Youre asking for accuracy of the claims made? Well, the numbers dont lie, and the results are in the noise as far as sampling goes. Even for the same sample, 6% standard deviation should be considered normal.
Si isnt the only indicator of dirt. Some places might have other components in the dirt at higher levels.
The K&N is more "open", which then requires the use of an oil to help grab and hold particles. Even if it does hold these, the wild cards are if it holds as much, what happens as the oil dries, and if the oil migrates. Thats what worried me the most - oil migration on the MAF sensor, and knowing when to clean and how much to oil it.
Additionally, adding the same size panel doesn't fix any restriction issues that might be intrinsic. Intakes seem to be better configured these days, but if there is a restriction to be overcome, the panel is only a minor piece in the puzzle.