It sounds like the survey was done post-pandemic and then asked respondents whether they could've done their jobs without the Internet pre-pandemic. This is a

It sounds like the survey was done post-pandemic and then asked respondents whether they could've done their jobs without the Internet pre-pandemic. This is a very different survey: it means that the ~20-30% of Americans who lost their jobs in the pandemic are not included in the survey. This is almost half of the labor force (which has been hovering around 60-65% of Americans since 2009). I'd suspect that there is strong selection bias, in that truckers, factory workers, cashiers, taxi drivers etc. were a lot more likely to lose their jobs in the pandemic and the ones that kept them are much more likely to work for "modern" companies that do require Internet. If you assume 37% of Americans are not in the labor force, 23% were unemployed and hence not in the survey, that implies 70% of the 40% remaining (~28%) require a home Internet connection, with about 3% being those whose employers recently switched to requiring Internet. That seems a lot more reasonable.