Eg: In Q17, the respondent have to say that they work outdoors every day. If the majority does only say that they works outdoors some days - or even most days - the algorithm infers that the job can be done fully from home.

And the second top level comment from that:

>The coding used by this survey is a joke. If 40% of an occupation says they work outdoors that's below 50% so it's bucketed into the work from home category on that criterion. I think if even 20% of a job is saying they work outdoors that's pretty strong evidence you're missing some information about what the other 80% are doing in their job that prevents it from being a work from home job. The net result of their setup is obviously a vast overestimate so take the 37% number with a grain of salt.