A Bayesian Case for Jesus as God
Given that God (of some form) exists, how probable is it that Jesus was God incarnate and rose from the dead?
Why it matters
Probability-based reasoning strengthens the historical case: instead of treating miracle claims as automatically improbable, we ask what God (if any) would be likely to do and what kind of public evidence we would expect to see.
The main case
's two-step argument: (1) If there is a God with the attributes classical theism ascribes to him, there is significant prior probability that he would become incarnate to reveal himself, identify with human suffering, and provide atonement. (2) The historical evidence for Jesus of Nazareth (his life, teaching, crucifixion, and purported resurrection) is exactly the pattern we would expect if such an incarnation occurred. When you plug both steps into Bayes' theorem, concludes P(Jesus is God incarnate | total evidence) is high. The focus narrowly on the resurrection evidence and argue the Bayes factor is astronomical.
Argument map
If God exists, God has reasons to become incarnate and provide public identifying marks.
Jesus uniquely fits those expected marks: perfect life, profound moral teaching, claims to divine authority, and miracles.
The resurrection functions as God's public "endorsement" of Jesus' identity.
The Bayes factor for the resurrection evidence (creeds, witnesses, martyrdoms, empty tomb) is very large.
Given theism and the historical evidence, it is highly probable that Jesus was God incarnate.
You cannot assign meaningful priors to divine actions.
Swinburne argues theism plus general reflection on God's goodness yields non-trivial expectations. We use analogous probabilistic reasoning in legal and historical inference.
The evidence is religiously biased and cannot be trusted.
The core facts (1 Cor 15 creed, crucifixion, James, Paul, empty tomb) are accepted by a strong majority of critical historians, not just Christians.
Any miracle claim has a prior so low it swamps the evidence.
This begs the question against theism. If theism has non-trivial prior probability, miracles as identifying acts of God do not start at near-zero.
Bayesian resurrection calculator
Adjust the dials yourself. Your prior is how probable you considered Jesus\' resurrection before looking at the evidence. For each piece of evidence, estimate how likely it would be if the resurrection happened and how likely it would be if it did not. The calculator applies Bayes\' theorem sequentially and shows how your belief should update. Follow the work of Timothy & Lydia McGrew (Blackwell Companion to Natural Theology).
Tip: even a prior of 1 in 10,000 yields a posterior > 99% when the combined Bayes factor exceeds roughly 106.
- Forensic DNA identification evidence in court typically runs at about 1 in 109 to 1012. A cumulative factor of 1040 is about a trillion trillion times stronger than the DNA match that convicts a defendant beyond reasonable doubt.
- Particle physicists declare a "discovery" at 5-sigma, about 1 in 3.5 million. 1040 is equivalent to roughly a 13-sigma result — the level at which physicists stop calling it evidence and start calling it the new baseline.
- It is the Bayesian weight of guessing a specific atom in a kilogram of matter, blindfolded, on the first try — and the evidence here points that strongly toward one hypothesis.
- Suppose you think resurrection is so unlikely the prior is 1 in a billion (10-9). A Bayes factor of 1040 still leaves you with posterior odds of 1031 to 1 in favor.
- To defeat this evidence with priors alone you would need to be more certain miracles are impossible than any human being is ever justified in being about anything — more certain than you are that the sun will rise, that other minds exist, or that you are not dreaming.
- Hume's "no testimony is sufficient" argument is often quoted here — but Hume was arguing before Bayes' theorem was widely understood. Applied rigorously, his standard quietly demands a prior so extreme it is unreachable.
If you would convict a stranger of murder on a 1012 DNA match, consistency requires you to take a 1040 cumulative case seriously. Either the ordinary standards of evidence apply here too — or you are making an exception precisely because of what the evidence points to.
Put the Bayesian calculator on your site
Let your readers adjust the same dials themselves. The embed is a standalone, responsive iframe.
<iframe
src="https://christwise.org/embed/bayesian"
width="100%"
height="1400"
style="border:0;border-radius:12px;max-width:100%"
loading="lazy"
title="Put the Bayesian calculator on your site"
allow="clipboard-write"
></iframe>Claim · Evidence · Objection · Response
1.Given theism, incarnation has non-trivial prior probability.
DebatedEvidence
- A perfectly good God has reasons to identify with human suffering and demonstrate love.
- Public incarnation provides a uniquely reliable way to communicate God's character.
- No major competing religion offers the same combination of incarnation + atonement + resurrection as divine vindication.
Strongest objection
"A priori reasoning about what God "would do" is speculative."
Response
agrees it is not deductive, but argues it is not zero either. Even modest prior probabilities combine with high likelihood ratios to yield strong posteriors.
- The Resurrection of God Incarnate — Richard Swinburne (2003)scholarlyFind on Amazon
- Was Jesus God? — Richard Swinburne (2008)popularFind on Amazon
2.The resurrection evidence yields a very high Bayes factor.
DebatedEvidence
- P(creed of 1 Cor 15 arising within 5 years | resurrection) is much higher than P(creed | no resurrection).
- P(empty tomb first reported by women | resurrection) is higher than P(empty tomb | invented legend).
- P(Paul's conversion | resurrection) is much higher than P(Paul's conversion | no resurrection).
- The independence of these data multiplies the overall Bayes factor.
Strongest objection
"Naturalistic explanations (hallucination, legend, theft) can be combined to account for each datum."
Response
Each naturalistic explanation lowers one Bayes factor but usually at the cost of another (e.g., hallucination explains appearances but not the empty tomb). The argue no disjunction of plausible naturalistic hypotheses matches the resurrection's explanatory scope.
- "The Argument from Miracles: A Cumulative Case for the Resurrection of Jesus of Nazareth" — Timothy & Lydia McGrew (2009)scholarlyFind on Amazon
- The Resurrection of God Incarnate — Richard Swinburne (2003)scholarlyFind on Amazon
- The Case for the Resurrection of Jesus — Gary Habermas & Michael Licona (2004)scholarlyFind on Amazon
What scholars debate
and represent the most developed Bayesian treatments of the case for Christ. Skeptics like Graham Oppy and John Earman have engaged them directly. Even critics often concede the framework is sound; they contest the numerical inputs, especially the likelihoods of the resurrection data on naturalistic hypotheses.
Reflection
- 1.If you assume God exists, does that change how probable a miracle becomes?
- 2.Which of the resurrection data points would most shift your own probability?
- 3.What evidence would count against the Bayesian case?
Key sources
- The Resurrection of God Incarnate — Richard Swinburne (2003)scholarlyFind on Amazon
- Was Jesus God? — Richard Swinburne (2008)popularFind on Amazon
- "The Argument from Miracles: A Cumulative Case for the Resurrection of Jesus of Nazareth" — Timothy & Lydia McGrew (2009)scholarlyFind on Amazon
- 1 Corinthians 15:3-8 (early creed) — Paul of Tarsus (c. AD 53-55)scripture
Featured thinkers
A leading philosopher of religion who has applied Bayesian probability theory to arguments for God's existence, the incarnation, and the resurrection.
Leading resurrection scholar who developed the Minimal Facts approach, cataloging claims accepted by a broad majority of critical historians.
Historian specializing in the resurrection, ancient biography, and Greco-Roman historiography.
Pass it on
Share this lesson
One honest argument can change a conversation. Send it to a friend, a skeptic, or your small group.
Want a hi-res image card instead? Build a shareable evidence card sized for X or Instagram Stories.