> Do you have evidence you could point to, or is this a instinctual opinion?
The Sony PlayStation is based on a fork of FreeBSD. It is entirely proprietary. Lots of different embedded device also use BSD because they are permissively licensed. Anybody who gets such a device is screwed because they cannot modify it or even know what the device is doing. This is true for the majority of IoT devices (which is a serious problem going forward). To be fair, there aren't nearly enough GPL enforcement lawsuits as there should be (which is why I donate to the Software Freedom Conservancy) so it's not like the copyleft side is /much/ better but at least copyleft does help achieve the goal.
In addition, GPLv3's anti-tivoisation clauses are quite unique in the free software world and are specifically written to ensure that users have freedom over devices that try to undermine the spirit of the GPL. You won't find any permissive license that has such protections for your freedom (because it's not enforceable without making the license copyleft).
This argument doesn't hold for standards, generally. Because what matters for a standard is adoption. Sure, you're making it possible to have a proprietary fork of your reference implementation but at least users can be sure that the specification is actually used (and if the spec is Apache licensed then you're safe from patents which is something that can't be said of proprietary specifications). But Apache OpenOffice isn't a standard, so this is not really all that relevant.