This is a good question. I have never encountered it myself either.

I will add the following.

If you have a sample and want to know the true proportion of participants voting "yes" and your estimator or sample proportion is 'p' then '1-p' will represent those voting "no" and "maybe".

Likewise

If you have a sample and want to know the true proportion of participants voting "no" and your estimator or sample proportion is 'p' then '1-p' will represent those voting "yes" and "maybe".

and finally,

If you have a sample and want to know the true proportion of participants voting "maybe" and your estimator or sample proportion is 'p' then '1-p' will represent those voting "no" and "yes".

That's my thinking, happy for others to correct me.