Reconciliation of graph data in the Readme with raw data in CSV

[pdehaye]

I am interested in understanding the intra-model variation between device at broadcasting.

The CSV contains averaged data only (this is really per broadcasting phone model, correct?).

My understanding was that the plot titled Pictured above: Raw RSSI readings of devices at 2m taken in an anechoic chamber in the README could help me infer that.

However the data clearly doesn't match. For instance the plot shows for the Samsung A10S an average of around -64dB while the table gives -66.57dB. I picked this model because the values seem so consistent, so the difference is small, but the others are even less coherent (see for instance Oppo AX5S).

What am I misunderstanding?