Okay, yesterday's poll was full of fail, because it does depend a lot on context. I'm normally the one who can't get more than two questions into a questionnaire without going "ARGH BUT IT DEPENDS" and throwing the whole box of "neither agree nor disagree"s across the room in disgust, so I apologise for giving everybody else the argh this time. I think I was working on the assumption that the hypothetical communications in question were a) personal, or at least something where you personally have an interest (in either sense) in getting a reply, and b) something that obviously invited a reply but didn't specify the time-limit; e.g. if you say "please let me know by the end of the day" then obviously you're kind of expecting a reply by the end of the day, and if you send something that doesn't normally need/expect/invite a reply, then, um, you're not expecting a reply. But I did completely fail to show working.
Not sure what to do now: 48 people have managed to respond despite the fail, so I don't want to make people fill it all in again; but the results are probably a lot more meaningless than I hoped. That'll teach me to try to design surveys when I'm tired and rushing to meet an arbitrary deadline. :-}