There is no such practice. If you need as a parameter a double you declare it double,. If you need as a parameter a vector of doubles (more often a reference to a vector of doubles) you declare it as a vector of doubles.
If you need as parameters a range of iterators of a vector you declare iterators as parameters.
Basically what everyone is saying: If you want to use a sequence of doubles, pass a vector by reference as you will have less problems, but if you only want a single value pass only that value, either by reference or not.