Science is a method; period.
Science is a method humans use in order to produce a reproducible experiment -or sets of experiments- in order to collect data. The collected data is only as good as the tools we utilize to do our measuring. As such, when our tools improve, so does accuracy of the data we collect.
There are a number of issues with this. The data can be incorrect, or incorrectly attributed as a causation instead of a correlation. The data can be manipulated in order to prove a bad theory. The tools can be flawed or not precise enough. The peers reviewing the data can be manipulated into consensus. The theory itself may be fundamentally flawed, but the inaccuracies of the tools used can seem to prove the theory (see: electron shells and superposition).
Science -at its most basic- is a checklist. That checklist ensures that an experiment is both reproducible and as accurate as our data gathering tools allow -at the time.
When our tools improve, change, or become obsolete our understanding must also. That's why to some people science always seems to have two heads and is double-tongued.
As with most other things, education is the key to solving this bad-perception of science, scientists, and the data collected from experimentation and observation.