TDR – Why the difference between short and open with reference to the load?

When setting up for a TDR measurement, I've noticed a significant step difference between the open and shorted measurements at the probe tip compared to a calibrated 50-ohm load measurement. Why is that?
The setup:
1) OPEN - Probe only, the probe is not contacting any surface.
2) SHORT - Probe only, the probe is shorted on the nearby ground plane.
3) LOAD - Probe only, the probe is on a calibrated 50-ohm structure.
4) OPEN - Probe and structure under test.
5) SHORTED - Probe and structure under test.

Trace colors:
Red - Probe tip open.
Green - Probe tip on 50-ohm structure.
White - Probe tip shorted to ground plane.
Blue & Violet - Structure under test.

What do you mean by Step difference? They each have about a .18 amplitude difference for the High impedance, and Low impedance.

Your right, actually .19mV. But, with the probe and cable 50ohms shouldn't the match be to the end of the probe?

Because the TDR uses a Pulse to send down the cable to trace for the open or short, there is a certain amount of standing wave or loading that is affecting the initial internal cable impedance. The internal coaxial run to the connector is set to 50ohm so that it would appear infinite witha calibrated load, with the exception of the connections showing the small bumps. When you short out the probe tip, it is reducing the overall impedance in the same fashion a low value parallel resistor does to another resistor. With the standing wave generated from the pulse it reads higher in impedance, yet physically isn't. This is an effect caused by the circuit that monitors the return loss of the signal.

The TDR I used in the Military had the ability to compensate for PTFE, and other dialectric values. This setup different correction factors on the measurement circuit that allowed for measuring RG75 and RG58 properly. This also showed when doing the calibrations, in the PTFE setup was slightly larger than Teflon setting.