That's actually a pretty decent point and illustrates one of the limitations of the scientific method. However, if this assumption is discarded, what you have is no longer science. Science basically says, "hey, we're only going to deal with things we can measure and test with statistics; we can't comment on things outside of that." Some people twist this or misunderstand this as "if it can't be measured or tested with statistics, it isn't important, isn't valid, or doesn't exist." An honest scientist won't say that; they'll recognize that materialism is one of the limits of the scientific method. It's why science can't disprove something like the existence of the Christian god or many other metaphysical claims. It irritates me when folks say that it can.

There are a few instances where science can deal with "immaterial" things, particularly in the soft sciences. They're soft for exactly that reason; they make more assumptions about things and more frequently run into the problem of falsifiability. For instance, Freud's ideas about the id, ego, and superego are fantastic, but they can't be proven false. It's a useful way of understanding human psychology, but it's not the only way we can think about it (as evidenced by the many other major branches of thought about human personality).

Social sciences are great. They remind me in many ways of working with the occult. You create a paradigm - a map - and use it to interpret the territory. You understand the map isn't the territory, but the map can be very useful for certain purposes.