How to Weigh a Testimony¶

The problem¶

A miracle claim is almost always a claim about testimony. Someone says that something happened: a tumour vanished, a lame man walked, a word of knowledge named a stranger's secret, a dead child breathed. We were not there. We have only the report.

That is not a reason to dismiss it. Most of what we know arrives by testimony: that Caesar crossed the Rubicon, that the Hindenburg burned, that your grandfather was kind. History is testimony all the way down. The question is not whether a claim is testimony but whether the testimony is good.

This page sets out the tests this casebook uses. They are the ordinary tests of historical and legal reasoning, not special pleading for religion.

The tests¶

1. Proximity¶

How close is the witness to the event? A doctor who examined the patient before and after is closer than a preacher who heard about it a year later. A contemporary newspaper is closer than a centennial retrospective. Proximity is not everything — a distant witness can be honest and a close one can lie — but it is the first filter.

2. Independence¶

Do the accounts come from separate sources, or from one source repeated? Ten people who all heard the story from the same evangelist are one witness, not ten. Independent attestation — a patient, a physician, a sceptical journalist, a hospital record — is much stronger.

3. The hostile witness¶

The strongest testimony comes from someone with no interest in the conclusion. A secular physician on a church's medical committee, an anticlerical newspaper reporting a crowd's experience, a government inquiry that would rather find nothing — these are worth more than a hundred friendly anecdotes. This is why the source library weights the peer-reviewed studies and the hostile press so heavily.

4. Documentation made at the time¶

Was anything written down or recorded when it happened? A notarised deposition, a hospital chart, a radiograph, a newspaper printed the next day, a letter posted that week. Documents created later, from memory, are weaker — not worthless, but weaker.

5. Publicity and testability¶

Could other people examine the claim? A healing in a public square, with the patient known to the town, is more testable than a private impression. A prophecy delivered publicly and written down before the event can be checked; one delivered in a closed room and remembered afterwards cannot.

6. Specificity¶

"The Lord healed me" is a testimony. "The consultant at [hospital] diagnosed a slipped disc at L4/L5 on 3 March; the MRI on 12 May showed no disc herniation; he signed me back to work on 20 May" is evidence. Specificity is not proof — specific claims can be false — but vagueness cannot be checked at all.

7. Alternative explanations¶

Has anyone considered the ordinary ones? Misdiagnosis, spontaneous remission, the placebo effect, psychosomatic symptoms, the body's own repair, fraud, exaggeration, the redefinition of terms. A claim that has survived a serious attempt to explain it naturally is stronger than one that has never been questioned.

8. The witness's stake and character¶

Does the witness gain from the claim? Money, status, a platform? Are they known for honesty? This is not a licence to dismiss the poor and obscure; it is a reason to weigh the famous and the funded more carefully.

9. Falsifiability¶

Did the claim put itself at risk? A prophecy that says "something good will happen to someone here" cannot fail. A prophecy that names a person, a date, and an event can. A healing claim that submits to medical review can be falsified; one that refuses review has chosen not to be.

10. Persistence under investigation¶

What happened when someone looked? A peer-reviewed medical journal published the Mozambique study, and its authors set out the design's weaknesses in the paper itself. That is a point in its favour. A ministry that will not submit to any examination is telling you something about itself.

Hume, and the prior-probability problem¶

The classic philosophical objection is David Hume's: "No testimony is sufficient to establish a miracle, unless the testimony be of such a kind, that its falsehood would be more miraculous than the fact which it endeavours to establish" (An Enquiry Concerning Human Understanding, section 10).

The argument has force as a warning about credulity. It has less force as a universal veto, for three reasons.

First, it assumes what it needs to prove. Hume's argument depends on the uniformity of natural law, but the question at issue is whether the laws of nature are the whole story. If God exists and acts, a miracle is not a "violation" but an act of the personal agent who sustains nature.

Second, the prior probability is not zero. For a naturalist, the prior is vanishingly small, and no testimony will move it. For a theist, the prior is not small at all. The disagreement is not about the evidence; it is about the prior. That is why the casebook puts the exegetical argument first and the evidence second: the evidence is weighed inside a worldview, and the worldview is what Scripture addresses.

Third, Hume's own criterion is empirical, not a priori. He allows that testimony could establish a miracle if the falsehood would be more miraculous than the event. That is a high bar, not an impossible one. A medically documented cure of a condition that never regresses, attested by sceptical physicians, is exactly the kind of case the criterion was designed to consider.

The Bayesian literature has made this precise: the question is not "is a miracle probable?" in the abstract, but "is this evidence more likely given a miracle than given the alternatives?" (See John Earman, Hume's Abject Failure, 2002.)

What this means for this casebook¶

Three rules follow, and they govern the rest of the site.

First, no single case is load-bearing. The argument does not depend on any one healing, any one revival, or any one saint. It depends on the pattern: the exegetical case from Scripture, the persistent testimony of the church, and a body of documented cases that are hard to explain.

Second, grade the evidence and say so. The source library labels every entry by type and deniability. The Mozambique study distinguishes what was measured from what it means. The Azusa page distinguishes press report from medical record. Honesty about the grades is part of the argument.

Third, expect mixture. Some claims are true, some are false, some are exaggerated, some are unexplained. A mature continuationist is not required to defend the whole heap. He is required to stop treating "we have reports" as proof and "we have doubts" as disproof.

The alternative to proof¶

The sceptic often says, "You cannot prove a miracle." That is true. You cannot prove most particular historical events; you can only make them more or less probable. The question is what best explains the evidence: a world in which God sometimes heals, or a world in which every report across twenty centuries is mistaken, fraudulent, or misdiagnosed.

The casebook does not claim the first explanation is proven. It claims it is the better explanation, and that the burden of proof lies with the view that says God has stopped.

What this does and does not prove¶

It proves: that testimony can be graded; that the tests are ordinary and public; and that the best cases are not the loudest ones.

It does not prove: that any particular claim is true, that Hume was wrong about everything, or that evidence can create faith. It proves that the evidence deserves to be weighed rather than waved away.