Once again, David Colquohoun (a noted physiologist at University College London), has a number of important points to make about the decline of logical and scientific rigor in daily discourse and institutions. Read his post in the Guardian, and check out his website.
As we approach science education, at all levels, we need to move away from generating the impression that science is about authoritative statements, and more about a critical, progressive, and self-correcting social process.
The first step is to encourage students to go beyond simple answers, and require them to state explicitly what assumptions they are making or are implied by their answer. They should then have the opportunity to consider which assumptions are well-supported, which are shaky, and where they might like (need) more data. In a way, you are asking them to perform a simple Bayesian analysis, in which they calculate the probability that their assumptions are correct (and so the strength of the data needed to make confident conclusions - weakly probable assumptions required more data!)
The process including making sure that they when they use a scientific term, they can explain what it means.