Is Belief in the Resurrection Rational?
A Response to Michael Martin
by Stephen T. Davis
The following article is from Philo,
Volume 2, Number 1.
Abstract: This essay is a response to Michael Martin's "Why the
Resurrection Is Initially Improbable," Philo, Vol. 1, No. 1. I argue that
Martin has not succeeded in achieving his aim of showing that the Resurrection
is initially improbable and thus, by Bayes's Theorem, implausible. I respond
to five of Martin's arguments: (1) the "particular time and place
argument"; (2) the claim that there is no plausible Christian theory of
why Jesus should have been incarnated and resurrected; (3) the claim that the
Resurrection accounts in the New Testament are unreliable; (4) Martin's
assumptions about how one establishes the initial probability of Resurrection;
and (5) the use Martin makes of Bayes's Theorem to discredit belief in the
I consider it an honor to have had some of my own arguments about the
Resurrection subject to careful review in the initial issue of Philo.
I was happy to note that with one exception (to be noted below), Professor
Martin gets my argument right. But I do not think he succeeds in making his
case. Let me discuss five points of disagreement between Martin and me.
1. The particular time and place argument. As part of his program for showing
that any alleged miracle, and especially the alleged miracle of Jesus'
Incarnation and Resurrection, is very highly improbable, Martin argues that,
even if it is likely that there be an Incarnation:
God could have become incarnated and have died for sinners on an indefinite
number of other occasions. There does not seem to be any a priori reason to
suppose that He would have been incarnated at one particular time and place
rather than at many others. Consequently, even if some incarnation or
Resurrection or other is likely, there is no a priori reason to suppose that He
would have become incarnated and died as Jesus in first-century Palestine.
Indeed, given the innumerable alternatives at God's disposal it would seem a
priori unlikely that the Incarnation and the Resurrection would have taken place
where and when they allegedly did.
But this argument does nothing to render the Resurrection of Jesus unlikely,
because it equally counts against the probability of virtually any future event.
Suppose (as I assume) the probability is nearly 1 that at some time in the next
three years, Michael Martin will sneeze. But on Martin's argument, since there
are very many moments between the present moment and three years from now, the
probability that Martin will sneeze at any one of those moments, and thus the
probability that Martin will sneeze at some time or other in the next three
years, becomes vanishingly low. But that can't be right; surely if as originally
assumed the probability that Martin will sneeze at some time in the next three
years approaches unity, then at some moment in the next three years Martin will
sneeze, despite the improbability of that particular moment being the favored
Martin's "particular time and place argument" is so strange that it
causes me to wonder what notion of probability he is working with. Again, if the
probability of the Resurrection of Jesus is said to be low, based on this
argument, then virtually every future event equally becomes vanishingly
improbable. Is that what Martin wants? I am truly puzzled here. Perhaps the
argument is simply a mistake on Martin's part; perhaps he is simply confusing
(a) the improbability of specifying in advance where and when some event takes
place with (b) the probability of the event's occurring (at some time and place
or other). But if this is not a mistake on Martin's part, if he really means his
argument to count against the Resurrection of Jesus, then at the very least he
is going to have to say a great deal more to explain and defend the (at least
apparently) bizarre notion of probability he is presupposing.
How did God decide when and where in history the Logos would become incarnate
as a human? I have no idea. Just for fun, let's suppose that God used a
randomizing device to pick the favored time and place. Even if so, it is
important to note that some temporal moment and some geographical location or
other would have been picked. And it is not a serious argument against the
occurrence of the event to point out, as Martin does, the improbability of it
occurring at precisely the time and place that it (allegedly) did occur?
2. There is no plausible theory of why the Son of God should be incarnated
and resurrected. Here Martin argues that Christians are unable to offer
convincing theological arguments why the Incarnation, death, and Resurrection of
the Son of God should occur, i.e., why they should be considered antecedently
probable. All the historically important theories of atonement in Christian
history, he says, "either fail to explain why God sacrificed His Son for
the salvation of sinners or else make the sacrifice seem arbitrary."
Now the purpose of a theory of atonement is to explain how or why it is that
the death and Resurrection of the Son of God brought about the possibility of
salvation for human beings. There are several important atonement theories in
Christian history. Martin chooses the "ransom theory" to discuss and
criticize, which I find extremely odd, since almost nobody has held or defended
this theory for centuries. The earlier and cruder version of the theory (from
the church father Origen) claimed something like this: due to our sinfulness,
human beings are in the clutches of the Devil; in order to convince the Devil to
release us, God offered to pay the Devil the ransom of the death of the Son of
God; the Devil agreed to this bargain and released us when Jesus died on the
cross; but God fooled the Devil by subsequently raising Jesus from the dead.
Martin then offers several theological criticisms of the theory, some of which I
assume would be accepted by the vast majority of contemporary Christians. I have
never heard of any Christians after, say, Anselm who accept this theory as the
best explanation of the atonement.
Doubtless Martin selected the ransom theory to criticize as a kind of easy
target. But rather than object to this move, I am going to suppose that the
crude ransom theory is true, despite the objections of Martin (and of virtually
all recent and contemporary Christians). That is, I am going to suppose that
something like this really is the reason that God brought about the Incarnation,
crucifixion, and resurrection of the Son of God.
Now if the ransom theory is true, we have a reason for considering the
Incarnation, Crucifixion, and Resurrection antecedently probable. "But the
theory can't be true," Martin will respond, "look at my
objections." Well, I don't think the ransom theory is true either, but, if
I were a defender of it, I think I could answer Martin's criticisms. Maybe the
Devil really does exist (contrary to Martin's intuitions); maybe God is
arbitrary; maybe God does allow the Devil to take forensic possession of
sinners; maybe morally perfect beings can deceive other beings in certain
circumstances; maybe only those folk who believe in and follow Jesus are the
ones who salvifically benefit from atonement and are released by the Devil.
It seems to me that Martin holds that atonement will work as an explanation
of salvation and as a way of making Incarnation and Resurrection antecedently
probable only if somebody is able to come up with a theory of atonement that is
theologically plausible to Martin. Now maybe there are possible worlds where
somebody is able to do that, but I strongly suspect that the actual world is not
one of them. But even if I am right about that, it will not much matter. The
logic of the situation is this: since I was arguing in my book that belief in
the Resurrection is rational not for Martin but for Christians, then the real
question is whether Christians have got a theory that is plausible to them and
that makes Incarnation and Resurrection antecedently probable. I do not intend
to discuss here my own preferred view of atonement; the point is that Origen and
his ilk (whatever Martin and I may think of their theory) have good reason to
consider Incarnation and Resurrection antecedently probable.
3. Martin's specific objections to the claim that Jesus was raised. Here I
must mention the one point in his paper where Martin misrepresents me. He thinks
I admit that the biblical testimony to the Resurrection is unreliable; he quotes
me as saying that the biblical testimony "was written years after the event
by unsophisticated, myth-prone people who were more interested in formulating
statements of faith and furthering Christian ends than writing accurate history.
Furthermore, the evidence they present is contradictory."
But in this quotation I was not speaking in my own voice; I was citing what I
take to be the one of the strongest arguments of nonbelievers in the
Resurrection. Although there certainly are discrepancies in the accounts of the
Resurrection in the New Testament (one or two of them are rather difficult), I
do not hold and would never say that the accounts are contradictory, and
certainly not about the essential lines of their witness. Moreover, I think the
writers were interested in recording accurate history (as well as, of course, in
furthering Christian ends). Indeed, one sure way for their efforts to have
impeded rather than furthered Christian ends would have been for them to have
written false accounts. The biblical testimony is, in my view, very reliable.
So I think the case for the Resurrection of Jesus based on the accounts in
the New Testament is exceedingly strong-far stronger than any of the alternative
theories like fraud, myth, hallucination, wrong tomb, swoon, conspiracy, etc.
That was the burden of my argument in chapters 1-4 of Risen Indeed.
One Bayesean way of making this point is this: the probability of the
overall evidence for the Resurrection being as it is on any one of the
alternative theories is exceedingly low. By far the best way to account for the
state of the evidence is to say that God did, as claimed, raise Jesus from the
As part of his aim of showing that the biblical case for the Resurrection is
not overwhelming, Martin notes that there are people within the Christian circle
who deny that the Resurrection of Jesus occurred (in the full-blooded sense of
the word Resurrection that we are using here). He points out that they are not
naturalists; thus (he says) the crucial issue in debating the Resurrection is
not the presupposition of naturalism versus supernaturalism (as I claimed in
Risen Indeed). But that there are supernaturalists who deny the Resurrection of
Jesus is no surprise to me, nor ought it be to anybody. The vast majority of
supernaturalist Jews deny that Jesus was raised; so do all the Muslims I've ever
heard of. But so what? I argue that belief in the Resurrection is rational for
Christians, not epistemologically obligatory. (Whether it is theologically
obligatory for Christians is of course another question.) Moreover, many of the
Christian scholars whom Martin has in mind are not supernaturalists at all but
something more like Deists. That is, although they believe in God, they do not
allow that interventionist miracles ever occur, and thus certainly not the
Resurrection of Jesus.
Martin raises five specific criticisms of the biblical accounts. Let me list
each of them and respond briefly. First, there were no eyewitnesses to the
Resurrection. What is correct here is that there were no eyewitnesses to the
Resurrection event itself whose direct testimony we have in the New Testament or
elsewhere. But presumably Jesus himself was an eyewitness, at least to that part
of the event that occurred after his revivification. But more important, even if
Martin is right, it will not much matter. If you saw me today with my hair a
certain length and then saw me next week with much shorter hair, you would be
with your rights in concluding that I had had a haircut, even if you did not see
the event occur.
Second, apart from Paul, there were no contemporary eyewitness reports of
Resurrection appearances of Jesus. Well, I suspect there were plenty of
contemporary eyewitness reports; the problem is that the eyewitnesses did not
write them down or did not write them down in a form that has survived to our
day. What is true in Martin's claim is that apart from Paul there are available
to us today no contemporary eyewitness reports of the Resurrection of Jesus.
Still, I think Paul's report counts a great deal. The apostle also supplies a
list of other individuals and groups to whom the risen Jesus appeared (1 Cor.
15:5-7). Some of the appearances Paul lists, above all those to Peter and to the
"Twelve," are confirmed elsewhere. An eyewitness himself, Paul names
the other eyewitnesses that he knows of.
Third, the empty tomb accounts are inconsistent, secondhand, and late. I
tried to answer these points in chapter 4 of Risen Indeed. Yes, some of the
written accounts are late (e.g., John 20:1-18), but most biblical scholars agree
with me that the empty tomb tradition goes back to the earliest proclamation of
the Resurrection. And although there are a few (not many) discrepancies that are
difficult to harmonize, the four empty tomb stories all agree on the main
points, viz., that early on the first day of the week certain women, above all
Mary Magdalene, went to the tomb; they found it empty; they met an angel or
angels; and they were either told or else discovered that Jesus was alive. In
addition, there is striking agreement between John's gospel and one or more of
the synoptic gospels that the women informed Peter and/or other disciples of
their discovery; Peter went to the tomb and found it empty; and the risen Jesus
appeared to the women, and gave them instructions for the disciples. In short,
the empty tomb tradition is very well established, and its central claims are
Fourth, New Testament scholars differ on when the stories of the empty tomb
entered the Christian tradition. This is true, but as I just noted, the majority
have come to the view that the tradition is very early indeed.
Of course truth is not decided by majority vote; still, in response to Martin's
fourth point, it is important to note the fact that I have just cited.
Fifth, why should we consider the alleged eyewitnesses to the Resurrection
appearances, or those who heard and passed on their accounts, or those who
recorded their accounts, trustworthy? Well, I suppose one decides questions like
this on the basis of (a) the inherent believability of the accounts that they
give, and (b) independent information that one may have about their character
and degree of integrity. I am sure that Martin is unsatisfied with these people
on both counts. But I am satisfied. And I do not know how to adjudicate this
difference between us apart from rehashing the same points that we have been
4. The intellectual difficulty of Resurrection. Martin is right that I insist
on this point. That God would raise a man from the dead is surprising indeed.
Here I was arguing against those apologists for the Resurrection who hold that,
in the light of any rational assessment of the evidence, unbelief in the
Resurrection is irrational. I do not think that this position attends closely
enough to the intellectual difficulty of the very idea of Resurrection. But is
the Resurrection initially improbable? Well, when asked in that way, the
question is not well formed and affords no clear answer. The answer will depend
on who is asking. If the one who is asking is a naturalist (somebody like
Bertrand Russell), then the Resurrection of Jesus will be judged to have a very
low initial probability. And as I noted in Risen Indeed (and as Martin also
points out), even supernaturalists are suspicious of most miracle claims. Strong
evidence must be produced to render the Resurrection believable.
But what if the one who is asking is a Christian supernaturalist (one who,
let's say, believes most of what Christians believe but is unsure whether to
believe that God raised Jesus from the dead)? Then the situation is different.
Then the initial probability of Resurrection will not be judged to be low at
all. As Richard Swinburne correctly points out, Christians expect that God will
want to redeem human beings, and one aspect of redemption is redemption from the
power of death. And part of the Christian theological
inheritance is the Old Testament, in many parts of which the promise of eternal
life is held up (see, among other texts, Daniel 12:1-3). So
given Christian theism, I think interventionist miracles are to be expected, as
This raises the question: Whose background knowledge is allowed to count in
determining the initial probability of the Resurrection of Jesus? Martin uses
the phrase, "our background knowledge," and
that makes me wonder who is included in the range of the pronoun
"our." I am worried that only people with beliefs like Martin's are
meant to be included. My suspicion is deepened by Martin's phrase,
"background beliefs shared by atheists and believers alike."
Presumably the beliefs that atheists and believers share are beliefs like:
Grass is green;
San Francisco is north of Los Angeles;
7 + 5 = 12;
The moon revolves around the earth.
But there are also lots of beliefs that atheists and believers do not share,
God wants to redeem human beings;
Daniel 12:1-3 is part of inspired scripture;
God occasionally performs interventionist miracles.
If I am right that Martin means to include only beliefs of the first sort,
then I respectfully reject his entire program. I will not allow anyone to tell
me that beliefs of the second sort—which I hold and consider rational—cannot
count in figuring the initial probability of the Resurrection.
Now I have argued on a number of occasions that one strong consideration in
favor of the Resurrection of Jesus is that nonbelievers in it are unable to come
up with a coherent and plausible story that accounts for the evidence at hand.
All of the alternative hypotheses with which I am familiar are historically
weak; some are so weak that they collapse of their own weight once spelled out.
In response to this point, Martin concludes his paper with the following
But Davis is mistaken. It is not necessary for a supernaturalist to give a
likely alternative in order to reject the Resurrection. Bayes's theorem
indicates why. If the initial probability of hypothesis H is extremely low on
the basis of background knowledge K, then the Evidence E does not have to be
very probable on the basis of the falsehood of H and the truth of K to indicate
that the probability of H is below .5 given E and K. The biblical evidence need
not be highly probable in terms of alternative hypotheses to show that the
probability of the Resurrection is below .5. A very modest probability, indeed,
will do. 
My response to this is that the initial probability of the Resurrection is
not "extremely low," once we look at it from what I regard as the
correct set of assumptions, viz., those of Christian supernaturalism. (From the
point of view of naturalism or Deism, Martin is correct, as I have always been
prepared to admit.) Moreover, it is important to note
that evidence can only fail to make a given hypothesis H probable if it renders
probable instead the disjunction of all the competing hypotheses. But suppose
none of them are any good. Perhaps this will be because they all seem
historically implausible. Perhaps it will be because they cannot explain why
there is little historical evidence in their favor. Perhaps it will be because
they are unable to account for known facts (e.g., that Jesus truly died on the
cross, that soon thereafter the earliest Christians were firmly convinced that
Jesus had been raised by God). Then the first hypothesis retains its overall
What exactly is the correct way to assess the probability of the truth of
testimony to extraordinary events? Following Hume and many others, Martin thinks
that one assesses it in terms of (a) the probability of the event in question
and (b) the probability that the witnesses are telling the truth. But that can't
be the whole story, otherwise we get into situations where we would have to
disbelieve a witness whom we know to tell the truth 99% of the time who reports
that the number 893420 was the winning number in yesterday's lottery. That is,
the probability must also be determined in the light of (c) the probability of
the witness reporting as she did had the event not taken place.
In the case of the Resurrection, that means that we must assess the familiar
alternative hypotheses, e.g., fraud, myth, hallucination, wrong tomb, swoon,
conspiracy, etc. Thus, even if the initial probability of the Resurrection is
not high, the probability of the testimony's being true may still be high enough
to make it rational to believe it.
To put the point more formally, Bayes's Theorem requires that
Prob (R/K) 3 Prob (EH/K&R)
Prob (R/K & EH) =___________________________________________________
Prob (R/K) 3 (Prob (EH/K&R) + Prob (AT/K) 3 Prob (EH/K&AT)
(where R = the Resurrection hypothesis; K = the background evidence; EH =
specific historical evidence; AT = various alternative hypotheses (myth,
conspiracy, etc.); Prob (R/K&EH) = the overall epistemic probability of the
Resurrection; Prob (R/K) = the initial probability of the resurrection
hypothesis; Prob (EH/K&R) = the explanatory power of the resurrection
hypothesis; Prob (AT/K) = the initial probability of the alternative hypotheses;
and Prob (EH/K&AT) = the explanatory power of any alternative hypothesis).
Now Bayes's Theorem implies that the resurrection hypothesis will have a high
epistemic probability (i.e., Prob (R/K&EH) > 0.5) just in case it has a
greater balance of inherent probability and explanatory power than any of its
alternatives (i.e., Prob (R/K) 3 Prob (EH/K&R) > Prob (AT/K) 3 Prob
And this is in fact the case. In my opinion, the alternative theories that
have been proposed are not only weaker but far weaker at explaining the
available historical evidence than the claim that God raised Jesus from the
dead. That is, there is a patch of first-century history that makes sense from a
Christian perspective but not from a naturalist's perspective. Those who push
the alternative theories normally do so, in my opinion, because their view of
the background knowledge rules out interventionist miracles or renders them so
highly improbable as to be not worth taking seriously. That is, most such folk
are metaphysical naturalists or Deists.
Moreover, even if it is true that supernaturalists are suspicious of most
miracle claims, that does not entail that supernaturalists hold that the
probability of the Resurrection is low. What is being ignored here is the fact
that the Resurrection hypothesis involves the free choice of an agent, viz.,
God. This is why the rarity of Resurrections (which everyone will grant) cannot
be equated with improbability. Suppose I want to buy a car, and I enter a lot
where there are a thousand cars for sale, of which only one is red. Now what is
the probability that I buy the red one? Clearly, that probability is not just a
function of the infrequency of red cars in the sample. This is obviously because
my selection of a car might not be entirely random as to color. Indeed, I might
freely choose to buy the red car precisely because of its uniqueness. Getting
back to the resurrection, this means that if God wanted to vindicate Jesus, God
may have chosen to raise him from the dead precisely because resurrections are
so rare and striking. Thus, the very infrequency of resurrections may actually
increase the probability of the resurrection of Jesus.
5. Martin's use of Bayes's Theorem. I have said little thus far about the way
Martin uses Bayes's Theorem, which by and large is beyond reproach, at least
until he starts supplying actual values. But there are problems in the crucial
footnote 19. They look like typos, but when corrected, have substantive
significance, and support me rather than Martin. Martin
wants the result that Prob (R/K&EH) = .36. Now we arrive at that result only
if we read the values of Prob (R/K) as 0.1, Prob (EH/~R&K) as 0.2, and prob
(~R/K) as 0.9. (That is, we need to shift the place of the decimal point. Note
also that the T of note 14 has suddenly become the K of note 19.) But now it
appears that the figures confirm my claim that it is necessary for a naturalist
to supply a likely alternative to the resurrection of Jesus in order rationally
to reject it. For Martin's figures indicate that, even if we accept his (to my
mind highly implausible) ascription of a value to Prob (~R/K) of nine times as
much as the value we ascribe to Prob (R/K), we would need to have a value of
something like 0.11 for Prob (EH/~R&K) for the value of Prob of (R/EH&K)
to fall below .5. That is, the disjunction of alternatives to R as explanations
of EH would have to give a probability of at least 0.11 to EH for it to be more
probable than not that some alternative to the resurrection of Jesus is the true
account. I find that extremely implausible.
In general, Bayes's Theorem is a useful tool in some epistemic situations,
but it is a blunt instrument when used in discussions of the resurrection of
Jesus. One problem is the assumption, made by Martin and others, that you can
read probabilities from frequencies ("Even Christians admit that miracles
occur infrequently; therefore miracles are improbable."). But as John
Earman has pointed out, the attempt to do that is almost universally recognized
in the philosophy of science as unsuccessful. An observed
frequency may be flatly zero (cf. an event of proton decay, never before
observed, but which scientists are spending huge amounts of money and effort to
detect), but it would be simple-minded accordingly to set the probability at
Another problem is the fact that, in actual cases of purported religious
miracles, there is nowhere near enough undisputed data to give reliable
estimates of the relevant frequencies. So it becomes necessary for critics who
want to use Bayes's theorem to argue by analogy to cases of testimony to events
where we know the relevant frequencies. But, as Earman points out, "there
are no accepted rules for such analogical reasoning,"
and people are obviously going to differ in the values they attach to the priors
and likelihoods. There seems to be no objective way of adjudicating such
I conclude that Martin has not succeeded in achieving his aim. For theists,
and especially Christian theists, the evidential burden that must be overcome
before one can rationally believe in the resurrection is not nearly so heavy as
Martin argues. Even if the Prob (R/K) is low, there is no reason to think that
this cannot be overcome by the fact that Prob (R/K&EH) > Prob
(not-R/K&EH). Indeed, as I have argued, the probability is not only greater
but very much greater. Belief in the resurrection of Jesus continues to be
rational for Christians.
1. See Michael Martin, "Why the Resurrection is
Initially Improbable," Philo, 1, no. 1 (Spring-Summer 1998): 63-73.
2. Martin, p. 68.
3. In both his Summae, Aquinas speculated about the
possibility of the Incarnation and resurrection occurring earlier or later than
they in fact did. But that in no way inclined him to doubt their occurrence, nor
should it have done.
4. Martin, p. 67.
5. The quotation is found in Stephen T. Davis, "Is It
Possible To Know that Jesus Was Raised from the Dead?," Faith and
Philosophy, 1, no. 2 (April 1984): 153.
6. Stephen T. Davis, Risen Indeed: Making Sense of the
Resurrection (Grand Rapids, Mich.: Wm. B. Eerdmans, 1993).
7. I do not know how many theologians and scripture scholars
have written about the empty tomb in, say, the last thirty years. But William L.
Craig, who has exhaustively studied the literature, points out that a strong
majority support the historicity of the empty tomb tradition, and he lists
twenty-eight scholars who do. See William L. Craig, Assessing the New Testament
Evidence for the Historicity of the Resurrection of Jesus (Lewiston, N.Y.: The
Edwin Mellen Press, 1989), p. 373.
8. See Richard Swinburne, Revelation (Oxford: Clarendon
Press, 1992), pp. 71-72; Responsibility and Atonement (Oxford: Clarendon Press,
1989), pp. 148-62; and The Christian God (Oxford: The Clarendon Press, 1994),
9. See Alan F. Segal, "Life After Death: the Social
Sources," in The Resurrection: An Interdisciplinary Symposium on the
Resurrection of Jesus, Stephen T. Davis, Daniel Kendall, S.J., and Gerald
O'Collins, S.J. (Oxford: Oxford University Press, 1997), pp. 90-125.
10. Martin, p. 64.
11. Martin, p. 66.
12. Since Martin is prepared to concede "for the sake
of argument" (p. 67) that interventionist miracles are probable given God's
existence, I will not respond in detail to his arguments (see pp. 65-66) to the
effect that there are compelling reasons for God to eschew them. I will only say
that God is going to have to bring about an interventionist miracle if (a) it is
a natural law that people who die stay dead, and (b) God wants to raise some
particular person from the dead.
13. Martin, p. 71.
14. As I argued in Risen Indeed, no matter what the state
of the evidence, a naturalist can always say, "Yes, the evidence for the
resurrection of Jesus is strong; I can't produce a good alternative explanation
of what happened, but a resurrection just couldn't have happened" (pp.
20-21). This is, on my view, a rational (albeit incorrect) position.
15. I borrow this argument from Richard Swinburne. See his
"Evidence for the Resurrection," in The Resurrection: An
Interdisciplinary Symposium on the Resurrection of Jesus, Stephen T. Davis,
Daniel Kendall, S.J., and Gerald O'Collins, S.J. (Oxford: Oxford University
Press, 1997), p. 200.
16. See S. L. Zabell, "The Probabalistic Analysis of
Testimony," Journal of Statistical Planning and Inference, 20 (1988):
17. In fairness I should point out that in October 1998
Martin notified me of his awareness of the error in footnote 19 of his paper.
This was well after I had noticed the error, and indeed had drafted the present
paragraph. I am leaving that paragraph as it was only because, as noted, I
believe the error can be taken to have substantive implications. Editor's note:
Footnote 19 should read Prob(R/K)=.1, Prob(EH/~R&K)=.2, and Prob(~R/K)=.9
and not Prob(R/K)=.01, Prob(EH/~R&K)=.02, and Prob(~R/K)=.09. Also, footnote
14 should read:
Prob(R/EH&T) = Prob(R/T) 3 Prob(EH/R&T)
[Prob(R/T) 3 Prob(EH/R&T] + [Prob(~R/T) 3 Prob(EH/~R&T]
Prob(R/EH&T) = Prob(R/T) 3 Prob(EH/R&T)
[Prob(R/T) 3 Prob(EH/R&T] + [Prob(~R/T) 3 Prob(EH/~R&T]
18. See John Earman's "Bayes, Hume, and
Miracles," Faith and Philosophy, 10, no. 3 (July, 1993): 293-310.
19. Earman, p. 301.
20. I thank Kelly James Clark, William L. Craig, Gerald
O'Collins, Alan Padgett, Victor Reppert, and Richard Swinburne for their helpful
comments on an earlier draft of this essay.
Stephen T. Davis is Professor of Philosophy and Religious Studies at
Claremont McKenna College.