You’re both right. Thanks for the catch – I’ve updated the original text to reflect what Caroline proposed.

Andy

 All roses are flowers
Some roses have thorns.
Therefore, some flowers have thorns.

If you leave common knowledge out of this, saying “some flowers are roses” could allow for the possibility that some roses are not flowers. Then you could not assume that the roses that do have thorns are the same ones that are flowers.

1.  Understand what objective evidence is.  My definition:  objective evidence results when all observers who follow the same protocol achieve the same results, regardless of their personal beliefs. (“Same” means within a given statistical variation determined by the limits of the protocol).
2.  When using logic, understand the important of the “basis” (or first axiom) for your argument.  This basis must be considered true by all for the purposes of the argument, and cannot be the result of the proof.
3.  The endpoint of recursive or reductive logic must be defined up front.  It cannot merely be a stopping point that is convenient for the argument.  Also, the stopping point cannot be what you were trying to prove in the first place.
4.  Most importantly,  remember that logic is an invention of the human mind.  Any number of things that are not true in the “real” world are quite logically consistent.  While logic is very useful for modeling the behaviour of the universe,  in no way does it control its behaviour.
5.  Remember that probabilities always sum to unity.  Do not suggest that something is improbable unless you are just as willing to defend the proposition that the complement is highly probable.  For instance, if you feel that it is highly improbable that self-replicating molecules could arise in a nature, then you are implicitly stating that you can prove that with a high probability that molecules can never form self-replicating structures.  The latter is a much harder proposition to defend, as you must find some law of physics the supports this.
6.  Never forget the Lottery Fallacy.   What is improbable for the individual may be highly probable (or a certainty) given the entire  population.   The human mind seems hardwired to commit this error.
7.  Remember that statistical error never improves with the repetition of independent samples.  It does not matter how many UFO’s, ghosts, religious miracles, etc are reported.  The errors inherent in such data remain the same (or get worse).
8.  Remember that uncalibrated experimentation is useless.  Until someone can calibrate their EMF meter or EVP recorder to detect when a ghost is not present,  the devices are useless in determining that a ghost is present. No matter how many times you use it!  (see #7).  This includes using a “psychic” to confirm EMF or EVP data.