Experiments on Distant Intercessory Prayer, God, Science, and the Lesson of Massah

John T. Chibnall, PhD; Joseph M. Jeral, MD; Michael A. Cerullo, MD

ARCHIVES OF INTERNAL MEDICINE

November 26, 2001

Vol 161, No 21, pp 2517-2640

For if you should enter into the temple for no other purpose than asking, you shall not receive. And if you should enter it to humble yourself, you shall not be lifted . . . it is enough that you enter the temple invisible. . . . God listens not to your words save when He Himself utters them through your lips.Kahlil Gibran, The Prophet

EXPERIMENTAL studies on the health effects of distant intercession (prayer) ignore important facets of construct validity, philosophy of science, and theology while focusing on issues like randomization and double-blinding. These tendencies reflect a desire on the part of researchers to remove nature as a causal factor when intercession seems efficacious. We argue that close attention to construct validity of cause-and-effect variables invalidates distant intercessory prayer as a scientific construct. Further, the application of statistical techniques to metaphysical causal phenomena is critiqued. We conclude that research on the effects of religion and spirituality on health should avoid attempting to validate God through scientific methods.

INTRODUCTION

There exist only a handful of recent experimental studies designed to test the efficacy of distant intercessory prayer for health problems.^1-7 Wishing to help fill the gaps in this deficient literature, wea "representative" cross section of the 3 major Western Judeo-Christian traditions (Catholicism, Judaism, and Protestantism)attempted to design an experimental distant prayer intervention study for major depression. This exercise led to disenchantment with the endeavor rather than a study design. Beginning with questions about research validity, we progressed into arguments about the nature of reality and the philosophy of science. Ultimately, we became convinced that the very idea of testing distant prayer scientifically was fundamentally unsound. The remainder of this article discusses each of these steps in some detail.

ISSUES OF CONSTRUCT VALIDITY

The feeling that all was not right with the intercessory prayer research emerged early and centered around our attempts to address the construct validity of our proposed study. In their seminal work on research validity, Cook and Campbell⁸ defined construct validity (with respect to causes) as the extent to which the operations meant to represent the causal factor (ie, the treatment or independent variable) actually reflect some theoretical construct of interest. When operations fail to incorporate critical dimensions of the theoretical construct, "construct underrepresentation" results (ie, low construct validity).

So, we began by attempting to avoid what Cook and Campbell called "inadequate preoperational explication of constructs."⁸ The way to do this is to choose operations depending on the result of a rigorous theoretical and conceptual analysis of the essential features of the causal construct. But what were the critical dimensions of the construct prayer? Prayer, by definition, is "an address or petition to God or a god in word or thought."⁹ But what else matters?

The list we generated began as follows: Is the amount of prayer important? Is the type of prayer important? The form? The duration? The frequency? The level of fervency? The entity to whom it is directed? The number of prayers per unit of time? Does the number of intercessors matter? Does a team vs individual intercession method matter? Does the faith tradition of the intercessor and/or intercessee matter? Does the power of the intercessor matter? Do the beliefs and experiences of the intercessor and/or intercessee matter? Does the worthiness of the intercessor and/or intercessee matter?

This list, unfortunately, generated questions of its own: If type or form is important, just how many types or forms are there? On what basis would you distinguish them? If "fervency" is important, how would you ever measure it to be able to manipulate it? The same is true if the power or worthiness of the intercessor is important: how would you ever measure them? An examination of the distant prayer studies did not help and only reinforced our growing suspicions. Prayer was not explicated well in these studies. Typical studies used constellations of intercessors from various faith traditions praying in decidedly individual ways.^2-7 For example, the intercessors (Protestants, Catholics, and Jews) in the study by Walker et al⁶ were asked to make "general positive intentions as opposed to specific requests" but "were free to pray in whatever way they chose." In an attempt at standardization, the intercessors in the study by O'Laoire⁴ were instructed in a 1-hour training session about "theories of intercessory prayer"; the importance of relaxation, positivity, focus, imagery, and "confident but effortless intentionality" when praying; and the difference between directed and nondirected prayer (directed prayer ["very specific as to outcome"] vs nondirected prayer [seeks "alignment with divine will"]). However, the intercessors were not described with respect to who they prayed to; what tradition guided their own prayer; what they actually did, said, and felt on the days that they prayed; or how long they prayed. The intercessors in the study by Harris et al³ were required to believe in a personal God who is "responsive to prayers for healing made on behalf of the sick."³ They then prayed daily for a "'speedy recovery with no complications' and anything else that seemed appropriate to them." The healers in the study by Sicher et al⁵ all had "previous healing experience at a distance" and included Christians, Jews, Buddhists, Native Americans, shamans, and graduates of "schools of bioenergetic and meditative healing." They were instructed to "direct an intention of health and healing to the subject." Matthews et al⁷ reported that 2 Christian prayer ministers (from Christian Healing Ministries, Jacksonville, Fla) offered"for the health of the patient""at least 10 minutes of daily prayer" individually and shared prayer (ie, both ministers prayed together in person or by telephone) at least weekly.⁷ The intercessors in the study by Byrd² were "'born again' Christians (according to the Gospel of John 3:3)," including Protestants and Catholics who prayed "daily for a rapid recovery and for prevention of complications and death, in addition to other areas of prayer they believed to be beneficial to the patient."² While all studies seemed to keep track of the volume of prayer offered over the course of the intervention, they only manipulated (crudely) the occurrence or nonoccurrence of distant prayer. Why?

This state of affairs did not seem to be accidental. Perhaps prayer was inadequately explicated because it could not be adequately explicated. We found ourselves wondering what theoretical conceptualization of prayer could ever answer all of these construct validity questions. More important, the following question arose: If prayer is a metaphysical concept linked to a supernatural being or force, why would its efficacy vary according to parameters such as frequency, duration, type, or form? The very concept of prayer exists only in the context of human intercourse with the transcendent, not in nature. The epistemology that governs prayer (and all matters of faith) is separate from that which governs nature. Why, then, attempt to explicate it as if it were a controllable, natural phenomenon?

Consider an analogy. A clinical trial is planned to study the effect of antibiotic agents on a new strain of respiratory bacteria. The investigators go to great lengths to ensure random assignment, placebo control, and double-blinding (as did our distant intercessory prayer researchers). But this is not all they go to great lengths to ensure. They control the type of antibiotic, the dose of the antibiotic, the dosing schedule of the antibiotic, the coadministration of other medications that have antibiotic effects or interfere with antibiotic effects, and the duration the antibiotic is taken, to name a few of the most obvious parameters. Why? Because all of these factors are critical to the construct validity of the cause; that is, they are part of the definition of "antibiotic." If the researchers ignored these characteristics of the causal agent or let them vary capriciously, the results of the study, no matter what the outcome, would be greatly reduced in interpretability. But this is the very state of affairs inherent in distant prayer research. The explanation for this state is that there is no reasonable theoretical construct to which to link prayer because of, we would argue, its very nature. No model guides our understanding of intercessory prayer as a treatment in the way we know that drug pharmacokinetics, type, dose, schedule, interactions, and treatment length are critical to an antibiotic as a treatment. In fact, we believe no scientific model can guide it.

A CONSTRUCT RESISTANT TO EXPLICATION

The problem becomes more fundamental than a simple lack of construct validity. It is deeper than the question, "What are the essential features of prayer?" The problem is in the construct itself. The more important query is the follow-up question, "How can we know?" Even if you could identify what you considered to be an essential feature of prayer, on what scientific basis, with respect to what theoretical model, could you support your position? For example, we propose that multiple prayers are more efficacious than a single prayer. Our colleague asks the obvious question: Why should it matter? What logic, reasoning, model, or theory could answer our colleague's question? Unless we answer theologically, we cannot answer at all. Here is another example: we propose that a prayer offered by 2 people is more efficacious than the same prayer offered by 1 person. Why should it matter? Again, there is no scientific answer. The fact is that there is no scientific or theoretical way to answer these questions. They are about personal relationships to God and are answered only by moral, theological, philosophical, or spiritual methods that go beyond mechanistic causality. If we cannot answer the questions scientifically, then the construct cannot be studied through empirical research.

Yet, one might argue that if an empirical effect is found, it proves something. On the contrary, we would argue that an empirical effect, which (as the later section "Statistical and Measurement Issues" will show) is not hard to find when dealing with human beings, means nothing to science outside some link back to construct validity and nature. Hempel,¹⁰ in his Philosophy of Natural Science, has outlined basic criteria for scientific explanations. He describes explanations of phenomena that provide psychologically or spiritually valid answers (including God's will, power, or plan) as inadequate for the purposes of science. Their inadequacy derives from their failure to meet 2 essential scientific criteria: explanatory relevance and testability.

Explanatory relevance refers to the requirement that an explanation for a phenomenon must constitute grounds for expecting a given result under a given set of conditions. When we offer that a rainbow will appear any time light from behind an observer is refracted through water droplets in front of the observer, the explanation has relevance to the phenomenon being observed. That is, even without ever having seen a rainbow, the explanation allows us to expect that under the conditions specified, one will appear. (This criterion is similar to that of "prediction" as proposed by Lambert and Brittan,¹¹ who state that for an explanation to be adequate in science, it must indicate "why, given certain antecedent conditions, the event to be explained could have been expected to occur.") Alternatively, if we offer that rainbows are the bridge that souls cross to the afterlife, our explanationnow and foreverhas no link to the observed phenomenon and allows us no conditions under which to expect or predict the rainbow's occurrence (which, of course, only matters with respect to science and not with respect to spirituality, literature, culture, personal meaning, etc).

The second requirement is testability, or that a given explanation must be empirically testable, in principle if not in reality. That is, a scientific hypothesis must have certain "test implications," such that empirical findings can either support it or contradict it. The fact that you can create a research design for a given hypothesis does not ipso facto make the hypothesis scientific. Hempel¹⁰ offers the example of entelechies or "vital forces" as explanations for the "organic directiveness" that occurs in embryonic development. Experiments that demonstrate repair of embryonic damage can and have been done. Yet a hypothesis that this process is a function of entelechies would be neither supported nor contradicted by the experimental results because the entelechal explanation cannot make differential predictions regarding when these forces will manifest and in what manner they do their directing. The vital force is just "there." Hempel calls such hypotheses "pseudo-hypotheses" (with respect to science, but not with respect to their intuitive, spiritual, or psychological validity). Distant intercessory prayer is no more testable than entelechies in this sense: God is there when effects are found or not there when they are not. To derive specific implications of such a hypothesis is not possible. It is akin to qualifying an explanation of the origin of species through evolution with the proviso that, while evolution may work within the principles of biology and chemistry, it is really God who directs it. The proviso, while potentially spiritually meaningful, is scientifically irrelevant (ie, it is without explanatory relevance and testability). As Hempel¹⁰ notes, it is not permissible to propose a hypothesis and call it scientific, despite its lack of test implications or suggestion by auxiliary hypotheses, solely for the purpose of adding legitimacy to a deeply held conception.

The scientific explication of the construct prayer also contradicts many spiritual and theological conceptions of prayer. It contradicts many individuals' habits and beliefs regarding prayer and faith. For example,

· What have we found if we find a significant difference in a distant intercessory prayer study attributable to the number of intercessors or frequency of prayer? Have we found a way to use science to manipulate God, the omniscient, omnipotent creator of the universe? Or have we found that God's "brain," like those that evolved in humans, appreciates an additive, linear effect (eg, two 10-minute prayers by 1 intercessor equals one 5-minute prayer by 4 intercessors)?

· What have we found if we discover a significant difference between Christian and Jewish intercessors? Validation of Christ? (No, as it turns out. The futility of empirically validating the mystical has been clear to many, including the deeply religious. At the turn of the last century, William James,¹² philosopher and psychologist, asserted that "we cannot distinguish natural from supernatural effects; nor among the latter know which are favors of God, and which are counterfeit operations of the demon.") What are the theological and social implications of a study pitting Jewish intercessors against Moslem intercessors? (Would anyone dare such a study? Would the National Institutes of Health fund it? Would any institutional review board approve it?)

· When Catholics each Sunday at Mass offer a communal special intention to God to "heal all who are sick and suffering," is the probability of efficacy higher or lower than a single Catholic praying for a single individual (who would also be included in "all who are sick and suffering")? How about a group of Catholics praying for a single individual? These questions are not even askable, let alone answerable. In one author's experience (J.C.), the priest himself has been known to follow this petition with the proviso: "We have asked for what we think we need; give us what you know we need."

· In the major religious traditions, prayer that tests for a response from God in the way the intercessor requires would not be considered prayer at all because it requires no faith, leaves God no options, and is presumptuous regarding God's wisdom and plan. Where is faith if science can validate the power of prayer? Where is faith if the results of research can influence whether, why, and how the devout (or nondevout) will pray or if more and more precise explication of the construct prayer can eventually produce a scientific law, thereby putting more and more constraint on God? If distant prayer studies can make the results of prayer consistent, predictable, and replicable, then faith has become science.

So, we are left with a causal construct that violates the scientific requirements of explanatory relevance and testability, a construct for which we are unable to discern essential features and their implications without violating basic assumptions of faith. The only possible essential feature of prayer is, perhaps, whether it occurs or not, and maybe whether it occurs sincerely or not (but, again, how could you ever tell?). There is no possible way, of course, to control its occurrence (remember all those Catholics praying for all the sick every week), and if its sincere occurrence is its only essential (but uncontrollable) feature, then all other features are arbitrary and therefore not manipulatable. Prayer resists scientific explication and, unfortunately, nature has nothing to say about the ways of God.

A CONSTRUCT OUTSIDE THE BOUNDS OF SCIENCE

God is the purported causal mechanism in these studies, or God's equivalent. Why else would the intercessors need to be blinded? Keeping the patients and their physicians blind to the conditions is a reasonable approach; this constitutes "double-blind" in the classic sense. Yet, all of the prayer studies reviewed by Astin et al¹ went to great lengths to ensure that the intercessors had no contact with their targets or even knew who they were (some studies have provided pictures and/or first names or initials of the targets, while others have not, but no study identified the prayer targets). But why should that matter? If the intercessor is blind to the target and prayer is the treatment, what is the mechanism by which the effect occurs if it is not God, or at least some metaphysical mediator or mediating process that functions omnisciently (eg, we are told that the "healers" in the study by Sicher et al⁵ were not specifically directed to pray to God, but were told to "direct an intention"; however, many of them were religious healers)? We maintain that studies of distant prayer have had this feature not to test the effect of prayer on human illness, but to test for God's intervention in response to human intercession.

Intercessor-blind studies of prayer are analogous to therapist-blind studies of psychotherapy or practitioner-blind studies of therapeutic touch. They do not make sense. The mechanism of action cannot be differentiated from the sentient human being who delivers it. In reference back to the antibiotic treatment analogy, a blind intercessor is akin to an antibiotic stripped of the property that enables it to inhibit cell wall production in a bacterium: the intercessor knowing for whom he or she is praying is as much a part of the "treatment" as the antibiotic's "knowledge" of how to inhibit cell wall production, and only in bacteria cells. If we "blinded" the antibiotic, by what process, if not miraculous, would it destroy bacteria? A prayer study with blinded intercessors makes sense only if some omniscient mediating phenomenon or process knows how and where to apply the mechanism of action (and chooses to do so). If intercessors are blind, the researcher finding significant differences has only 1 recourse to an explanation: God. (In the study by Byrd,² no attempt was made to hide this fact.) All possible natural phenomena (bioenergy transference, magnetic wave affects, etc, all of which are worthy of investigation) have been ruled out.

Moreover, why is it vital to these prayer studies that patients not know that they are being prayed for? Why not study the placebo response directly? For example, Do subjects who know that someone is praying for them do better than subjects who do not know? Such a study would help to explicate the nature of the mind-body interface, and would contribute to a fascinating and powerful literature on expectancy effects in medicine.^{13, 14} Such research would neither deny nor support miracles, the hand of God, or the providential effect of prayer; it would simply say nothing about them. Recently, Matthews and MacNutt (both well-published and unabashed advocates for the healing power of prayer), along with colleague Marlowe,⁷ found interesting (and, in some cases, substantial) effects of in-person prayer teams on symptoms of rheumatoid arthritis outpatients (while, incidentally, finding no effect for distant intercessory prayer). They acknowledged the potential role of placebo in their findings and speculated on the mechanisms of action involved in the changes they documented (including immunological effects). They also challenged future researchers to differentiate the effect of in-person prayer from other forms of sympathetic clinical attention, both spiritual and nonspiritual. None of these notions requires a supernatural construct. Consider, by contrast, the hypothesis of Sicher et al⁵ that distant healing might involve "previously undescribed energy transfer" or "mind-to-mind communication." Why then, if they are willing to consider a natural phenomenon as the causal agent, were their intercessors blind? If the intercessors were transferring energy or somehow healing disease by communicating telepathically, would it not be imperative that there be a known destination for this energy or a known mind with which to communicate? (These questions can be asked as well of much of the secular, nonprayer distant healing research.^{1, 15}) Why not have the intercessors surreptitiously sit outside the examination room and direct their intention right at the patient? Why not look for disturbances in the electromagnetic field around patients when an intercession is occurring? Why not test in-person healing with a random assignment of healers directing intentions vs nonhealers (pretending to be healers) reciting nonsense syllables in their heads? Clearly, there are alternatives to testing for God's intervention, alternatives that are fascinating, useful, and, above all, consistent with a scientific approach without being theologically untenable.

The fact that it is God (or God's equivalent) that is being tested is the fundamental distinction between explaining the results of distant prayer studies and explaining the mysterious findings of other research (in which the history of medical research abounds). Darwin did not have access to Mendelian genetics when he developed the theory of evolution by natural selection. He did not know how offspring acquired subtle differences from their parents in ways that favored survival. Yet, this did not invalidate his theory. The difference is that what Darwin observed did not presuppose a supernatural construct as an explanation; it allowed nature and experimentation at some point in time to furnish an explanation. If we find contradictions in nature it is not because contradictions exist in nature, it is because we do not know nature well enough. As Verschuuren¹⁶ notes, Darwin's theory had an "in-built capacity to adapt to new developments, to grow into new fields, and to provide a delimited and reduced mechanism of speciation that allowed many small-scale experiments. . . ." (italics ours). Unfortunately, in practice the distant prayer studies remove nature from the equation, substituting God in its place. But as we shall see, God is neither confined by the natural world nor amenable to being experimented on, low P values notwithstanding.

STATISTICAL AND MEASUREMENT ISSUES

If God Is Infinite, So Are and

The role of God as the mechanism by which distant prayer works separates prayer research from science in other ways as well, most notably in the way statistical and measurement issues have been applied. For example, Why are statistical tests necessary to detect differences if God is in the works? Does it make sense that God would operate probabilistically? Why would there be any overlap in the score distributions for those getting the treatment vs those not? And, judging from the effect sizes in the distant intercessory prayer research, there is plenty of overlap. On an individual human being basis, what are we to make of the 40% of the control group who did better than the mean of the treatment group (when, on average, the control group did worse) and the 40% of the treatment group who did worse than the mean of the control group (when, on average, the treatment group did better)? If the mechanism is miraculous and metaphysical, the need for probability statements and the overlap of distributions is difficult to explain. A related issue is the very idea of subjecting God to probabilities as we set them up. With respect to God's intercession in the world, it is clear from many of the major religious traditions that God is not bound by natural laws or probabilities. Rabbi Luzzatto¹⁷ makes this clear in his explication of God's workings in the world, including miracles, from a Jewish perspective:

The spiritual consists of all entities which are not physical and which cannot be detected by physical means [I:5:1]. . . . Every physical entity and process is under the charge of some type of angel. These angels have the responsibility of maintaining each of them, as well as bringing about changes within them according to G-d's decree [I:5:2]. . . . When he wills, G-d can change the order of creation at any time. He can bring about various miracles and wonders, as He desires and deems beneficial for creation, according to the time and circumstances [II:5:6]. . . . The Highest Wisdom decreed . . . that these Forces [Kochos, a type of transcendental being] should be able to act upon the physical world in another manner . . . that conforms to their own innate laws, rather than the physical laws of nature. It is through this mode of action that the normal laws of nature can be suspended and altered on the physical plane [III:2:2]. [Italics ours.]

Attempts to impose statistical order on God become more absurd when we consider the proposition on which null hypothesis significance testing is actually based: "assuming the null hypothesis (of no effect, no difference, no relationship, parameter µ = X, etc) to be true, how probable are the results obtained?"¹⁸ What question are we asking with distant prayer research? "Assuming that God cannot heal at the bequest of human intercessors, what is the probability of getting these results?" (Keep in mind that a low probability does not then tell us that ipso facto the null hypothesis is false. Significance testing assumes the null hypothesis to be true; it cannot, therefore, yield a direct measure of the falsity of the null.^18-20 Like many other researchers, distant prayer researchers make the mistake of believing that a low probability level allows them to "linguistically transform" their P value into a meaningful finding for a given study.²⁰ In actuality, statistics are a tool for evaluating error probabilities over the long run; they are not an "index of truth."²¹) It is, perhaps, absurd to even have a null hypothesis (that God has no effect), if there is no way for it to be true (of course God can heal, when so inclined). Statistical hypothesis testing also fixes probabilities for mistakes, the classic type I error (ie, , or the probability of rejecting a true null hypothesis) and type II error (ie, , or the probability of failing to reject a false null hypothesis). But can God be held to these parameters? Must acts of God conform to a normal distribution as so many natural phenomena do (or are assumed to do)? It makes no sense to conduct a power analysis and set to a certain level if the laws of probability can be rescinded at any time. If God is the mechanism, the flipped coin does not have to, over time, yield half heads and half tails. If God is the mechanism, the distribution of means from successive random samples drawn from a population can take on any shape imaginable, at any time, at any moment.

Consider an analogy. Suppose that nature consists of a large glass jar filled with 50% blue marbles and 50% red marbles. As scientists, we want to find out what proportion of marbles is red. We do not, of course, have direct access to all the marbles. In fact, the only way to get marbles from the jar is by using a marble interferometer. We test the hypothesis that the proportion of marbles favors red (ie, that the proportion of marbles is not 50:50). We set up a mathematical model assuming that the proportion of red marbles to blue marbles is equal, then calculate theoretical comparison distributions based on an infinite number of random samples of various sizes. This concludes the theoretical work (deductive reasoning), and we are ready for the empirical work (inductive reasoning). We use the marble interferometer to obtain a sample of marbles. We compare the proportion of red marbles in our sample with what would be expected if the proportion of red marbles was 50%. This is the simplest form of hypothesis testing. Where are the flaws in the prayer studies relative to this analogy? The primary problem relates to what happens when we obtain our sample. The prayer studies have no equivalent to the marble interferometer. The process is more analogous to an omniscient being (God) choosing specific marbles and handing them to the experimenter. The actual sample is not chosen in the way that the theoretically infinite random samples are chosen (with an interferometer-like process) to generate a statistical comparison distribution (in the case of both parametric and nonparametric statistics). But it is with these comparison distributions that our actual samples are compared in hypothesis testing. Statistical comparison distributions come from a mechanistic model of the universe, while the actual sample in a prayer study comes from a teleological model of the universe. Conventional statistics just do not work in this model. (But can we test whether the world is teleological by determining whether our data conform to mechanistic models? No, because there are an infinite number of alternative naturalistic hypotheses to explain any discrepancy. Science presumes a mechanistic world.) With respect to statistical conclusion validity,⁸ the distant prayer studies were invalid before any data were even collected.

When the dutiful distant prayer researcher says that "there is an 80% probability of a P value of less than .05 for differences of at least X," do the powers of God suddenly become constrained? We think not. If God is infinite, so are and .

Measurement and the "Crud Factor"

A related issue involves the nature of the outcome variables for distant prayer research. A review of these studies shows that the outcome variables are generally abundant, sometimes "soft," and never theoretically and conceptually grounded or differentiated. Cohen,²² the godfather of such primary statistical concepts as power analysis, has discussed in detail Meehl's concept of the "crud factor."^{18, 23} Basically, thanks to the crud factor, on any given day everything can be related to everything else in the context of soft theory, abundant opportunities (the infamous "fishing" technique⁸), and/or soft outcomes. Consider the study by O'Laoire⁴: 11 outcomes, including state and trait anxiety, self-esteem, mood, depression, and self-perceived changes in creative expression and physical, emotional, intellectual, spiritual, and relational health. All measures were self-reported and the latter 6 were "home grown." (Regardless, O'Laoire failed to detect an effect for distant intercessory prayer on psychological health.) Consider the study by Harris et al³: there were no less than 35 parameters (for which the investigators made some adjustment in the analysis), ranging from need for an electrophysiology study to death. As it turned out, no single component achieved statistical significance as a function of blinded distant intercessory prayer, but an unvalidated "weighted and summed scoring system" did (but that only weakly). And in the study by Sicher et al⁵ 11 outcome variables with large standard errors (SEs) yielded a hodgepodge of inconsistent, marginal results.⁵ Although a multivariate approach was used in an attempt to account for correlation among outcomes, the recommended correction to the obtained univariate probability values was not made.²⁴ In addition, changes in statistical tests to account for severe distributional anomalies of certain variables increased some P values beyond conventionally accepted levels. Of the 11 outcomes, 4 changed significantly at an unadjusted univariate level. Of these, 1 was a self-reported indicator of distressed mood that, when adjusted for baseline differences, indicated more improvement in the control group. The only truly objective variables, CD4⁺ levels and death, had the highest P values (.55 and >.99, respectively). Finally, what of the oft-cited study by Byrd?² More than 2 dozen uncorrected univariate comparisons of variables with very low frequencies or large SEs (among which 6 statistically significant effects emerged) were made, followed by a multivariate procedure that maximizes chance findings (stepwise regression).²⁴ An unvalidated "hospital course" grade ("good, intermediate, or bad") of unknown reliability was also significantly different at a univariate level, but it overlapped considerably with the 29 individual outcome parameters, particularly the bad grade, which encompassed 4 of the 6 individually significant factors. We propose that a crud factor–like mechanism is more than sufficient to explain these and other positive distant prayer results, as have others.^25-27

But even if it were not, what are we to make of prayer studies where only a small fraction of the outcome measures change significantly? The related conceptual and theoretical issues are at least as compelling as the statistical ones. In no intercessory prayer study has there ever been an a priori specification of outcome criteria. Which outcome variables do the researchers expect to change, by how much, and, most important, why? This gets back to the previous discussion of construct validity: the outcomes of prayer resist explication as well (construct validity of the effect). The result is a hodgepodge of variables, some of which change and some of which do not. But there is never an explanation as to why some did and some did not; no mechanism is ever proposed to address the differences. Even the tentative explanation of Sicher et al⁵ that distant healing might be a function of "some previously undescribed energy transfer" cannot explain how the energy knows where to go and why it affects outpatient visits and not CD4⁺ levels.⁵ Consider again the study by Byrd.² The "Judeo-Christian God" apparently saw fit to require only 3 of the 192 intercessory prayer patients (1.6%) to need antibiotics, while permitting 17 (8.5%) of the 201 controls to experience this fate. Yet this same "beneficial therapeutic" God did not significantly alter the number of patients requiring, to name but a few, antiarrhythmic agents, pressure monitoring, readmission to the cardiac care unit, vasodilators, pacemakers, major surgery, and antianginal agents. Further, mortality rates were not different as a function of prayer; neither were rates of sepsis, unstable angina, or gastrointestinal bleeding; neither were cardiac care unit days, total hospital days, or number of discharge medications. Why did God pick antibiotics? Could this result have been predicted before the study? Could differential predictions have been made regarding the mass of outcome variables? The answer to each of the latter 2 questions is "No."

The choice of outcome variables is a telling and unfavorable commentary on distant prayer research. Why are not more convincing approaches to outcome constructs used in this line of research? Consider a distant intercessory prayer study for blind persons, directed at the restoration of sight, with secondary measures of change in subjective parameters such as well-being, mood, or quality of life. A differential prediction could be made, and a single objective outcome criterion exists. The same study could be done with paraplegics with severed spinal cords, amputees, or the dead. If the mechanism of action is Godly and miraculous, there is no logical reason to shun these studies other than the most important reason: that they trivialize and distort the nature of prayer, God, and our relationship to the transcendent by bringing them into a sphere in which they do not belong.

ALTERNATIVES TO TESTING PRAYER

As our concerns with distant prayer research proliferated, we began to consider the extent to which this line of research detracts from other more productive activities regarding religion and medicine. Recently, numerous commentators and researchers have presented pertinent information on a variety of topics in this area, including physician attention to patient religious needs, spiritual history taking in the clinic, sensitization of medical students and residents to the importance of religion or spirituality to patients facing health crises or death, epidemiological findings on the health benefits of religion, the centrality of religion to the elderly patient, and the universal role of religion in interpreting crises of meaning (eg, premature death or disability) in the lives of human beings.^28-37

Much of this work has been done in the spirit of patient-centered medicine,²⁸ within a model that acknowledges a portion of the physician's role concerned with supporting, healing, caring, and palliating. This work has, for the most part, been decidedly nondenominational, seeing the very diversity of patient faiths as one driving force in the recognition of patient religion. Moreover, researchers like Levin³³ have proposed hypotheses for the relationship of religion to health that are grounded in nature, including the role of religion in promoting good health behaviors (eg, substance use and abuse, diet, and sex), the psychoneuroimmunological effects of ritual and worship, the social support and relationships engendered by religious affiliation, and the psychological effects of religion as mediated by feelings of hope, forgiveness, love, optimism, and joy.³³

Distant intercessory prayer studies project a distorted image (especially when filtered through the media) of how science is accomplished and, perhaps more important, what "truth" means in the context of belief, spirituality, and religion.³⁸ The effect is to distract us from the more real and more human interaction of religion and medicine: the acknowledgment that religion and spirituality are often vitally important to our fellow human beings in their experience of illness, disability, suffering, dying, and death. Considering ways of bringing medicine closer to a recognition of that importance would be more beneficial than applying the scientific method to prayer.^{39, 40}

The continued desire of some to do the latter misses the point anyway. As Verschuuren¹⁶ suggests, science is a collection of abstract maps of the world, not a substitute for our own experience of the world (our own intuitive map).¹⁶ Each specialized map must omit everything that is irrelevant, out of scope, and out of scale. What is the result? Science does not deny God, miracles, and the like; rather, it merely neglects them. Further, the maps of science can never "exhaust the original itself" so that although a "map-world may show joy and sadness, hatred and love, it never can actualize them. . . . For the same reason it is impossible that God would occur as an entity on a map" (italics ours). The important message is to stop fighting the maps (and this includes scientists who "trust their own maps more than the world itself" as well as "people wanting to use them for unwanted destinations"). Science cannot actualize our spirituality, so why do we ask this of it?

CONCLUSIONS

In chapter 17 of the Book of Exodus (KJV), Moses must confront one of many potential rebellions by the Israelites, this one prompted by a lack of water. After smiting the rock and producing the water, Moses names the place Massah, the Hebrew word for "trial." Verse 7 tells the story: "And he called the name of the place Massah . . . because they [the Israelites] tempted the Lord, saying, Is the Lord among us, or not?" Later, in the Book of Deuteronomy where many of the Jewish laws are explicated, this event carries enough weight to warrant its own command: "Ye shall not tempt the Lord your God, as ye tempted him in Massah" (6:16, KJV). And much later, when Yeshua is having his visions in the wilderness and wonders whether God's angels would actually bear him up in their hands if he leaped from a pinnacle of the temple, he chastises his beguiler with these very words: "It is said, Thou shalt not tempt the Lord thy God" (Luke 4:12, KJV).

The lesson of Massah is that God cannot be compelled by our research designs, statistics, and hypotheses to answer our demand, "Is the Lord among us or not?" Massah makes it clear that our intercessions must be a matter of faith and trust in God, of putting our hope in God, of knowing we are part of God no matter what the outcome in our experience of the physical world. As a metaphor for the testing of distant prayer through the scientific method, Massah tells us not only that God should not be tested but, more important, that God cannot be tested.^{41, 42} As Sloan and Bagiella⁴⁰ have written, mixing experimental method with faith degrades both concepts. We do not need science to validate our spiritual beliefs, as we would never use faith to validate our scientific data.

Author/Article Information


John T. Chibnall, PhD
Department of Psychiatry
Saint Louis University School of Medicine
1221 S Grand Blvd
St Louis, MO 63104
(e-mail: [log in to unmask])

Joseph M. Jeral, MD
Washington, DC

Michael A. Cerullo, MD
Lexington, Ky

We are indebted to Jennifer Accardo, MD, for her critical insights.

REFERENCES

1.
Astin JA, Harkness E, Ernst E.
The efficacy of "distant healing": a systematic review of randomized trials.
Ann Intern Med.
2000;132:903-910.
MEDLINE

2.
Byrd RJ.
Positive therapeutic effects of intercessory prayer in a coronary care unit population.
South Med J.
1988;81:826-829.
MEDLINE

3.
Harris WS, Gowda M, Kolb JW, et al.
A randomized, controlled trial of the effects of remote, intercessory prayer on outcomes in patients admitted to the coronary care unit.
Arch Intern Med.
1999;159:2273-2278.
ABSTRACT  |  FULL TEXT  |  PDF  |  MEDLINE

4.
O'Laoire S.
An experimental study of the effects of distant, intercessory prayer on self-esteem, anxiety, and depression.
Altern Ther Health Med.
1997;3:38-53.
MEDLINE

5.
Sicher F, Targ E, Moore D II, Smith HS.
A randomized double-blind study of the effect of distant healing in a population with advanced AIDS: report of a small scale study.
West J Med.
1998;169:356-363.
MEDLINE

6.
Walker SR, Tonigan JS, Miller WR, Comer S, Kahlich L.
Intercessory prayer in the treatment of alcohol abuse and dependence: a pilot investigation.
Altern Ther Health Med.
1997;3:79-86.
MEDLINE

7.
Matthews DA, Marlowe SM, MacNutt FS.
Effects of intercessory prayer on patients with rheumatoid arthritis.
South Med J.
2000;93:1177-1186.
MEDLINE

8.
Cook TD, Campbell DT.
Quasi-Experimentation: Design and Analysis Issues for Field Settings.
Boston, Mass: Houghton Mifflin Co; 1990.

9.
Simpson J, ed, Weiner ES, ed.
Oxford English Dictionary.
2nd ed. Oxford, England: Oxford University Press Inc; 1989.

10.
Hempel CG.
Philosophy of Natural Science.
Englewood Cliffs, NJ: Prentice-Hall International Inc; 1990.

11.
Lambert K, Brittan GG.
An Introduction to the Philosophy of Science.
Englewood Cliffs, NJ: Prentice-Hall International Inc; 1970.

12.
James W.
The Varieties of Religious Experience.
New York, NY: Random House Inc; 1999.

13.
Benson H.
Timeless Healing: The Power and Biology of Belief.
New York, NY: Scribner; 1996.

14.
Kirsch I, ed.
How Expectancies Shape Experience.
Washington, DC: American Psychological Association; 1999.

15.
Harkness EF, Abbot NC, Ernst E.
A randomized trial of distant healing for skin warts.
Am J Med.
2000;108:448-452.
MEDLINE

16.
Verschuuren GMN.
Investigating the Life Sciences: An Introduction to the Philosophy of Science.
Oxford, England: Pergamon Press Inc; 1986.

17.
Luzzatto MC.
The Way of G-d.
4th ed. Kaplan A, trans. Jerusalem, Israel: Feldheim Publishers; 1983.

18.
Cohen J.
The earth is round (p<.05).
Am Psychol.
1994;49:997-1003.

19.
Falk R, Greenbaum CW.
Significance tests die hard: the amazing persistence of a probabilistic misconception.
Theory Psychol.
1995;5:75-98.

20.
Goodman SN.
Toward evidence-based medical statistics. 1: The P value fallacy.
Ann Intern Med.
1999;130:995-1004.
MEDLINE

21.
Chibnall JT.
Some basic issues for clinicians concerning things statistical.
Int Psychogeriatr.
2000;12:3-7.
MEDLINE

22.
Cohen J.
Statistical Power Analysis for the Behavioral Sciences.
Mahwah, NJ: Lawrence Erlbaum Associates Inc; 1988.

23.
Meehl PE.
Why summaries of research on psychological theories are often uninterpretable.
Psychol Rep.
1990;66(suppl 1):195-244.

24.
Tabachnik BG, Fidell LS.
Using Multivariate Statistics.
3rd ed. New York, NY: HarperCollins Publications Inc; 1996.

25.
Hoover DR, Margolick JB.
Questions on the design and findings of a randomized, controlled trial of the effects of remote, intercessory prayer on outcomes in patients admitted to the coronary care unit [letter].
Arch Intern Med.
2000;160:1875-1876.
FULL TEXT  |  PDF  |  MEDLINE

26.
Sandweiss DA.
P value out of control [letter].
Arch Intern Med.
2000;160:1872.
FULL TEXT  |  PDF  |  MEDLINE

27.
Posner G.
Another controversial effort to establish the medical efficacy of intercessory prayer.
Sci Rev Altern Med [serial online].
2000;4(1). Available at: http://www.hcrc.org/sram. Accessed March 6, 2001.

28.
Barnard D, Dayringer R, Cassel CK.
Toward a person-centered medicine: religious studies in the medical curriculum.
Acad Med.
1995;70:806-813.
MEDLINE

29.
Chibnall JT, Brooks CA.
Religion in the clinic: the role of physician beliefs.
South Med J.
2001;94:374-379.
MEDLINE

30.
Chibnall JT, Call CB, Jeral JM, Holthaus C.
Student religiosity and attitudes toward religion in medicine at a private Catholic medical school.
Fam Med.
2000;32:102-108.
MEDLINE

31.
Fricchione GL.
Illness and the origin of caring.
J Med Humanit.
1993;14:15-21.

32.
Larson DB, Swyers JP, McCullough ME.
Scientific Research on Spirituality and Health: A Consensus Report.
Rockville, Md: National Institute for Healthcare Research; 1998.

33.
Levin JS.
How religion influences morbidity and health: reflections on natural history, salutogenesis and host resistance.
Soc Sci Med.
1996;43:849-864.
MEDLINE

34.
Matthews DA, McCullough ME, Larson DB, Koenig HG, Swyers JP, Milano MG.
Religious commitment and health status.
Arch Fam Med.
1998;7:118-124.
ABSTRACT  |  FULL TEXT  |  PDF  |  MEDLINE

35.
Maugans TA.
The SPIRITual history.
Arch Fam Med.
1996;5:11-16.
MEDLINE

36.
Post SG, Puchalski CM, Larson DB.
Physicians and patient spirituality: professional boundaries, competency, and ethics.
Ann Intern Med.
2000;132:578-583.
MEDLINE

37.
Puchalski CM, Larson DB.
Developing curricula in spirituality in medicine.
Acad Med.
1998;73:970-974.
MEDLINE

38.
James W.
The Will to Believe and Other Essays in Popular Philosophy and Human Immortality.
Mineola, NY: Dover Publications Inc; 1956.

39.
Rosner F.
Therapeutic efficacy of prayer [letter].
Arch Intern Med.
2000;160:1875.
FULL TEXT  |  PDF  |  MEDLINE

40.
Sloan RP, Bagiella E.
Data without a prayer [letter].
Arch Intern Med.
2000;160:1870.
FULL TEXT  |  PDF  |  MEDLINE

41.
Galishoff ML.
God, prayer, and coronary care unit outcomes: faith vs works [letter]?
Arch Intern Med.
2000;160:1877.
FULL TEXT  |  PDF  |  MEDLINE

42.
Hammerschmidt DE.
Ethical and practical problems in studying prayer [letter].
Arch Intern Med.
2000;160:1874-1875.
FULL TEXT  |  PDF  |  MEDLINE

Edward E. Rylander, M.D.

Diplomat American Board of Family Practice.

Diplomat American Board of Palliative Medicine.