Very skeptical of the way the data is used. There are too many broad or vague caveats to really make much sense of what is being looked at.
For example this is based on info from apps running the AdDuplex SDK whatever that's worth. It may in itself bias the result because only a certain category of app uses the SDK (coupon apps for example).
This is only among Windows devices not all OSes, which would significantly deflate the real world representation of the numbers. Windows 8/RT only has a 2.9% share of all tablet OSes in 2012 and then Surface would only be 7.6% of that small slice. IDC: Windows Tablet Share To Hit 10.3% In 2016
There is no distinction between Windows 8 and Windows RT tablets.
There is no definition of what is considered a tablet. What are the HP tablets that are doing so well?
[Edit: As some users are suggesting, these numbers are skewed to RT devices since it measures RT app usage via the Store. The opposing theory says that those on Windows 8 proper simply don't use RT apps as much as those with a Surface.
That is certainly plausible although without any firm numbers of usage patterns of non-RT users, it's hard to say for sure. Due to the "in your face" approach of the RT interface and Store we're not entirely convinced of the distinction just yet.]