@ADTech. All I did was attach my antenna to a KWORLD hybrid USB tuner on my laptop, tune in to the audio (and video) test signal being broadcast by XHTJB-2, then let PolarPlot collect data as the antenna rotated 360 degrees.
I have no insight into how PolarPlot calculates dBi from the test signal, but just to be sure I was not picking up ambient noise, I did the same thing using a 665Mhz dipole fabricated from RG-59 COAX. This was the result:
Your point about multi-path potentially impacting the PolarPlot results is well made. In fact, I DEPEND on multi-path to deliver OTA TV service in apartment complexes. LOS is simply impossible in high-density locations, so tuning into the strongest signal (wherever it is coming from) is best served with an omnidirectional antenna.
Just so you are aware, for marketing purposes, I only claim 4.3dBd VHF and 9.17dBd UHF, not the 20+dBi shown in PolarPlot. I have EZNEC Demo version 5.0, but my attempts to model my antenna with it produces errors stating that I have incorrectly connected some of the wires. Perhaps it is those "incorrect connections" that give my antenna its performance?
For me (and my clients), on-the-ground real-world results are a lot more important than clinical test numbers. I guess I will just continue tilting against the CEA windmills.
Cheers, and Happy 2015!