Sorry until I see the data that "test" (using test in loosest sense of the word) is based on I say it's hogwash. The test was likely more research of covenience which often comes from that corner of the net. They also just tested if things ranked better with or without <H1>. It should include semantic relationship between Document structure and title and other HTML elements. IMO, that test is hardly a reason to change something I've believed is important since my first attempts at determing factors back in 96/7. You can't prove if it helps with a test because there are so many factors everything is pretty hard to prove definitively especially if it is interplay between html elements.

I won't say it is so important it must be there just... there are a finite number of elements ( my last count was 28 ) that can legitimately be "optimized" so any element that's a possibility should be optimized. Isn't that what we are supposed to do? Why figure out exactly what to optimize... do em all and don't sweat whether you've missed an opportunity. IMO, it can't hurt so.... who the F cares if it's a magic bullet? there are none of those either or someone would have found it by now!

__________________
Follow me on Twitter! On the Trail with SOSG How I became a Social Media Convert and Twitter and Agents of Influence and now regular poster at Cloudmixer where We're Mixing New Media Ideas.