More on Evidence vs. Faith

Victor Reppert sees the evidence as a relationship between the likelihood of a fact and some postulated state of affairs.  Given the postulated state of affairs Y is true, if a particular fact X is more likely to exist, then X is said to be evidence in favor of Y.  To put it in Victor's own words:

I understand evidence in Bayesian terms. For me, X is evidence for Y just in case X is more likely to exist given Y than given not-Y. By this definition, something can have evidence for it and be false. - Reppert 

In his view, an individual fact is regarded as evidence for or against the postulated state of affairs.  It may then be possible to judge whether that postulation is true, perhaps on the basis of a single fact in evidence, or perhaps by weighing several pieces of evidence for and against the postulation.

In my view, this way of looking at evidence is wrong, and quite likely to result in incorrect conclusions.

Let's take a simple example.  You live in an isolated area in the woods.  The fact in evidence is peanut shells found in your yard.  You postulate a state of affairs thus: birds broke open the peanuts and ate them, leaving the shells in your yard.  Your husband postulates that the squirrels did it.  So we have two competing hypotheses:

H1: Birds ate peanuts and left the shells behind.
H2: Squirrels ate peanuts and left the shells behind.  

We can call the fact in evidence E: Peanut shells were found in the yard.  But using Victor's Bayesian approach to examining the evidence, both H1 and H2 would make E more likely.  That is to say, E is evidence in favor of both H1 and H2.  Where do we go from here?  We probably need to start looking for other pieces of evidence.  If we are an H1 supporter, we will probably look for more evidence in favor of our theory, and likewise if we are an H2 supporter.  In the end, the one who has gathered the most pieces of evidence in favor of his favored theory wins, right?  Wrong.  The true state of affairs is not to be found by taking this approach, as we shall see.

How should one evaluate evidence in a manner that is most likely to reach the truth?  You need to look at it in a completely different way.  Let's start by gathering as much evidence as we can.  Not evidence in support of one hypothesis or another - just evidence, which consists of any and all facts that result from the true state of affairs.  In reality, any actual state of affairs results in a set of facts, or evidence, being left behind, and that evidence is just what we are able to observe as a result of that state of affairs.  Any complete collection of evidence is always consistent with the true state of affairs.  If we want to determine what that true state of affairs is, we need as many facts as possible, and we need to look at them all together so that we can postulate the one hypothesis from among all candidates that is most likely to produce that complete collection of evidence.  In this view, we consider the relationship between the full body of evidence and the likelihood of a hypothesis, not the relationship between the hypothesis and the likelihood of a single fact.  It is generally not reasonable to consider a single piece of evidence in isolation in support of some hypothesis, unless it happens that there is no more evidence available.

When we create a hypothesis, we must be cognizant of the total collection of evidence.  The hypothesis must be consistent with all the facts in evidence.  On the other hand, any potential hypothesis can be ruled out if there is one single piece of evidence that in inconsistent with it.  So to continue with our example, we start by gathering the evidence.

E1: Peanut shells found in the yard.
E2: Shells are in a single, neat pile, next to the bench.
E3: Peanuts are kept inside the house, and none were placed outside.
E4: There are no neighbors where animals could have obtained peanuts.
E5: No animals can get into the house.
E6: Kids live in the house.
E7: Your husband says he thinks squirrels did it.

Now, if we consider the hypothesis H1, we can rule it out, because birds would scatter the shells rather than leaving them in a pile.  H2 seems a bit more likely, but we know that squirrels can't get into the house, and there are no neighbors they could have taken peanuts from, so we rule out that hypothesis as well.  But we can come up with another hypotheses that is fully consistent with all the evidence:

H3: The kids took peanuts from the house and ate them in the yard.  

Is this hypothesis guaranteed to be correct?  Not necessarily.  It's still possible that

H4: Your husband did it, and lied about thinking it was the squirrels.  

All the evidence we have would also be consistent with H4, so we are unable to rule one or the other of those hypotheses.  Obviously, we need to consider additional evidence.  And in doing so, we need to remain open to the idea that a completely different hypothesis may be in order.

Again, we must gather facts without regard to their support for one hypothesis or another.  If we are leaning toward one hypothesis, we are likely to ignore some important evidence.  You may want to believe that your husband wouldn't lie, for example.  You consider another piece of evidence

E8: the peanuts were bought for the children to eat. 

And therefore, you may want to settle on H3 as your preferred hypothesis.  But objectivity is crucial when gathering evidence.  So you must take note of some additional facts.

E9: The peanuts were purchased yesterday.
E10: The kids weren't told about them yet.
E11: Your husband drank some beer on the bench outside last night.
E12: The shells were found first thing in the morning, before the kids went outside.

At this point, it's not looking good for your husband.  But it is worth nothing that both H3 and H4 are still consistent with all the facts in evidence.  And there is still one additional fact to come to light.

E13: When asked, the kids tell you they found the peanuts and ate them yesterday.

In the end, H3 appears to be the best hypothesis.  But that doesn't mean that H3 has been proven absolutely.  It could still be the case that the kids are covering for their dad, for example.

What can we make of all this?  There's nothing particularly wrong with using Bayesian analysis to evaluate evidence, but it is wrong to limit consideration to any single piece of evidence, or to anything less than all the available evidence.  A hypothesis should be formulated to take into account every available piece of evidence.  A single piece of evidence can be used to exclude a hypothesis from further consideration.  An isolated piece of evidence should not be seen as favoring or opposing a hypothesis.  Rather, the totality of evidence supports the best hypothesis.  The totality of evidence is always consistent with the true state of affairs.  If the true state of affairs includes some deception, then the evidence may be deceptive, but the evidence is always consistent with the truth.

To put this in perspective, Victor still insists that religious belief is based on evidence, not faith.

Oh please. So, the famous book in defense of Christianity was called "Faith that demands a verdict?" - Reppert

Faith is what leads one to ignore evidence - to fail to account for the full body of available evidence.  The book Evidence that Demands a Verdict does just that.  It has been extensively reviewed, and as the reviewers say, the jury is in and verdict has been handed down.  Infidels.org has compiled a series of essays that examine this evidence much more objectively than Josh McDowell has done, and place it in context within the greater body of evidence available to anyone who isn't blinkered by religious faith.  In addition, they have made a list of other reviews of McDowell, found here.  The bottom line is that if Christians like Victor (and Josh McDowell) want to claim that their belief is based on evidence, then they need to look at all the evidence, not just the facts that support their favored hypothesis.  Otherwise, they're just letting faith govern their view of evidence.