“Observations are not simply vetted by presuming the reliability of their source, but also the by probative value which they may serve to impart to the question at hand. If you reject highly probative observations, simply because you have made an assumption as to their reliability – this is a practice of weak science. ” — The Ethical Skeptic at https://theethicalskeptic.com/2018/01/07/anecdote-the-cry-of-the-pseudo-skeptic/
So, if reliable and probative are independent concepts (orthogonal), it follows that we can make a two dimensional matrix for them, and classify various things on both dimensions, to see where they fall. Maybe instead of the five dollar word probative, we should use the one dollar word ‘relevant.’ For the word reliable, we could use ‘certainty that it is correct.’
So, what fits into the matrix cells? I have no examples at present. However, the idea seems interesting.
|Unreliable, low certainty of correctness||Highly reliable, high certainty of correctness|
|Strongly probative, relevant to the situation||Not so good: relevant, but not at all certain||Best evidence: both certain and relevant|
|Weakly probative, irrelevant for the situation||Poorest evidence: neither certain nor relevant||Not so good: certain, but no bearing on the situation|
The ability of a piece of evidence to make a relevant disputed point more or less true.
For example: In a trial of a defendant for murder, the defendant’s dispute with his neighbor (unrelated to the crime) has a no probative value because it provides no relevant information to the trier of the fact. However, if the defendant’s dispute was with the victim, this has a much higher probative value as it could be a motive for the murder.”