The SD is only very slightly reduced, this may be experimental error.
I feel like I am back in school in a lab course. Here is a better experiment: Out of focus beige wall taken at ISO 800 and re-developed at 80,100,200,400,800,1600. The results show that there is no noise reduction at ISO 100. There is mostly luminance noise reduction at ISO 200. At 400-1600 there is increasing chroma noise reduction.
First line for each image is results on the RGB image, next is Y, next is CbCr (color).
Sorry that the columns don't line up. StdDev is the important column.
Area Mean StdDev Min Max
ISO80 890028 138.940 11.135 83 202
Y---- 890028 141.582 10.983 83 202
CbCr 890028 126.832 2.196 115 136
ISO100 890028 138.794 11.192 82 202
Y---- 890028 141.833 11.038 85 202
CbCr 890028 126.435 2.198 115 136
ISO200 890028 137.664 8.133 90 191
Y---- 890028 141.076 7.875 94 194
CbCr 890028 126.067 2.137 114 135
ISO400 890028 138.593 5.764 98 180
Y---- 890028 142.541 5.702 102 184
CbCr 890028 125.590 0.852 121 129
ISO800 890028 138.177 4.278 110 166
Y---- 890028 141.847 4.255 115 170
CbCr 890028 125.873 0.595 123 128
ISO1600 890028 195.545 3.557 171 214
Y---- 890028 199.407 3.463 175 218
CbCr 890028 125.641 0.738 123 128