Are There Demonstrable Effects of Distant Intercessory Prayer? A Meta-Analytic Review
Kevin S. Masters, Ph.D.
Glen I. Spielmans, Ph.D.
State University of New York–Fredonia
Jason T. Goodson, Ph.D.
Utah State University
common in the United States. Practitioners of these interven-
tions find them compatible with personal philosophies. Conse-
most commonly practiced alternative interventions and has re-
cently become the topic of scientific scrutiny. Purpose: This
study was designed to provide a current meta-analytic review
of the effects of IP and to assess the impact of potential modera-
tor variables. Methods: A random effects model was adopted.
Outcomes across dependent measures within each study were
pooled to arrive at one omnibus effect size. These were com-
bined to generate the overall effect size. A test of homogeneity
and examination of several potential moderator variables was
conducted. Results: Fourteen studies were included in the
meta-analysis yielding an overall effect size of g = .100 that did
the effect size reduced to g = .012. No moderator variables sig-
nificantly influenced results. Conclusions: There is no scientifi-
cally discernable effect for IP as assessed in controlled studies.
Given that the IP literature lacks a theoretical or theological
base and has failed to produce significant findings in controlled
trials, we recommend that further resources not be allocated to
this line of research.
(Ann Behav Med
for illness has grown exponentially in recent years. Perhaps the
greatest evidence of this trend is the establishment of the Na-
tional Center for Complementary and Alternative Medicine
(NCCAM) within the U.S. National Institutes of Health. In
2002, 62% of Americans reported using some kind of alterna-
tive medicine (1). In a study examining the characteristics of in-
dividuals who use alternative treatments, Astin (2) found that
they tended to be well educated and in poor health status but not
largely dissatisfied with conventional medicine. Rather, these
individuals found that the alternative treatments were more con-
gruent with their own values, beliefs, and philosophical orienta-
tions toward health and life in general. NCCAM has been spe-
cifically interested in studying therapies for which there is no
plausible biomedical explanation currently accepted by the bio-
logical and medical communities. This category includes many
traditional folk health practices as well as practices based on
ideas related to quantum physics or new age ideologies.
Of the 10 most often utilized alternative medicine proce-
dures in the United States, prayer for self (43%) and prayer for
others (24.4%) are the 2 most commonly named therapies, and
and philosophical orientation toward life and the cosmos. Nev-
ertheless, it is important to note that prayer in and of itself may
not qualify as a treatment that lacks a plausible biomedical ex-
planation. For example, when ill individuals engage in prayer
ities of the experience could trigger recognized psychophysio-
logical and psychoneural processes that plausibly influence
health and healing. Thus, this type of prayer certainly falls into
the category of an alternative medical procedure but not one
lacking theoretical credibility.
Prayer for oneself is, however, not the focus of this investi-
gation. Rather this meta-analysis investigated the effects of dis-
tant intercessory prayer (IP). Simply defined, IP is prayer said
the intercessor) is not present with the recipient of the prayer,
thus making the prayer distant. Practitioners of many of the
world’s religions offer up prayers for both the ill and those who
are not ill, for example, when traveling mercies are requested.
uals, such as laying on of hands or anointing with oil, where the
individual is present and aware that the prayer and ritual are oc-
curring. Another type of prayer occurs when intercessors pray
for individuals who are not present with them and may, in fact,
not even know that others are praying for them. These are also
common in many faith traditions and are a form of IP. In short,
many believe that prayer is effective in and of itself, and it does
not require that the recipient of the prayer be an active partici-
pant in either the prayer or the knowledge that someone is pray-
ing for him or her. This point is significant as an important fea-
Jason T. Goodson is now a postdoctoral fellow at Dartmouth Univer-
Convention of the American Psychological Association, August 2005,
Reprint Address: K. S. Masters, Ph.D., Department of Psychology, 430
Huntington Hall, Syracuse University, Syracuse, NY 13244–2340.
© 2006 by The Society of Behavioral Medicine.
participants (mostly patients) did not know if they were receiv-
prayer or control group. Likewise, their health care providers
were also blind to prayer condition. Further, in some studies the
participants did not know that they were participating in a study
at all, constituting a type of triple-blind design. By implement-
ing these rigorous methodological procedures, the authors of
these studies removed known psychological processes or pla-
cebo effects from the set of possible explanatory mechanisms
that could account for significant findings.
ture pertaining to IP, but this meta-analysis is the most compre-
study not available to previous reviewers. It is also notable that
prior authors reached highly discrepant conclusions regarding
the proper future course for this research. Whereas Powell,
Shahabi, and Thoresen (4) concluded that there was some evi-
Cochrane Review of the IP literature (6) first published in 2000
view provided no guidance regarding the effectiveness of IP.
The purpose of this study was to provide a current meta-an-
alytic review of the effects of IP and to assess the impact of po-
tentially significant moderator variables. Studies included in
this review were specifically chosen because they investigated
IP rather than (or in addition to) other forms of distant healing
that have appeared in the literature. We were only interested in
IP due to its prominence in the culture and currently topical, but
controversial, stature in behavioral medicine.
To locate all relevant studies, PsycINFO and Medline data-
cles published prior to August 2005 were eligible for inclusion.
References in relevant review articles (6–9) were also searched
as were reference lists from articles included in the meta-analy-
sis. To meet inclusion criteria, studies must have (a) used IP as
an intervention to treat any type of medical or mental health
problem, (b) provided data that allowed for calculation of an ef-
fect size, (c) compared IP to a control group, and (d) blinded
participants as to their experimental condition. It was not re-
quired that participants be unaware of their participation in a
study. This strategy yielded 15 studies (noted by an asterisk in
the reference list). One study was excluded because it examined
the impact of “retroactive” IP on patients with a prior blood in-
fection (10). The outcomes of participants had been established
prior to the implementation of IP; consequently, we did not un-
derstand how this could be considered a prayer intervention and
hence decided that it did not merit inclusion. All other IP inter-
vention studies were included.
Outcomes across all dependent variables were pooled
within studies to provide one omnibus effect size for each study.
Effect sizes across studies were weighted by their inverse vari-
ance to provide an overall effect size estimate that most accu-
rately represented the true population effect size (11).
Level and detail of data reporting varied widely across pri-
mary studies. Consequently, effect sizes were computed from
means and standard deviations when possible. In their absence,
effect sizes were calculated from t tests and F tests. Effect sizes
from dichotomous outcome measures were computed using
procedures described in Hasselblad and Hedges (12), who pro-
vided a method for transforming odds ratios from dichotomous
data into effect sizes by using the following formula:
After all effect sizes were calculated, they were converted to
Hedges’s g, which corrects for a small bias in Cohen’s d (11).
All effect sizes were calculated using Comprehensive Meta-
Analysis software (13).
A random effects model was deemed most appropriate for
this research area given its heterogeneity and limited number of
studies. This method allows for greater generalization than the
fixed effects model (14), but a random effects meta-analysis is
typically more conservative than a fixed effects meta-analysis
effect size is zero). Knowing this, and given that this research
area could be described as early stage, we opted for flexibility
by also calculating effect sizes using the fixed effects model.
virtually identical; consequently, only the results of the random
effects model are reported.
A test of homogeneity, using Q, was performed to examine
whether the set of effect sizes were distributed around a com-
mon mean. Although a test of homogeneity that fails to reject
the null hypothesis argues against the existence of moderator
variables, such a test often lacks sufficient statistical power
to detect moderators (15). Thus, in addition to homogeneity
tests, we conducted moderator analyses based on participant
characteristics and study design features, as both of these sets
of variables are often related to findings in other areas investi-
tial impact of participant type, method of allocation of partici-
pants to conditions, frequency of prayer, and duration of prayer
Participants were divided on the basis of whether they were
from a patient population (i.e., individuals seeking medical or
psychological intervention for some type of physical or mental
problem) or were more representative of a healthy-analogue
sample. In the latter case, the research participants were not
seeking treatment but received prayer for another reason. For
22Masters et al.
Annals of Behavioral Medicine
log * 3.
example, in one study (16) the participants reported a difficult
life situation that was then the object of prayer. The effect of
prayer on participants currently receiving treatment for illness
(medical or mental health services) was compared to the effect
on participants with no identifiable health problems. This mod-
erator analysis utilized a Q test for comparing effect sizes be-
tween groups, a frequently used meta-analytic analog to the
analysis of variance (17).
tion was examined by comparing the mean effect size of studies
utilizing random assignment to studies that did not use random
assignment. To assess the impact of prayer frequency, the effect
of prayer was examined for participants who were prayed for
daily versus those prayed for less than daily. Both of the afore-
mentioned analyses utilized Q tests. Due to the wide variation
in intervention duration, the impact of this variable was exam-
ined using weighted least squares (WLS) regression, with each
study’s inverse variance as the weighting variable.
It also seems possible that type of prayer (directive,
nondirective, silent meditation, etc.) could influence the effi-
cacy of prayer intervention. Unfortunately, due to lack of spe-
cific information presented in the studies, we were only able to
broadly lump type of prayer into either directive or nondirective
categories (Table 1), potentially leaving much variance within
classes. After making this classification it became apparent that
because of the relative lack of pure nondirective prayer condi-
that the religious beliefs of intercessors may moderate the ef-
fects of intercessory prayer. Eight studies explicitly examined
the effects of intercessory prayer as performed only by Chris-
tians. It may thus appear possible to compare effects between
Christian intercessors and others. However, such an analysis
would be misleading, as it is likely that most of the intercessors
in the studies in which intercessor faith was not clearly reported
were Christian. Indeed, only one study mentioned including
non-Judeo-Christian intercessors (3). Thus, the moderating in-
fluence of intercessor faith was not examined.
Important characteristics of the individual studies included
were diagnosed with some sort of medical or mental health con-
dition whereas three studies used healthy participants. In all 14
studies participants were blind to treatment condition and in
many cases the study blind was so thorough that neither patients
to conditions. Other studies used a multiple baseline across sub-
jects design (18), matched assignment (19), and a combination
of random assignment and Assignment × Time (20). Studies
varied widely in the lengths of their interventions, from a mini-
mum of approximately 1 week (actually was timed to hospital
discharge, which averaged 7.6 days) (21) to 15 months (22). A
great deal of heterogeneity was present in the frequency of
prayer across studies ranging from once weekly to daily, with 8
studies implementing daily prayer. The studies demonstrated
high variability regarding the number of dependent variables re-
ported with one listing 40 measures (23) and another utilizing
only 1 (24).
Effects of Intercessory Prayer
Across the 14 studies, using a random effects model, the
cance (see Table 2). The findings for each individual study and
Volume 32, Number 1, 2006
Meta-Analysis of Prayer 23
Intercessory Prayer Studies Included in Meta-Analysis
StudyPrayer Type Intercessors’FaithConditionga
Aviles et al. (2001)
Cha & Wirth (2001)
Harris et al. (1999)
Joyce & Welldon (1965)
Krucoff et al. (2005)
Mathai & Bourne (2004)
Matthews et al. (2000)
Matthews et al. (2001)
O’Laoire et al. (1996)
Palmer et al. (2004)
Tloczynski & Fritsch (2002)
Walker et al. (1996)
riety = Christian, Muslim, Jewish, and Buddhist; CAD = coronary artery disease; MH = various mental health problems; RA = rheumatoid arthritis; kidney
dialysis = patients on kidney dialysis.
aPositive value for g represents a positive effect for intercessory prayer.bIn this small sample size study, two patients, both in the control group, had a differ-
ent and more deadly form of leukemia than the other patients. They were excluded from the analysis.
the overall result are graphically portrayed in Figure 1. A mod-
erator analysis based on patient health achieved borderline sig-
nificance, suggesting more positive change for patients during
IP intervention than for healthy participants (Q = 2.86, p =
.091). When only studies examining ill patients were consid-
ered, the mean effect size, although small, achieved statistical
significance, whereas healthy participants showed no benefit
from intercessory prayer (Table 2).
One study with a large effect size supporting the efficacy of
IP (25) turned out to be unusual in a number of ways (26,27).
When originally published, the study had three authors. How-
ever, one author removed his name from the study, stating that
he only provided editorial input for the investigation which did
not merit inclusion as an author. Another author, Wirth, was re-
cently convicted of fraud-related charges unrelated to the study
and a newspaper report (27) casts considerable doubt on wheth-
er the prayer groups in the study were actually conducted. We
thought that the strange circumstances surrounding this article
warranted examination of the overall results both including this
study and excluding it from consideration.
When Cha and Wirth (25) was removed from the analysis,
the omnibus effect size across studies diminished to nearly zero
icant difference between sick and healthy participants (Q = .82,
p = .37). Further, the previously reported small effect size in
support of the efficacy of IP among sick patients was no longer
A comparison of the effect of IP on participants randomly
assigned to conditions versus participants nonrandomly assigned
revealed no significant difference (Q = .038, p = .85). In addi-
tion, the mean effect size of studies that utilized daily prayer
showed no significant difference compared to studies that uti-
lized less frequent prayer (Q = .33, p = .57). A WLS regression
analysis revealed no relationship between mean effect size and
duration of prayer intervention (β = –.02, p = .94). The afore-
mentioned analyses indicate that study design characteristics
did not serve as moderator variables in this analysis.
A test of homogeneity yielded nonsignificant results (Q =
15.07, p = .30), indicating that the studies were likely clustered
study in the set.
This study provides a quantitative review of the research
addressing the potency of IP for improving one’s physical con-
dition or life circumstance. The most parsimonious statement to
be made from this literature is that there is no scientifically
discernable effect that differentiates the status of individuals
who are the recipients of IP from those who are not. Further,
there was no evidence to suggest that these results were influ-
enced by potential moderators such as method of allocation of
research participants to groups (random vs. nonrandom), prayer
dosage, that is, whether prayers were offered at least daily or
less often, or how long the prayer intervention lasted. Although
an initial analysis seemed to indicate the possibility that an ef-
fect could be found that differentiated studies on the basis of
whether the recipients of prayer were actual medical patients or
24Masters et al.
Annals of Behavioral Medicine
Effects of Intercessory Prayer Summarized Across Studies
Condition No. Comp.ga
Patient (without Cha
& Wirth, 2001)
Overall (without Cha
& Wirth, 2001)
aPositive value for g favors effect for intercessory prayer.
No. comp. = number of comparisons.
favors intercessory prayer (IP) intervention.
Forest plot of effect sizes and confidence intervals for individual studies and overall. Note that, due to data entry, a negative weighting
vanished when one highly controversial and somewhat doubtful
study was removed from the analysis.
It should be pointed out that all of these studies suffer from
a major and unsolvable methodological flaw, that is, receipt of
prayer cannot be controlled and therefore it is impossible to
know to what degree individuals in the control groups were ac-
tually the recipients of the “intervention” (IP) from loved ones,
family members, clergy, or others apart from the research inter-
cessors. Krucoff et al. (3), in a multicenter study that is certainly
tients (undergoing percutaneous coronary intervention or elec-
tive catheterization) were aware that prayer on their behalf was
being offered outside of the study protocol. Further, 64% of
those not assigned to the prayer group believed that they were
lieved that they were not. Thus it could be argued that virtually
all participants receive prayer or at least believe they do. Given
this situation, perhaps the conclusion cited earlier should be
amended to state that there is no scientifically discernable effect
of IP initiated by a research team from those who do not receive
research team initiated IP.
fensible theoretical rationale gains greater prominence if this
line of research is to continue. However, the development of
such a rationale may, in and of itself, prove challenging. Chib-
nall, Jeral, and Cerullo (28) offered a brilliant criticism of the
theoretical and methodological underpinnings of this research
logical difficulties with IP studies. Masters (29) noted that the
preponderance of IP studies have been carried out by those of
Christian heritage. He then offered a critique from a distinctly
Christian perspective that suggested there is no basis in Chris-
tian theology or its worldview to support further IP research.
Sloan (5) incorporated many of these same arguments into his
critique and further places this research outside the bounds of
what is scientifically discernible. Interested readers are referred
to these sources for detailed analysis of this research at the theo-
retical and methodological levels. We believe that when these
factors are considered in concert with the results of this meta-
analysis, there is simply no reason to continue supporting re-
search on IP if the purpose of that research is to demonstrate an
effect for the recipients of IP when compared to nonrecipients.
Lacking a theoretical foundation, it appears that this area of
research has continued at least in part as a function of the misin-
terpretation of visible early studies. For example, the Byrd (21)
study is often referred to as the starting point for this research
cacy of IP. However, this study actually included 30 dependent
variables and found significant effects, without using any type
The one finding from this study that is most often cited in sup-
port of the potency of IP is the global rating of the patients’hos-
to replicate Byrd’s finding using the global scale Byrd devel-
oped, they were unable to do so. Further, Byrd found no differ-
ences on seemingly very important variables such as days in the
cardiac care unit, days in the hospital, or mortality. Interpreta-
tion of this study as supportive of the benefits of IP seems dubi-
ous at best.
First, we did not study prayer for one’s self or prayer said in the
presence of the prayer recipient. This research has nothing to
say about these practices. We are also not trying to persuade be-
lievers to stop their practice of IP as it is carried out within their
faith traditions. We see no health risk to patients or others from
IP, and, in fact, believe that there may be as yet unspecified and
unstudied benefits for the intercessors themselves. We encour-
defensible constructs and related models. Nevertheless, given
that the IP literature lacks a theoretical or theological base and
has failed to produce significant findings in controlled trials, we
recommend that further resources not be allocated to this line of
*References marked with an asterisk indicate studies included in the
(1) Barnes P, Powell–Griner E, McFann K, Nahin R: CDC Advance
Among Adults: United States, 2002. Washington, DC: National
Center for Complimentary And Alternative Medicine, 2002.
(2) Astin JA: Why patients use alternative medicine: Results of a
national study. Journal of the American Medical Association.
(3) *Krucoff MW, Crater SW, Gallup D, et al.: Music, imagery,
Monitoring and Actualisation of Noetic Trainings (MANTRA)
II randomized study. The Lancet. 2005, 366:211–217.
(4) Powell LH, Shahabi L, Thoresen CE: Religion and spiritual-
ity: Linkages to physical health. American Psychologist. 2003,
and health. Retrieved July 2005 from http://www.metanexus.
(6) Roberts L, Ahmed I, Hall S: Intercessory prayer for the allevia-
tion of ill health. The Cochrane Database of Systematic Re-
views: Reviews 2000. Article No. CD000368. Retrieved July
2005 from http://www.mrw.interscience.wiley.com/cochrane/
(7) Astin JA, Harkness E, Ernst E: The efficacy of “distant heal-
nal Medicine. 2000, 132:903–910.
(8) Ernst E: Distant healing—An “update” of a systematic review.
Wiener Klinische Wochenschrift. 2003, 115:241–245.
(9) Townsend M, Kladder V, Ayele H, Mulligan T: Systematic re-
view of clinical trials examining the effects of religion on
health. Southern Medical Journal. 2002, 95:1429–1434.
outcomes in patients with bloodstream infection: randomized
(11) Hedges LV, Olkin I: Statistical Methods for Meta-Analysis. San
Diego, CA: Academic, 1985.
Volume 32, Number 1, 2006
Meta-Analysis of Prayer25
(12) HasselbladV,HedgesLV:Meta-analysisofscreeninganddiag- Download full-text
nostic tests. Psychological Bulletin. 1995, 17:167–178.
(13) Biostat: Comprehensive Meta-Analysis Version 2.2. Englewood,
NJ: Biostat, 2002.
(14) Hedges LV: Fixed effects models. In Cooper H, Hedges LV
(eds), The Handbook of Research Synthesis. New York: Russell
Sage Foundation, 1994, 285–300.
(15) Hedges LV, Pigott TD: The power of statistical tests in meta-
analysis. Psychological Methods. 2001, 6:203–217.
of the effects of remote intercessory prayer: Interactions with
personal beliefs on problem-specific outcomes and functional
status. The Journal of Alternative and Complementary Medi-
cine. 2004, 10:438–448.
(17) Lipsey MW, Wilson DB: Practical Meta-Analysis. Thousand
Oaks, CA: Sage, 2001.
(18) *Tloczynski J, Fritzsch S: Intercessory prayer in psychological
well-being: Using a multiple-baseline, across-subjects design.
Psychological Reports. 2002, 91:731–741.
(19) *Joyce CRB, Welldon RMC: The objective efficacy of prayer:
A double-blind clinical trial. Journal of Chronic Diseases.
(20) *Matthews DA, Marlowe SM, MacNutt FS: Effects of interces-
sory prayer on patients with rheumatoid arthritis. Southern
Medical Journal. 2000, 93:1177–1186.
(21) *Byrd RC: Positive therapeutic effects of intercessory prayer in
a coronary care unit population. Southern Medical Journal.
(22) Collipp PJ: The efficacy of prayer: A triple-blind study. Medi-
cal Times. 1969, 97:201–204.
(23) *Harris WS, Gowda M, Kolb JW, et al.: A randomized, con-
trolled trial of the effects of remote, intercessory prayer on out-
Internal Medicine. 1999, 159:2273–2278.
(24) *Walker SR, Tonigan JS, Miller WR, Comer S, Kalich L: Inter-
cessory prayer in the treatment of alcohol abuse and depend-
Medicine. 1996, 3:79–87.
(25) *Cha KY, Wirth DP: Does prayer influence the success of in vi-
tro fertilization-embryo transfer? The Journal of Reproductive
Medicine. 2001, 46:781–787.
(26) Guterman L: Author of disputed Columbia U. study on preg-
nancy and prayer pleads guilty to unrelated fraud charges. Re-
trieved March 15, 2005 from http://chronical.com/free/2004/
(27) Lyons JS: Research into prayer, fertility link now doubted. Re-
trieved January 18, 2005 from http://www.mercurynews.com/
(28) Chibnall JT, Jeral JM, Cerullo MA: Experiments on distant
intercessory prayer. Archives of Internal Medicine. 2001,
(29) Masters KS: Research on the healing power of distant interces-
sory prayer: Disconnect between science and faith. Journal of
Psychology and Theology. 2005, 33:268–277.
Other references included in the meta-analysis but not cited earlier
(30) *Aviles JM, Whelan SE, Hernke DA, et al.: Intercessory prayer
and cardiovascular disease progression in a coronary care unit
population: A randomized controlled trial. Mayo Clinical Pro-
ceedings. 2001, 76:1192–1198.
(31) *Mathai J, Bourne A: Pilot study investigating the effect of in-
Australian Psychiatry. 2004, 12:386–389.
(32) *Matthews WJ, Conti JM, Sireci SG: The effects of inter-
cessory prayer, positive visualization, and expectancy on the
well-being of kidney dialysis patients. Alternative Therapies in
Health and Medicine. 2001, 7:42–52.
(33) *O’Laoire S: An experimental study of the effects of distant in-
tercessory prayer on self-esteem, anxiety, and depression. Al-
ternative Therapies in Health and Medicine. 1996, 3:38–53.
26Masters et al.
Annals of Behavioral Medicine