2. Substantive debate by knowledgeable proponents of all points of view: No good comes when an authority silences substantive debate on an issue.
3. There are no true authority figures: Scientific discourse takes place on a level playing field in which ideas are judged by their merits Summed up in this bathroom graffito: "Ideas by merit, not by source."
4. Use more than one hypothesis: There are often many possible patterns or explanations for patterns. They should all be examined, and the quickest way to get to the answer is usually to examine them simultaneously and not to assume that they are mutually exclusive.
5. Don't get too attached to your hypotheses: The whole point of testing a hypothesis is to try to falsify it. If we don't try in earnest, then we really haven't done much. Alas, the world, including the scientific world, contains true-believers who could never, under any circumstance, be convinced that they are wrong.
6. Quantify: Remember, subjective observations are useless to science. Sometimes, subjectivity can creep into even detailed observations. The true proof of objectivity is often the ability to count or measure.
7. A chain is as strong as its weakest link: When your argument requires a chain of logical steps, each step must be valid or the whole thing is hogwash.
8. Parsimony (AKA Occam's Razor): When there are more than one possible solutions, the simplest one will usually be correct.
9. And the really big one, Falsifiability: In science you have to have a definite answer to the question, "If I were wrong, how would I know it?" If you don't, you aren't doing science. That means it must be possible, in principle, to know if your hypothesis is wrong. If you have no way of knowing if you are wrong, there's really not much to test or discuss scientifically.
Appeals to ignorance:
In effect, this argument boils down to "I can't imagine how something took place, so therefore it didn't take place." Sometimes, this is used if the underlying processes behind an event are poorly known. The problem, of course, is that we can't say something didn't happen just because we don't know everything about how it happened.
This is Latin for "It doesn't follow." It's a favorite of advertisers. We apply the term to cases where someone presents two unconnected ideas as if one were a logical consequence of the other. A special case of the non sequitur is the red herring, a deliberate non sequitur thrown out to confuse one's opponent.
The Argument from Authority:
The idea that we should adopt an idea because some respected person tells us to. This is extremely common in advertising, where a spokesperson is chosen to advocate a product on the basis of popularity or perceived expertise.
The straw man argument is one in which a weak parody for the opposing position is created, then the parody is flamboyantly attacked and torn to shreds. If the deception is successful, listeners never notice that the substitution occurred and imagine that the original opposing argument was defeated.
When a forthright statement is more likely to offend than persuade, it is sugar-coated by the coining of new terminology. E.G.: "Ethnic cleansing," goes down easier than "Mass murder."
Statistics of small numbers:
Attempting to infer a large pattern from too few observations.
Reification: (Not on Sagan's list, but important)
From Latin res - "thing." To "thingify." To assume that because something has a name, it must be concretely real. Things that humans tend erroneously to reify include: IQ and race.
Arguments from adverse consequences:
These are arguments in which one is asked not to accept a position because doing so would require them to accept unpleasant consequences that stem from it.
Correlation equals causation:
It's easy to be convinced that because two things happen simultaneously, one must cause the other. This is not something we can assume. The hypothesis of non-correlation has explicitly to be tested and falsified.
Post hoc ergo propter hoc:
This is Latin for "it came after so it was caused by..." This is a special case of the correlation = causation fallacy in which when one event follows another it is claimed to have been caused by it.
Artificial resuscitation for an argument that has already been falsified by some reasonable standard. This usually assumes the form of an argument that the special circumstances surrounding a particular situation invalidate the falsification.
An unreasonable reluctance to accept a sound argument because of the fear of being drawn from it to accept a similar but less sound argument, and so on until you have been persuaded, after many iterations, to accepted a completely fallacious argument.