Efficacy of prayer: A demographic analysis of prayer for healing

Praying handsFirst, a clarification. I am not evaluating healing that results from an interaction with an individual that claims to have the gift of healing. To this end, when I say “prayer for healing” I intend it to mean prayer that is offered up by everyday people that requests divine intervention to heal a physical ailment. This does not include the laying on of hands, slapping people in the forehead, etc… Consideration for the gift of healing will be addressed elsewhere.

If you do some research to find evidence of the efficacy of prayer for healing you will undoubtedly encounter the numerous attempts to test this through scientific studies, which yield widely varying results that are selectively praised and dismissed by both sides. You’ll also encounter several objections to those studies:

  1. God wants people who are compelled to true love, which is grounded in faith, not scientific evidence. As such, God may withhold interceding in the context of a scientific study.
  2. The compulsory prayers offered up in a study are not as genuine as unsolicited prayer.
  3. The patients typically have friends and family that are praying separate from the study, which invalidates the control arm of the study.
  4. Patients who are aware that they are receiving prayer may alter their behavior or attitude accordingly, which affects their outcome.

Given these objections and the controversy that surrounds those studies, I’m electing to not rehash that content but rather to look for alternative input into this topic. In particular, I’m suggesting that one way we can look for evidence of healing is to compare mortality rates for a relatively indiscriminate ailment in populations that are distinctly divergent in their faith but with similar physiology and treatment options. In this way we can avoid most of the issues associated with the conduct of a particular study. To do this analysis, we look at one of the most indiscriminate of ailments, cancer. To further avoid physiological bias, we will look only at colorectal cancer, which has the lowest variance between racial groups. I will also look at female breast cancer, which is one of the only cancers that affects whites more than blacks.

What do we know?

The tables below present the mortality per incidence (MPI) for colorectal cancer and mortality per incidence for female breast cancer, partitioned by the five most and least prayerful states in the US.

Data for the five most prayerful states
State Prays Daily Colorectal Cancer MPI Breast Cancer MPI
Percent Rank (46) Percent Rank (50) Percent Rank (50)
Mississippi 77% 1st 38.8% 15th 20.8% 4th
Louisiana 76% 2nd 37.7% 20th 20.5% 7th
Alabama 73% 3rd 37.4% 21st 18.3% 24th
South Carolina 72% 4th 40.8% 5th 18.6% 20th
Kentucky 70% 5th 36.0% 33rd 19.3% 13th
Data for the five least prayerful states
State Prays Daily Colorectal Cancer MPI Breast Cancer MPI
Percent Rank (46) Percent Rank (50) Percent Rank (50)
Maine 40% 46th 32.1% 47th 17.5% 32th
Massachusetts 41% 45th 37.2% 26th 17.0% 34th
Alaska 41% 44th 31.0% 48th 23.2% 1st
New Hampshire & Vermont* 43% 43rd 32.9% 47th 13.8% 49th
Connecticut & Rhode Island* 47% 42nd 31.7% 48th 15.5% 43rd

* These states were combined in the prayer study. The cancer statistics use a weighted average based on population size to derive the mortality per incidence, which is then translated into a ranking.

I did not know what to expect before I looked at the numbers but I will admit that this was a bit unexpected. Not only is there not a link between population prayerfulness and lower mortality per incidence, but the relationship is actually reversed. These numbers clearly show that the mortality per incidence rate is almost always lower in the less prayerful states. This warrants some further discussion, which I’ll cover in the interpretations.

A couple additional points about this data:

  1. The better statistic would be something like the 5-year survival rate, or even better, a remission rate. I was unable to locate data which provided those statistics within geographic partitions. If anybody out there can summon that data, I’ll gladly accept the contribution. Regardless, the statistic I use here, mortality per incidence, looks at the death rate (cancer deaths per 100,000) over the incidence rate (cancer diagnoses per 100,000). I content that, assuming there are not dramatic changes in the cancer statistics from one year to the next, this provides a reasonable estimate of the percentage of cancer patients who end up dying as a result.
  2. The prayerfulness study is likely not reflective of the percentage of people who would pray for somebody that they know has cancer. That would certainly be skewed to much higher levels in all populations. However, the relative differences are substantial enough (78% higher in the prayerful states) that it should still give a good indicator of the relative likelihood that any particular cancer patient is being prayed for. That is, patients in the most prayerful states are almost certainly more likely to be prayed for than the patients in the least prayerful states.
  3. Hey, Alaska – it looks like you might want to investigate why you suck at treating breast cancer but are good at treating other cancers.

What is the Christian interpretation of the information?

The mortality per incidence analysis appears to be a poor result for one who believes in the power of prayer for healing. However, I also foresee a couple ways that the result could be interpreted to more closely fit the Christian worldview:

  1. In Christianity, death is the transition from being trapped in an imperfect, temporary body to being joined with God for eternity. In that sense, death is not a bad thing. In fact, this view could even be used to suppose an expectation that mortality rates could be higher in the more religious states, as was found to be the case. God would be rescuing these individuals from the pain that was being inflicted by their imperfect body.
  2. The apparent lack of effectiveness of prayer was also put in a positive light in a Christianity Today article which focused on the results of the STEP study. In short, the article suggests that Christians should be glad that God is not withholding his aid to those who are not receiving prayer. That is, God is being fair because He loves everybody equally and this is consistent with his nature.

Within the context of a Christian worldview, I would not consider either of these responses to be invalid, though they’re not very satisfactory. They also raise a significant question – if either of these are the proper interpretation then why pray for healing? This led me to look into why this is an accepted practice in Christianity. In short, the biblical case for healing through prayer looks to me to be weaker than one might expect. In fact, I was only able to locate one verse that specifically links prayer with healing, James 5:14-15

“5:14 Is anyone among you ill? He should summon the elders of the church, and they should pray for him and anoint him with oil in the name of the Lord. 5:15 And the prayer of faith will save the one who is sick and the Lord will raise him up – and if he has committed sins, he will be forgiven.”

This is a pretty clear statement, but it is also carries hints of what we might call “the gift of healing” rather than the more common prayer for healing that I’m interested in here. Also, some would interpret this verse to speak to spiritual healing (focusing on the forgiveness part – see one, two, three). There is a strong indication throughout the rest of the New Testament, and in particular in Jesus’ ministry, that physical healing is primarily a means to spiritual healing. The theme is that God is really only concerned with our soul, and so physical healing is only valuable if it serves a spiritual purpose – that is, toward securing somebody’s salvation. A sampling of the indicators for this is below:

  • Jesus used physical healing as confirmation of his authority to forgive sin (Mark 2:5-12)
  • Jesus stated that some afflictions existed so that he could use physical healing to demonstrate God’s power (John 9:1-7, 11:4)
  • The gospels often record Jesus’ affirmation of somebody’s faith as part of the act of physical healing (John 4:47-53, Matt 8:13, 9:22, 9:29, 15:28, Luke 17:19, 18:42)
  • Jesus suggested that self-mutilation (anti-healing, if you will) would be preferable to losing one’s soul (Matt 5:29-30)

This leads me to suggest a third interpretation of the apparent lack of efficacy of prayer for healing:

  1. God’s primary, and perhaps only, purpose for physical healing is to bring spiritual healing. In that case we would expect healing to be relatively more frequent in the less religious areas (as was found in the analysis). This also points toward a form of indifference toward the need for physical healing when salvation is not at stake, which would tend to make it a somewhat infrequent occurrence.

This appears to me to be the most biblical and consistent Christian interpretation of the apparent lack of efficacy of prayer for healing in the prior analysis.

What is the naturalistic interpretation of the information?

A naturalist should also be a bit surprised by the analysis. In the absence of other factors, a naturalist would expect a lack of correlation between prayerfulness and mortality per incidence. A negative correlation is troubling, though perhaps not as much as a positive correlation. As such, a naturalist would seek out other explanations for the reverse relationship.

Hypothesis #1: Poverty rates

One can’t help but notice that, at a glance, the most prayerful states would seem to map to some of the poorer states and the least prayerful states would seem to map to some of the richer states. If this is the case, then that could reasonably contribute to the observed discrepancy due to the fact that the poorer population is less likely to obtain supplemental care beyond that which is covered by insurance, and are also more likely to have no insurance at all, or lower quality insurance. It’s also reasonable to expect that the health care facilities in more affluent areas have more resources that enable better care. Any or all of these would impact the care that patients receive. The ranks for the median household income of the most prayerful states are Mississippi (50th), Louisiana (41st), Alabama (46th), South Carolina (40th) and Kentucky (47th). The ranks for the median household income of the least prayerful states are Maine (36th), Massachusetts (6th), Alaska (4th), New Hampshire/Vermont (7th/20th) and Connecticut/Rhode Island (3rd/18th). This supports the notion that the states which are most prayerful are also definitively less wealthy and demonstrates feasibility for the hypothesis that the higher mortality per incidence may be at least partially explained as a function of wealth.

Hypothesis #2: Those with faith are less likely to fight against death

As noted in the Christian interpretation, a Christian sees death as a transition and, though they may fear death, their belief in an afterlife is likely to dampen that fear relative to a non-believer. Fear is an incredible motivator and if somebody is more fearful of death, they will likely put more effort into avoiding it. This may offer additional insight into why the mortality per incidence rates are lower for the least prayerful states, though I’m not sure how to provide evidence for this. Even so, it is not an unrealistic hypothesis to help explain the observed discrepancy.

Which interpretation seems more probable?

If it were theologically evident that Christians should expect God to intercede in response to prayers for healing then I would identify the naturalistic view as being far more probable. However, it seems likely to me that the popularized view of prayer for healing may not be in line with biblical theology and that a prayer which is not focused on spiritual matters is not well aligned to God’s will. Despite this, it seems that Christians are extremely reluctant to espouse views that might in any way indicate that we shouldn’t pray for physical healing. I get the sense that this is for fear of positing a God that is either incapable of healing or a God who is not moved with compassion at our physical suffering (either of which would clearly contradict Jesus’ ministry). Similarly, I see the athiest position as one which misrepresents the Christian view by over-emphasizing the expectation of healing as a result of prayer. What we get then is that the most common views of prayer for healing come from either the traditional Christian view, which errors on the side of not undermining God’s power, or from the athiest view, which exaggerates a God who is dutifully obligated to respond to all prayers. I think that both of these may have missed the mark.

I should also point out that if the Christian view I presented above is true, where healing only occurs to bring about faith then, to be blunt, the majority of prayers for healing can be considered to be pointless and misguided. This would also underscore the division and differences within the church as to the proper application of prayer. This conflict in views and lack of agreement counts against the acceptance of the Christian interpretations I provided.

In the end, I find that both worldviews can present reasonable interpretations of the data but the naturalistic explanation is more easily supported and does not carry with it the kinds of conflicts that are present in the Christian interpretation. This leads me to assign the probabilities as follows:


Lastly, as this is my first topical post, I also need to point out that I do not intend these posts to be static. As discussions arise and new information is gathered I fully expect to update the posts and even change the probabilities I have assigned.


4 thoughts on “Efficacy of prayer: A demographic analysis of prayer for healing

  1. Hi Travis,

    I think most attempts at scientific studies of prayer are based on a poor understanding of prayer. Science deals with the predictable, physical mechanisms and the like. When science deals with people, it has to do so statistically because people are not so predictable and the mechanisms are not so clear. So how to apply this to what is essentially a omnipotent omniscient person deciding how to respond to human requests?

    The most common approach I have seen runs scientific tests on paired groups of people who are recovering from illness. One group is prayed for, one isn’t and the recovery rates are compared. That experimental design is really treating God as a psychological belief and seeing whether that belief makes any difference (just like it might test if having a loving spouse makes any difference). It’s hardly testing for miraculous answers to prayer, which are unlikely to occur in such situations.

    I have found as many such studies as I can (see Can prayer assist healing? and the results were 2:1 in favour of a small effect, for what that’s worth.

    I have never seen the study that you reference, but i think it is even dodgier than the ones I’ve just mentioned. Scientific studies try to eliminate all the extraneous variables, so the variable of interest can be tested. But from a quick glance, it seems doubtful that this study has done that. For example, what if people who were poor or sick were more likely to come from certain states or more likely to be religious or ask for prayer? What if socioeconomic factors were the main reason for any difference? Why correlate cancer mortality with religious belief, why not correlate it with hair colour or political views?

    So I think the study tells us something, but it must be a very weak conclusion.

    I think the only valid way to attempt to measure God’s action in healing is to examine cases of actual unusual recoveries and see whether they can be easily explained naturally, or not. If we find enough that cannot be explained, then the probability that it is God increases.

    • Dodgy, eh? The “study” here is just my own personal analysis using broad demographic data. That might explain its dodginess. Even so, some of the questions you asked were discussed in the post and some don’t seem relevant. In particular, note that there isn’t a study population, so religious belief and asking for prayer aren’t factors. It’s simply asking whether there is any correlation between locations which tend to pray more and the mortality from cancer.

      Regardless, thanks for the link again. You’ve compiled some good data that I’ll have to look at.

      • Sorry Travis, I didn’t realise you did the work yourself, and I certainly didn’t mean to be rude. I think it is good that you had a go at things in this way. But my other comments still seem to be true. There could be all sorts of cross correlations at work here, and you haven’t accounted for them.

        I like to analyse data like that too, so I wonder if there is anything you could do to analyse other factors? I wonder whether you could test the states for other factors which seem to correlate with the cancers? If you found other factors, it would show that prayer wasn’t necessarily important, but if you tested a few other factors and nothing else correlated, then you could at least say that prayer was the main factor of the ones you tested. The more others you tested, the stronger your conclusion would be. Just an idea.

        Anyway, sorry if I was rude. I woke up at 4:30 am this morning to attend an Anzac Day dawn service, and my head is even moire woolly than usual right now. I’ll claim that as my excuse rather then something endemic! 🙂

      • Eric,
        Don’t sweat it. If I has tried to pass this off as an actual study I would hope that somebody would cry foul. Just to be clear, the discussion did not speculate that prayer itself was somehow causative in the negative correlation. Wealth seems like the most likely causative factor to me.

What do you think?

Loading Disqus Comments ...