Anecdotal Evidence

ANECDOTAL EVIDENCE are just comments based on a simple high level of observation with no samples of typical cases to back it up.   In other words it is a sample of ONE or to few to have any true value.  It is not used in establishing factual information with the scientific method nor used in engineering.  

An analogy is something like driving a car with only one bolt holding the wheel rim to the wheel hub.  (Wheels are normally secured to a car using 5 special bolts that use reverse threads.)   The owner of the car observes that it works just fine driving the car around town.  Hence his conclusion is  it must be OK then to do.   A friend of his borrows the same car and decides to drive it on a race track for fun.   The car crashes because the wheel fell off.  In other words, the 5 bolts exist for wheel reliability under any and all operating conditions avoiding a wreak and possible loss of life.  The owner was wrong.

EMPIRICAL EVIDENCE can start out as a anecdotal based but has more direct observation with a lot of samples or cases and/or some limited experimentation to back it up. Good vs bad performance has to have a working definition to go with the evidence.  It also requires a good understanding of what your looking for and why.  If you do not know what to look for, the evidence has no clear value.  Empirical evidence is great in forming a theory but it still needs to be tested to prove or disprove it. 

SCIENTIFIC EVIDENCE (AKA FACTS): Only by following the scientific method involving carefully setting up the experiment, doing quantitative measurement, simulation and testing can one establish if this theory to be true or not.  In other words, only when we fully understand why the empirical evidence is true does the evidence become truly useful facts.

References:

http://en.wikipedia.org/wiki/Scientific_method
http://en.wikipedia.org/wiki/Anecdotal_evidence
http://en.wikipedia.org/wiki/Empirical_evidence