Regarding hypotheses:

A hypothesis is an explanation proposed for a set of observed data.  Why do we see the pattern that appears to be there?  Any hypothesis is only as good as the last piece of supporting data.  A hypothesis can never be proven, only supported or refuted.  Which brings up a problem that is all too common.  A good scientist is led by the data.  That means one does not set out to find data to support a proposed hypothesis.  Doing so leads to ignoring data that don't.  One collects data relevant to the question that is of concern and then tries to figure out an explanation that fits.  If the data don't fit the original hypothesis, then the hypothesis is wrong, the data are bad, or both.  You have to figure out which and adjust accordingly.  And you are obligated to continue to collect data to test and retest the hypothesis.

Regarding assumptions:

Much of my work involves solving puzzles with most of the pieces missing.  It also involves training others.  For both tasks, I have to ask what evidence is available.  What explanation best fits it?  Is there another explanation that fits as well?  Are assumptions being made for which there is no evidence?  Are those coloring the explanations?  I am continually frustrated by the fact that most people jump to conclusions based either on insufficient (or no) evidence and on assumptions that are insupportable.  I frequently see the same thing in publications and even more so on the internet.  I'm afraid I have very little patience with it in others and none in myself.  Any good scientist should feel the same way.  It means I frequently disagree with so-called "received wisdom" that doesn't measure up as well as with reports of the "latest discoveries," which all too often have half-baked conclusions.  By "received wisdom," I mean century-old hypotheses that have not been questioned in the light of the data that have accumulated since.  These become the assumptions on which new hypotheses are based, building a very shaky house of cards.  Fortunately, science is self-correcting, and sooner or later a re-evaluation will be forced.