Sep 30

Historical Method: Tell me what is probable.

This is my second post in a series on the historical method. In my first post, I rejected the claim that skeptics are committed to a double standard for having a bias against magical claims. The reason for this has to do with probability, which will be a recurring theme in this series. This post will simply introduce the idea that we have to consider probability if we hope to make any progress.

Tell me why a given solution to a historical question is probable, rather than possible. I know I say this a lot, but it bears repeating. This has to be the starting point for resolving any historical question.

Imagine you are a historian and I come to you with some records written about Julius Caesar claiming that he was a god. Now, this claim seems to contradict everything we know from science and other sources, which leaves us with a problem. Let’s consider some possible solutions to the problem:

  • The claim is correct, meaning Julius Caesar was a god and everything we think we know is false.
  • The author was part of a conspiracy to puzzle later historians.
  • The author was himself a god, but wanted to divert your attention elsewhere so he could be left alone.


Strictly speaking, these three solutions are all possible. But what do they have in common? They are all so unlikely that to describe them as “vastly improbable” would be an understatement. If we do not introduce probability, then we really have no reason to prefer any one possible explanation over any other. This means we should restrict acceptable proposed solutions to those that are relatively probable. So, the following explanations would be preferable:

  • Rulers often demanded that people describe them as gods.
  • People of the time did not have our current understanding that such gods probably do not exist, so the author did not know any better.


Hopefully you see why we might want to restrict solutions to the second kind of answer. Historical claims about Christianity are no different. If someone wants to propose a solution for something, like the virgin birth contradicting scientific knowledge or the differences in gospel genealogies, then they should provide reasons why their solution is probable. Then, we can compare it to reasons for an alternative solution. We may not be able to operate with the precision of comparing two options in Blackjack, but in many cases we can have a rough idea.

Similar Posts:


Skip to comment form

  1. Randy Everist

    Hi Dr. Mike! 🙂 Something interesting is that many, if not most, Christians who defend the resurrection do so not on grounds of probability, but on plausibility as an inference to the best explanation.

    They do not take the claim “God raised Jesus from the dead” and then try simply to decide whether or not it sounds good. They considered first century evidence.

    However, what makes something probable, in contrast to merely plausible, is related tightly with background knowledge. So, in order to reject the hypothesis “God raised Jesus from the dead” a priori as impossible, it must be impossible for God to exist or assumed that he doesn’t, something for which we have been given no argument to plug in. Now to reject it as improbable a priori requires the same issue. However, even if God’s existence is given to be improbable, a priori probability is only one part of the calculus.

    We must also consider the probability of the evidence being true given the hypothesis’ truth, and the probability of the evidence being true given the hypothesis’ falsehood. That is to say, given the “three minimal facts approach,” is it more or less likely the facts would be true were the hypothesis “God raised Jesus from the dead” to be true. I think it’s clear that the combined facts of Jesus’ being dead and buried in Joseph’s tomb, his tomb being discovered empty, and his postmortem appearances are raised in probability given the truth of the resurrection, and lowered given its falsehood. This seems to me to be obvious.

    After this, we then assess competing explanations, as we are trying to find out which hypothesis best accounts for the evidence. In order for an explanation to be better, it must raise the probability of the three statement’s being true given that particular explanation to a degree in the calculus that the explanation’s overall probabiltiy is higher than the hypothesis “God raised Jesus from the dead.”

    So that if we hypothesize that God miraculously saved Jesus from the cross, that makes the three facts enormously improbable given that explanation’s truth, which means that explanation has less probabilty and hence less explanatory power. One must not only provide something that is just as good or better, but they should have first century evidence to back it up!

  2. Mike

    Hi Randy. I personally like Carrier’s attempt to bring in a Bayesean (ian?) method into history. I know there are Christians who engage in the same sort of thing, like I think the McGrews do. I just want the assumptions on the table for evaluation, and if we can have some kind of objective rubric, then even better.

    Either way, I think plausibility and probability are getting after the same sort of distinction from explanations that are merely possible.

  3. Randy Everist

    Yes, I think Tim McGrew (with whom I have the pleasure to interact with on occasion), convincingly shows that even if God’s existence is seriously improbable, the Resurrection Hypothesis comes out way ahead on Bayesian probabiltiy, which I was trying to describe.

    There is a distinction between the two, and probability relies on numeric values, which are notoriously hard to place in accurately in a being who exercises free will, for example. Yet we may find some things quite plausible that we cannot quantify in probability with much confidence. I have no idea what the probability is that I will choose to sleep in later than my alarm tomorrow. It could be .7. I can tell you that I don’t sleep in too often, but it’s also the weekend, and I am staying up late. Thus, I think it’s plausible I will sleep in tomorrow, even if I can’t give you a non-arbitrary probability figure.

  4. Mike

    I’m not in principle against the McGrews’ approach to this and other matters. I may disagree on finer points, but I think they’re quite good at what they do.

    This is meant to be pretty high level, so I think it will suffice to say that we want to get at things like likelihood so that we will have an objective standard to eliminate certain alternatives.

  5. Mike

    Sorry, I was running around earlier. I just reread your first post and I think we’re in agreement there with two caveats. First, I think at least putting a number range helps to properly assess the weight something is being given. If someone can’t give a range, then perhaps we ought to wonder how important the data is. Second, if a posterior probability remains very low after including evidence and knowledge, then we don’t necessarily need to compare it against competing theories. If something is 0.1%, then we can be pretty confident that some other theory is right. If they’re all low, we may just have to say we don’t know what’s right for that case.

  6. Randy Everist

    Mike, I see. 🙂 I think we’re largely in agreement. One interesting aspect of the last sentence is that one could then embrace the RH as the best explanation, but essentially shrug one’s shoulders. If all competing theories are enormously improbable, including the RH, then one may say while it is the best explanation, one may come along later, and in any case it is still more probable it did not happen. I can respect that approach. It just so happens I think the probability comes out somewhere higher than .5. But I’m not so sure that should be our stopping point.

    Suppose every naturalistic and supernaturalistic alternate explanations for the relevant data all came out to, say, .05, at best. Suppose further the RH comes out to .45. Should we still accept the “best explanation” title but choose to abstain from belief? I’m not so sure. I would argue one would be justified in that case, for no competing alternatives that we can think of are any closer than nine times less probable than RH, and RH is very nearly just as probable as not, and considering margin for error in values, should be considered that way. That’s all hypothetical, but I don’t want people to get the impression that we ought to remain agnostic unless we are near-certainty or things like that.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Notify me of followup comments via e-mail. You can also subscribe without commenting.