Petitionary prayer produces no result beyond what chance and ordinary causes already predict, which is what we would expect if no one is listening

Petitionary Prayer and Measurable Results

The argument to be examined is not the crude claim that every unanswered prayer disproves God. Its stronger form is evidential and comparative. If a personal God hears and answers petitionary prayer, then the total pattern of outcomes should look at least somewhat different from what we would expect under chance, ordinary causation, human effort, and reporting bias. Yet, the critic says, the pattern we actually observe is just what we would expect if no divine responder were involved. Prayers for recovery, employment, safety, fertility, or relief sometimes appear to "work", but the successful cases cluster in precisely the places where coincidence and normal mechanisms already operate. A person prays before a job interview and gets the job. A patient prays while receiving modern medical treatment and recovers. A family prays during financial difficulty and receives help from relatives or a charity. Meanwhile, children praying to be spared abuse, starvation, bombing, or terminal illness often receive no rescue at all.

In its most forceful version, this is not merely an emotional complaint. It is a demand for proportional evidence. If believers cite answered prayer as evidence for a personal God, then the critic is entitled to ask whether the alleged answers occur at a rate or in a pattern that exceeds what we should already expect. Francis Galton made an early statistical version of this point in his nineteenth-century essay "Statistical Inquiries into the Efficacy of Prayer", asking whether groups publicly and regularly prayed for, such as sovereigns or clergy, showed the expected advantages. In contemporary terms, the point can be put in the language of comparative confirmation: an observation supports one hypothesis over another only insofar as it is more expected on that hypothesis. The same broad evidential discipline appears in Paul Draper's work on suffering, J.L. Schellenberg's work on divine hiddenness, and Daniel Dennett's broader criticism of insulating religious explanations from ordinary standards of inquiry. The point is not that these writers all make the identical prayer argument, but that they press a shared demand: claims that touch observable events should be compared with rival explanations.

The strongest popular exhibit is medical intercessory prayer. The most famous large controlled trial is the Study of the Therapeutic Effects of Intercessory Prayer, usually called the STEP trial, published by Herbert Benson and colleagues in the American Heart Journal in 2006. It involved patients undergoing coronary artery bypass graft surgery and used three groups: patients told they might or might not be prayed for who were in fact assigned prayer by the study; patients told the same thing who were not assigned prayer by the study; and patients told with certainty that they would be prayed for. The trial did not find a beneficial effect of the assigned intercessory prayer on recovery. The group that knew it was being prayed for had a higher rate of complications, but that result should be interpreted cautiously; possible explanations include anxiety, expectation effects, or chance, not a demonstrated harmful power of prayer.

There have been smaller studies and later reviews, and the serious literature is more cautious than online summaries often are. Some trials have methodological weaknesses, including small samples, unclear blinding, multiple measured outcomes, and the difficulty of controlling outside prayer. Some report positive effects, others do not. Systematic reviewers have generally not treated the evidence as establishing reliable medical efficacy. The sober conclusion is not that science has "proved prayer false", but that controlled studies have not provided a robust, repeatable signal showing that intercessory prayer improves health outcomes beyond ordinary causes.

The Steelmanned Argument

The argument can be stated as follows:

• 1. If petitionary prayer is answered by a personal God often enough, and in ways specific enough, to be offered as public evidence, then, other things being equal, we should expect some detectable divergence between prayed-for and not-prayed-for outcomes, especially where grave need is involved and ordinary explanations are weak.
• 2. We do not observe such a divergence. Apparent successes are readily explained by chance, selective memory, confirmation bias, medical care, social support, persistence, and ordinary human action.
• 3. The distribution of alleged answers is morally and evidentially puzzling. Trivial or ordinary successes are often credited to God, while urgent pleas from the most vulnerable frequently appear unanswered.
• 4. Controlled tests of intercessory prayer have not produced strong evidence of benefit, and the best-known large trial found none.
• 5. Therefore, reported answers to petitionary prayer supply little public evidence for a divine responder; and the observed pattern is more expected on naturalism than on any theistic model that predicts frequent, request-sensitive interventions visible in worldly outcomes.

This argument has real force. It targets a common apologetic move: "I prayed, the desired event happened, therefore God answered." The critic rightly notes that such reasoning is weak unless the event is unlikely under ordinary causes, clearly connected to the prayer, and not balanced by a much larger class of unreported failures. Human beings are poor intuitive statisticians. We remember hits and forget misses. We retell striking coincidences and remain silent about ordinary disappointments. We are especially prone to interpret ambiguous outcomes in ways that protect prior commitments.

The argument also presses a moral asymmetry. If one person treats finding car keys as evidence that God intervened, while another person's prayer to survive torture, famine, or a child's terminal illness is not granted, the selection of "answered" cases can look arbitrary. The point is not to mock gratitude for ordinary goods; gratitude may be morally valuable. The evidential problem is that small conveniences are counted as signs, while vastly more urgent non-rescues are often excluded from the ledger. If the claimed pattern of divine response cannot distinguish between trivial conveniences and desperate necessities, then it is hard to use that pattern as evidence of wise and compassionate intervention.

The Strongest Reply

The strongest theistic reply begins by denying that petitionary prayer is a mechanism whose outputs should be predictable in the way a drug's effects are predictable. On classical theism, God is not one cause among others inside the natural order, and prayer is not a technique for manipulating divine behavior. Prayer is instead understood as relational, morally formative, and expressive of dependence; a believer may receive the requested outcome, a refusal, endurance, changed desires, insight, or help through other people. On that understanding, the absence of a measurable effect in a clinical trial is not surprising. But this reply also narrows the evidential claim: prayer so understood should not be advertised as though it were a publicly measurable causal treatment.

A defender may also argue that the expected evidence is not straightforward. If God has reasons to preserve a stable natural order, to avoid coercive public manifestations, to respect human freedom, or to achieve goods not visible to us, then many sincere prayers may not be granted in the requested form. This reply is familiar from broader discussions of evil and divine hiddenness, especially skeptical-theist responses. It does not say suffering is unreal or easy to explain. It says that finite observers may not be well placed to infer from "I see no sufficient reason" to "there is no sufficient reason." The reply is philosophically serious, but also controversial, because if used without limits it can weaken ordinary moral and evidential reasoning.

There are also specific objections to prayer trials. First, "contamination" is nearly unavoidable. Patients in a control group may be prayed for by family, friends, or religious communities outside the trial. Second, the trial cannot measure the sincerity, moral state, intentions, or theological context of those praying. Third, if God is personal, divine action may not be triggered by randomization protocols. Fourth, some theists hold that deliberately testing prayer as though God were a laboratory variable misunderstands the practice. These points do not prove that prayer works, but they do weaken overconfident claims that such trials straightforwardly falsify petitionary prayer.

The reply has genuine strength. It is fair to say that many sophisticated theists do not predict that prayed-for patients in a randomized trial must recover more often. They also do not hold that every sincere request, even every desperate one, will be granted. So the empirical argument does not refute theism as such. It refutes only a cruder model in which God functions as a reliable request-granting agency whose interventions should be statistically obvious.

Where the Reply Is Less Satisfying

Still, the reply has a cost. The more prayer is protected from measurable expectations, the less it can be used as public evidence. A believer cannot consistently say, on Monday, that answered prayer is a visible sign of divine reality, and then say, on Tuesday, that no observable pattern should be expected when the claim is tested. If alleged answers count in favor of theism, then unanswered prayers and statistical indistinguishability must count for something in the other direction, even if not decisively.

This is an internal critique of evidential standards. The issue is not whether a theist may rationally pray. The issue is whether reported answers to prayer provide evidence strong enough to persuade someone who does not already accept the framework. Anecdotes rarely meet that burden. For every story of a remission, there are many non-remissions. For every providentially timed donation, there are many evictions. Unless the full comparison class is considered, the evidence is distorted.

The appeal to hidden reasons also needs discipline. It may be true that a personal God would have reasons beyond our knowledge. But if that move is made too quickly, it can make the claim immune to evidence. Any outcome becomes compatible with answered prayer: recovery, death, improvement, deterioration, rescue, non-rescue. Compatibility is not confirmation. A hypothesis that can accommodate every possible result may still be true, but the observed result gives outsiders little reason to accept it.

The suffering of children is the deepest part of the argument. It is not merely a statistical inconvenience. It connects petitionary prayer to the problem of evil, which is among the strongest arguments against theism. Critics such as William Rowe and Paul Draper argue, in different ways, that apparently pointless suffering is much more expected on naturalism than on theism. Theists have replied in different ways: Alvin Plantinga's free-will defense is aimed mainly at the logical problem of evil; Richard Swinburne emphasizes soul-making, freedom, and the goods made possible by stable natural laws; Eleonore Stump and Marilyn McCord Adams address suffering through more personal and narrative accounts of redemption and defeat of evil. These are serious philosophical positions, not evasions by definition. But they remain emotionally and intellectually costly. They do not turn the unanswered prayers of the desperate into evidence for theism. At best, they try to prevent such cases from being decisive evidence against it.

What the Argument Establishes

The argument establishes several limited but important conclusions.

• First, petitionary prayer, considered through public and statistical evidence, has not been shown to produce reliable effects beyond chance and ordinary causes.
• Second, many popular claims about answered prayer are evidentially weak because they rely on anecdotes, selective reporting, and failure to compare successes with failures.
• Third, controlled prayer studies do not support confident claims that intercessory prayer has measurable medical efficacy.
• Fourth, the uneven pattern of alleged answers, especially when trivial requests appear granted and extreme suffering remains, strengthens the evidential problem of evil and divine hiddenness.

But the argument does not establish that no God exists. It does not even strictly establish that no prayers are ever answered. A personal God, if real, might act rarely, for reasons not captured by statistical testing, or in ways not identifiable as public evidence. The argument also does not show that prayer lacks all psychological, communal, or moral value. Even a naturalist can acknowledge that prayer may calm, focus, console, or motivate some people, though those effects would not require a divine responder.

The fair conclusion is therefore modest but significant. Petitionary prayer is weak as public evidence for theism. Its reported successes have not, in general, been shown to exceed what chance, ordinary causation, and human interpretation already explain. The strongest theistic replies prevent this from being a simple disproof of God, but they also reduce the evidential value of "answered prayer" claims. If prayer is defended as a private relational practice, it may survive the critique. If it is offered as measurable proof that someone is listening, the evidence does not carry the weight placed upon it.