Today, April 23rd 2017. I carried out a similar experiment to yesterday but with WSPRlite G33CWI reduced in power by 3dB compared the WSPRlite G31CWI. 57 simultaneous S/N readings from stations all over 500km away were gathered.

The results distribution is shown below;

X=Number of samples, Y=difference in dB

The mean was 2.5dB, The mode was 3db and the standard deviation was 1.6dB.

As the distribution is fairly Normal, it can be said that the result lies between -0.6dB and +5.7dB with 95% confidence.

For comparison DX10 also clearly shows the difference over the same period: