The NDB is a simple animal - it sits in one place and transmits. Therefore, most errors you notice are more likely going to be receiver errors. I know that you can get coastal and/or terrain refraction, night error, dip and the ADF's habit of pointing at thunderstorms - but in theory, the receiver makes all the difference.
Of course, your receiver is going to be better at pointing at a stronger signal. I don't have the published protected ranges of the three beacons but I do remember from flying at Brize for may years that the BZ is quite powerful. And considering that the owners of the BZ have considerably more resources that the owners of OX and GST, there may be something in this too.
But with all these errors, we have to ask why in this day and age of GPS updated RNAV we are using a navaid designed in the 1920's. The procedures still exist and sometimes we have to fly them, so IO540s comment reagrding the NDB procedure being used with a GPS overlay is very valid. Last week, I flew into an airfield in China which like all airfields in this region is undergoing major WIP due to the fact that China's aviation market seems to double in size every two weeks. The airport managed to have both ILSs out of service at the same - with the VOR approach being reported as unreliable! An NDB was the only option. But as the airfield is in a steep valley and it was night and there were CBs around, there was no way we would have continued if the GPS hadn't been reporting 'accuracy high'.