Range Error in NLOS in the wrong direction

I have strange behaviour with the DWM1000 which I’m using together with a Raspberry Pi 4. When I measure a signal in LOS the range is indicated further away than when I put something at the same range into the direct flight path. E.g. I use a distance of 1 m which I also get as measurement in LOS but in NLOS it is then consistently only 0.95 m which makes no sense because the signal should need longer to travel through an object and therefore the measured range should also be longer (e.g. water bottle, so 20 cm of water to propagate through).

Has someone experienced something similar?
Does the chip try to counteract longer flight times of the signal when it detects NLOS in a way?
And if yes, can I somehow turn that off?

Thanks for the help and regards