Why should a hypothesis be developed before experiments take place?

A certain piece of data can have an infinite number of hypotheses to explain it. This leads to a condition known as, "underdetermination." As such, collecting various pieces of data indeterminably does not improve scientific investigation. In fact, collecting data in such a manner can overwhelm a researcher. However, having a hypothesis before experimentation sets up a researcher to look for certain specific types of data in the process of testing it. A researcher therefore avoids under-determination by having a hypothesis that guides towards looking for particular data.