Short answer - CC-BY is most common in OA generally but it's not
ubiquitous, and you'd not be alone if you didn't use it.
It's certainly a topic of some debate. Generally speaking, humanities academics/journals are more concerned about it than the sciences (see eg the Royal Historical Society position. Concerns about commercial reuse (NC) are usually more common than about derivatives (ND).
CC-BY is by now more or less the standard in the sciences. Almost all all-OA titles that I've dealt with in the earth sciences & biological sciences use this as the default, though there are a small handful that use CC-BY-NC or CC-BY-NC-ND as their standard license (CC-BY-ND is rare), and a few that run on a more traditional "no license but we're always free to read" basis. I have seen CC-0 and CC-BY-SA papers but this is very unusual.
Hybrid titles (subscription journals with a per-article OA option) tend to have more mixed arrangements, with some offering only CC-BY, some a choice of CC licenses, some a generic "free to read" option, etc. Hybrid titles published by the major commercial publishers tend to have a one-size-fits-all license policy regardless of their subject matter.
Some numbers to illustrate this: in 2014 my institution published 93 CC-BY papers, six CC-BY-NC, and four CC-BY-NC-ND. Four of the NC(-ND) papers were in hybrid titles with a choice of license, the others were required by journal policy. As noted, this is a mix of publishing in the earth & biological sciences.
Meanwhile, much of the drive for licensing comes from funding agencies. These tend to either (a) require the CC-BY license (Bill and Melinda Gates Foundation, Wellcome Trust, Research Councils UK), or (b) not specify a license as such (NSF, HEFCE). UK funders have been more aggressive about specifying a fixed license than US ones.