just one peak, but several peaks and valleys, corresponding to transmission line stub modes. This is actually a good sign, because it means your data will pass relatively unscathed, even well into that frequency range.
Tim