These stats are bad, but it's much worse than just sloppy data gathering in the survey. See the discussion here: https://news.ycombinator.com/item?id=17610221

You can get numbers much like HuffPo's if you compare urban San Francisco (population 800,000) to Toronto (population 3,000,000). Which is grossly misleading to begin with, but they then listed it as "Silicon Valley" in the headline and the "San Francisco Bay Area" in the body. Which are, obviously, three fundamentally different regions.

The CBRE data probably isn't perfect, but projecting their San Francisco percentage change back to 2013 does match Census Bureau numbers surprisingly well. Things get a bit confusing in 2018 because it looks like CBRE changed their grouping from SV and SF separately to the whole Bay combined.

Unfortunately, HuffPo didn't know or care about that distinction, so we got this article instead.