Science, in its intended form, relies on purity of understanding based on evidence and data. Anything else is biased and is therefore not pure science.

Data collections can be misconstrued, misused, and their samples can be flawed, the same with evidence, and perception can easily screw that one up. Models that predicate future occurrences while using past data, easily flawed, and in every single case I've seen, flat out false. Cannot ignore these.
Error and falseness is every bit a part of science and it's evolution as the things that are at one point shown to be true.
While I don't partake to the logic (or should I say lack thereof) ascribed to ghosts I don't think some understand that science and it's method are not absolute truths and needs the same amount of scrutiny and scepticism as everything else.