Yes, I pointed that out too: “many respondents use multiple OSes”.

But SO releases the data as CSV after a while, and depending on how they anonymize it the “N users ticked Windows+Ubuntu+Arch+Android” info might sill be available and convertible into “N users ticked Windows+Linux+Android”.

It’s a bit annoying that the survey has one “Windows” and one “MacOS” answer, but the “Linux” answer is split between “Ubuntu”, “Debian”, “Arch”, etc. I get that Android or WSL are their own category, but detailing the traditional distros makes them look less common.

A grouped “Linux” category would likely be bigger than MacOS and close to Windows (hard to say exactly because many respondents use multiple OSes).