Probability and Prayer

This is sort of a factual question, because I would like actual math, but it involves a subject of faith, so I’m putting it here. If it belongs in another forum, I humbly ask that it be moved accordingly.

I would like to know if a statistical probability can be assigned to a frequency of coincidence supporting the idea of answered prayer. That is, if someone maintains that a record of what they perceive to be “answers” to their personal prayers were set against a record of their “unanswered” prayers, the results would be statistically higher in favor of “answered” prayer than random coincidence would explain. Is there some magic number, like “20% of the time we can expect random coincidence to look like answerd prayer, therefore if >20% of my prayers are “answered”, it is a statistically significant indication that there is more than random coincidence involved.” ?

I’m not even sure if the way I’m asking this makes sense as I just think there is something wrong with this idea (assuming that answered prayer is a possibility - I’d rather stay away from that side of the argument), but I don’t know enough about statistics or probability to know what it might be.

It’s a bit more complicated than that. The universe of possible responses has to be considered, and what is to be considered an answer must be decided ahead of time. For instance, if you pray for someone to get well, does a three month remission count, or only a total cure?

Another problem is who is doing the praying. People tend to remember successes, so unless you limit people to praying for specific things, you might get them reporting prayers that succeeded and ignoring prayers that didn’t. There is also the problem of unfalsifiability - even if you don’t get a positive result, proponents of prayer can say that god won’t participate in an experiment like this.

Yeah, I figured it’s more complicated than that, which is why it smells all wrong to me. We’ve been through the unfalsifiability and the confimation bias to get to this point already. The person doing the praying is maintaining that keeping a such a record would eliminate confirmation bias and would allow that items she would tag as “answered negatively” would be, for statistical purposes, considered “unanswered.” I just don’t think there’s any meaningful way to assign probability to this sort of thing even under those conditions.

Prayer’s Power to Heal Strangers Is Examined (Washington Post story)

Well, first you have to determine what kind of prayers - freeform or ritualized? - by what people - Baptists, Unitarians, Muslims, Hindus, Mormons? - to what God(s) - Allah or Kali?

Leaving that aside, the OT has a nice test (of course it worked): some prophet or king wanted to know if he was on the right track with his plan of action, so he prayed to God to give him a sign by making a sheepskin placed outside overnight wet with water. The next morning, so and so many cups of water were wrong from the sheepskin, while the whole ground around it was dry.
Then, to be sure it wasn’t a freak natural phenomen, the prophet prayed again for the opposite, and the next morning, the whole ground was wet, and the sheepskin totally dry.

So that kind of test would be very convincing. But I don’t see a statistic base level - if this test (or a similar one) worked, esp. with testing the opposite to rule out normal condensation of dew on exposed cold surfaces, then every hit would count, me thinks.

Over a million people visit the shrine at Lourdes every year. This article describes the success rate.
http://www.theaustralian.news.com.au/story/0,20867,20009588-5002031,00.html

Extract below:
No less than 7000 miracles have been registered since counting began, but only 66 have been recognised as miraculous by the church and their stories are enshrined here. Eighty per cent of these miracles have been experienced by women, and most involve the curing of tuberculosis and multiple sclerosis. Of the 66, only 38 involved bathing in the Lourdes water. Six of those claiming miracles did not even go to Lourdes.

It’s more statistics than proability. You’d need to, ahead of time, decide on the number of experiments and how to determine the outcome of each. The person would have to do a lot of praying to make it significant. Lots of things don’t have yes/no responses - could you give some examples of things to be prayed for. A six coming up on a die would work well. Health, not so much.

If you solved those problems, there are well known methods that will tell you the probability that the results are due to chance.

groan I couldn’t even articulate the difference between “statistics” and “probability.” They are both large, thick, closed, dust-covered books to me.

The person I am discussing this with is not talking about running discrete experiments with dice or fleece, but about her personal experience of her prayer life. The kinds of things she offers as examples of answered prayer include:
A woman who wanted to get pregnant found out she had conceived.
A husband who had been abducted (!) returned safely home.
A dog who had run away in a large city was found and brought back.

So, random everyday stuff that believers who care about friends and family present to their God for intervention. She maintains that her “success rate” is around 90% and that coincidence could only account for about 20%. I think these numbers are PDOOHA. The keeping of a journal of such things, assuming it was honestly and scrupulously done, could eliminate the confirmation bias that conveniently forgets the “misses”, but I still think that the latitude for what constitutes an answer is pretty vague. Is that kind of construction even susceptible to the concept of probability? And how long would you have to do it for it to be statistically significant?

You know, I think before you explain difficult things like statistics and probability to her, you should explain the birds and bees to her. Because in most cases, it takes sex and not a prayer to become pregnant.

Had he been really abducted, or merely gone for cigarettes? And how much did the paying of ransom contribute to this?

And this is due to prayer and not the kindness of the person who found it? Or to the dog wearing a tag with the adress to bring it back to?

Honestly, I think you will have no luck convincing somebody with that outlook on life, and lacking of basic reasoning and common sense and how the world works.

If she’s convinced her prayer works so good (90%? please!), why doesn’t she want to do a test to convince you of the effectivness of her belief? If she thinks it’s disrespectful to ask God for these tricks, ask her why she doesn’t consider it disrespectful to treat God like a lackey to fulfil wishes when normal people can do these things just as well? (In other words, isn’t it bad to pray for a new car instead of working to get the money yourself?)

Even if 100% of the things asked for came to pass, you’d still have to be able to attribute it to god and not chance if you were to get an accurate statistic. If god continues to hide from those he supposedly loves, there’s no way to know if it was him who responded or if it was science or dumb luck that made your wish come true.

Once I asked my mom for \$5. She didn’t give it to me, but later that day I actually found \$5. That still doesn’t mean my mom gave it to me. It just means I found \$5.

I once heard the ultimate answer to “Why doesn’t God answer my prayers?” The explanation was that God DOES answer every single prayer. It’s just that, sometimes, the answer is “No.” And that “God moves in myterious ways.” So if you didn’t get what you asked for, it was for your own good, but nobody knows why. You can rationalize anything. God allowed the innocent child to die despite everyone’s prayers, so that others could appreciate life more, or gain inspiration, or some such thing. And of course, it helps if you pray for the “right” things, like strength, or hope, or courage, and not for the “wrong” things, like a million dollars or for your cow-orker to contract an embarrassing disease.

Yes, you CAN rationalize anything, but not when you’re talking about god and prayer.

Either we are misunderstanding each other or we’ll simply have to agree to disagree. Personally, I don’t think there is a superior being that influences our individual lives. If prayer helps someone to have strength, be a good person, feel comforted in grief, I’m not judging him/her. What I give a big :rolleyes: to, however, is the rationalization that causes folks to accept bad things without trying to do something positive. Or ignoring scientific evidence. Or confusing sequentiality with cause and effect. (“I prayed that I’d recover from my illness, and I got better, so therefore the prayer was effective.”)

Who said she doesn’t? She’s the one who offered to keep the journal to prove that prayer was so successful for her. I’m the one who’s questioning whether any numbers you got from such a “test” would be statistically significant.

Again, we’ve been through the unfalsifiability aspect of answered prayer. We agree that, ultimately, there is no way to tell whether it’s God or Chance that accounts for any results you get, you just gotta believe… She thinks that having numbers that are skewed significantly towards her desired outcomes is an indicator that it’s not Chance and that influences her willingness to believe. I’m not sure that’s even true, but if it were, how the heck do you come up with reliable numbers? There’s a way to determine the probability on a coin toss. What’s the way you determine the probability on finding your dog or getting a raise? It seems fundamentally off to me.

In one word, no, the numbers wouldn’t be any use at all. So you have to tell her that either she agrees to a scientific, real test - like a coin toss - or she failed the test, because the journal doesn’t prove anything.

Though with the woolly, wish-fulfilling “reasoning” you gave earlier, she may have problems to understand this. Just be adamant about it.

Just as background, in science the 5% level is generally regarded as being statistically significant, and 1% as being highly significant.

That is, there must be less than 1 chance in 20 that a result is due to a proposed experimental variable instead of to random chance to be even minimally acceptable. It takes a 1 in 100 chance to be considered really well demonstrated.

But to apply this, you would have know what the odds are of any particular outcome in the absence of the experimental variable, and you need to have a sufficiently large number of cases. For example, you would have compare the recovery rate of (let’s say) 100 lost dogs that were not prayed for, vs 100 that were, in order to get valid results.

In any case, the different “experiments” you describe probably vary widely in the basic probablity, so that it is impossible to lump them. You would have to consider them individually. If a woman is of normal fertility and is trying to get pregnant, the probablity is probably quite high that she will given a reasonable amount of time. The eventual recovery of a lost dog that had a tag would also be quite probable. On the other hand, remission of a highly advanced cancer would be improbable.

Ah, I think that’s what’s bothering me. She wants to consider her whole prayer life in a lump, and apply the statistic to “how often I get what I ask for” no matter how likely or unlikely any individual result might be. I don’t think there’s a valid metric for “how likely is it that I will get what I ask for?”