I think it's closer to -22 or -23. your 3rd collum is just
Pout(x) - Pout(x-1)
So this number looses its linear approximation. I use to measure -1dB Comp
with a 10dB Step attenuator. Is what you do is put the Step Attn. in front of
the input signal. Now set the Pin to (as Mazz said -20dBm) and read your
Output Pwr to see what is your Gain:
@ Pin = -20dBm ---> Gain = Pout - Pin= -9.780 - (-20) = 10.22dB
Now step up the Atten. by 10dB. The new Pin will be 10dB lower (-30dBm).
@ Pin =-30 --> Gain = Pout - Pin = -16.45 - (-30) = 13.55
If Pin = -20 was the -1dB CP then the Gain at -30 dBm ( -20 +10dB Atten)
should be 10.22 + 1dB = 11.22 ( 1dB higher than -1dBCP), But since the Gain at
Pin =-30 is measured to be 13.55, then we need to find were (Pout - Pin) =12.55
Assuming that 13.55dB is the nominal gain, it looks like the 1db Comp point
is between -23 & -24 .
Cheers