901. Sue Blackmore debates Alister McGrath

Comment #149152 by Bonzai on March 25, 2008 at 7:35 am

Dr. Benway

I don't think atheists have a problem with scholarship. They have a problem with thinking about the book as anything more than literature. They criticize the scholars for trying to have things both ways.

Then criticize the magic when it comes up. I have no problem with atheists calling believers out on what they actually believe, but I do have a problem with atheists telling believers what they should believe based on a very shallow way of reading ancient literature, which is essentially like a know nothing fundamentalist Bible thumpers (or "bashers" in the U.K) in the deep South.

epeeist

We have had a similar attempt at discussion with Artful Dodger. What is literal, what is metaphorical, how do you tell the difference between the two and how is the authority to declare which is which granted? He has done his usual post-and-run at this point. You have added another - what does the metaphor mean?

Those are all fair questions that an honest believer would have to answer. But that is quite different from saying that since there are ambiguities the only "authentic" Christianity is to take all words as literal truth, It is dumb and ridiculous. Supposedly rational and informed critics of religion should know better,

Thinking theists are happy to declare Genesis symbolic until one asks what therefore did Jesus die for, at which point Adam and Eve seem to acquire some level of literalness again.They want to eat their cake and have it.

Do you know for sure that the whole of Genesis was meant to be literal? In Hebrew "Adam" simply means "man" and I was told by some Jews that the ancient Jews didn't take the story of the Garden of Eden literally. Perhaps someone more knowledgeable in Judaism can shed some lights on it.

Your point seems to be that people used to take everything literally until science showed that literalism is untenable and they shift to more symbolic interpretations to avoid being nailed down. I am sure that happened but I don't think you can claim that is a general rule,--not as a bald assertion without better evidence from religious history anyway. It is not a self evident claim.

steve

It is. The sophisticated believers have more to explain. They have "New Bible with added 'Interpretation' - helps wash away the nasty bits".

Yes, they have things to explain and sometimes they do and sometimes they fudge. I am oppose to the categorical statement that only a literal interpretation is authentic and everything else is "fudging".This is naive and shallow,

Al

The "part describing the whole" being one example. These are easily identified, and are generally understandable to someone with even a modest understanding of semitic languages.

That would exclude most Bible thumpers.

So to say that these are cryptic in a way to make them indecipherable to translators or scholars, is a bit of a stretch.

Even translation from ancient Chinese to modern Chinese is not trivial, a lot is lost in translation, let alone translating into a different language. Maybe there is more continuation ancient in ME language so that their ancient forms are sufficiently similar to their modern descendants, I don't know but it is not intuitively obvious.

But even if you know to tease out all the metaphores from the literal,--based on our reconstruction of linguistic history, there is still an unspoken context, is it supposed to be an eternal command, or is it situational thing, etc?

I see no reason why an "authentic believer" must assume that all verses in the bible are universal and literal like reading instruction manuals,

902. Lying for Jesus?

Comment #149096 by Bonzai on March 25, 2008 at 5:51 am

bibanu

You are comparing things (e.g. dynamite) and their use with ideas/theories and their consequences. It is not the same thing. THEORIES have consequences - and I think that Coulter shows a nice and believable sequence there (and note the German scholar who she is quoting - another Richard :)).

Well, people can't be held responsible if their ideas are being bastardized, Darwin should be held responsible for eugenics no more than Einstein should be blamed for post modernism because of the idiotic slogan "everything is relative".

However, I do think we should extend the same considerations to the other side, to blame Christianity for Hitler is just as flimsy.

903. Lying for Jesus?

Comment #149092 by Bonzai on March 25, 2008 at 5:44 am

So Hitler believed in Thor.

BTW, not call Creationists are Christians or even followers of the Abrahamic faiths, Many pagans are creationists, so were most pre-Christians. They have different creation myths.

904. Sue Blackmore debates Alister McGrath

Comment #149080 by Bonzai on March 25, 2008 at 5:08 am

Steve,

But to claim that the source book is all true is a more consistent attitude than to knowingly fudge things and try and claim divine authority for that fudgeing.

Being "true" and being "literally true" are two separate matters, If I say someone was caught lying and his pants are on fire, that would be "true" if he was indeed caught and was embarrassed for the lie, even though his pants are not literally "on fire".

Ancient Middle Eastern languages were not direct and literal like English, they used a lot allusions and metaphores in a way that were weaved into normal speech seamlessly. It is not like in contemporary English where you can tell relatively easily which is which. Aside from the fact that English is a relatively straight forward language, the ease in parsing is partly due to an unspoken shared cultural references. To decipher what Biblical passages meant to the contemporary audience would involve a lot of linguistic and anthropological forensic work, which is the subject of Biblical scholarship.

I am not saying all moderate believers study Biblical scholarship but to say that the only consistent way to believe is to take the whole book word for word in translation is simply naive.

During the cultural revolution, the French intellectual André Malraux visited Mao, he asked Mao to describe the situation in China. Mao said, "it was like a Buddhist Monk carrying an umbrella." Malraux thought that was very romantic, he returned to France and wrote articles on the romantic image. When the Chinese read his articles, they laughed their pants off. You see, in Chinese "hair" ryhmes with "law" and "heaven" is a traditional reference for higher power, somewhat like "God" minus the personality. A Buddhist monk is shaved, he has no hair, hence no law, the umbrella blocks off the sky, so no heaven either. Mao's words means "total lawlessness and chaos".

Now there were many Chinese speakers in France who could have told Malraux that if he cared to ask and his translator was obviously not very competent. But at least he did know it was a parable, only he screwed it up.

Now you are telling me that one can be justifiable confident that he can figure out the intended meanings of the Bible by reading an English translation thousands of years later, in a naive literalist approach. Sorry, I think that is insanity talking even for a true believer,

By insisting one can only read the bible like a fundamentalist the atheist critique comes off as naive and shallow for the sophisticated believers. It is not "fudging" to acknowledge that language is complex.

P.S. In case the readers are wondering, the pants didn't really fall off for the Chinese readers of Malraux, it was just figuratively speaking.

905. Sue Blackmore debates Alister McGrath

Comment #149071 by Bonzai on March 25, 2008 at 4:35 am

I actually agree with the fundamentalists on this point (never thought I'd say that!). I have more 'respect' intellectually speaking for the fundamentalists in this regard. The fundies, highly deluded though they are, at least have what they regard to be evidence for their beliefs; i.e. the bible

Yeah, "fundamentalists" reading a translated Bible as if Jesus spoke English, what a joke. At least Muslims read the Quran in Arabic.

You may have more sympathy for the fundamentalists because you share their shallow, "born again" mindsets, only different "religion".

I am sick of idiotic atheists who insist that the only consistent believer is the fundamentalist. Their fucking book is not consistent internally if you take the naive, literalist approach and it needs to be interpreted, it has always been the case and that's what theology is about, at least a large part of it.There is no scriptural basis to say that "God" is an literalist even for the believers.


Even if you disagree with something you don't set up a strawman and then shoot it down to declare victory.

If you want to criticize religion at least try to understand it first.

End of rant,

906. The science of religion: Where angels no longer fear to tread

Comment #148912 by Bonzai on March 24, 2008 at 2:55 pm

It is called Buddhism.

Well as you know there are many strands of Buddhism, some do have gods.

I don't think the atheistic strands of Buddhism or something like Taoism (also atheistic in its pure form)would be able to take the place of God beliefs for all. There are Christians and even Muslims who do practise some kind of Buddhist meditation and read Taoist philosophy. But somehow that is not enough to fill the "God hole" in their brains. What does that feel like? I don't know, it is qualia ,can't be described.

907. Lying for Jesus?

Comment #148909 by Bonzai on March 24, 2008 at 2:44 pm

Norman,

Do you really think Richard Steigmann-Gall's "The Holy Reich: Nazi Conceptions of Christianity, 1919-1945" is merely desperate and simplistic?

Well what did he say? Do tell us, you, however do look desperate and simplistic with questions such as this:

Is there another kind [of Christianity]?

What does Edwin Black, "War against the weak" actually say?

Edwin Black provided documented evidence, including Hitler's private correspondence with American eugenicists that proves convincingly that Hitler was inspired by "science" in his "final solution". Look up the book, or find a review online ("science" is in quotes because it was a bastardized form of "Darwinism". It was actually a pseudoscience and an ideology)

Did I ever say otherwise? Only Hitler said otherwise and I quoted him. I didn't even mention Martin Luther's anti-semitism like Richard did.

Invoking Martin Luther is quite "flimsy" indeed, for Hitler looked at Jews in racial, rather than religious terms. If religion was the motivation he could have just forced the Jews to convert.

Also, Germany was actually one of the LEAST anti-semitic country in Europe before the Nazi took over, it is untenable to depict Nazi genocide against Jews as a logical extension of old world anti-semitism.

So, direct quotes from Hitler are flimsy?

If some guy told me "I am Napoleon", a direct quote, does it follow that he was really Napoleon?

Any one with some minimal historical knowledge and common sense would know quotes mean nothing if cited like bumper sticker slogans without the proper context. Since obviously you are not stupid you must think that we are.

908. The science of religion: Where angels no longer fear to tread

Comment #148902 by Bonzai on March 24, 2008 at 2:21 pm

I think it is the other way around. Religion may survive because of the sense of an inner life, but that is nothing to do with the idea of God.

That "otherworldiness" is the source of a lot of "spiritual experience" according to believers. It is the way they "feel" God and the "transcendence". Hence my comment. I think it is entirely possible to have belief in some God without formal dogmas and Churches, just look at the so called neo pagans and freelance believers who don't subscribe to any religion.

On the other hand I can't see how religion (understood conventionally) can survive without God.

909. It looks like Man crucified

Comment #148900 by Bonzai on March 24, 2008 at 2:15 pm

That was not what I was hoped you would see. I had hoped that you would notice what the creationists were up to

A "war" with creationism is not the same as some grand overarching war against any form of religious belief in the name of some "rational" utopia.

I sometimes have that feeling while reading this site, it is as if there is some kind of crusade going on by the "rational" zealots where many of whom,--not all,--mostly succeed only in setting up strawmen. The argument boils down to "if you say you're a Christian you must believe in what I say you believe in or you're just lying or being evasive." And usually "what I say you must believe in" happens to be the most crass forms of Biblical literalism only found in the Southern U.S. and very few places outside.

I can't speak for Spinoza, but it is probably the same thing that he was thinking about regarding the silly "war" rhetorics. The first casualty of any war is usually truth, we should bear that in mind if it is a war that we want to wage. I find very often the caricature of religious people is laughably simplistic, not very different from the caricature of atheists from "the other side" in crudeness.

P.S. This site is painfully slow at my end. So you'll have to forgive me for the typos and mistakes, I won't bother to edit and repost.

910. The science of religion: Where angels no longer fear to tread

Comment #148887 by Bonzai on March 24, 2008 at 1:23 pm

I think some kind of "God" will always survive just because we do experience an inner life. There is this "other wordiness" which arises from this awareness, (Belief in God is not the same as religion, which is a social institution)

This is the God of the qualia.

Science and philosophy would not be persuasive if they try to tell people that their sense of self and inner experience don't exist and that people are just deterministically programmed meat puppets.

Even if this is true,--which I doubt,--we would have to invent some kind of illusions to cope, either explicitly or implicitly by simply ignoring that part of science. "Truth" is *meaningful* only to intentional agents that can ascribe and grasp meanings, if "truth" says intention is an illusion, meanings will become void as well, in that event "truth" would mean nothing.

911. It looks like Man crucified

Comment #148881 by Bonzai on March 24, 2008 at 12:52 pm

Spinoza,

Good posts!

Mitchell Gilks,

Let me ask you, do you think that laymen can be intellectuals, while academics can not be?

This is not directed at me, but I do have an opinion, probably very close to yours.

I think being an academic and being an "intellectual" in the true sense are two different things. I know many academics, many of them are very tunneled vision people who don't know anything outside their narrow fields, worse, many don't even have the slightest intellectual curiosity. I think the publish or perish culture of the academe has much to do with it. I know plenty of people with Ph.Ds and I would be very ashamed to call myself a "doctor" in anything if I were like them, they don't even have a passing knowledge about major subareas in their own field outside their areas of specialization.

912. Biology prof expelled from screening of 'Expelled'

Comment #148877 by Bonzai on March 24, 2008 at 12:38 pm

So PZ got kicked out of a screening because they thought he was a trouble maker up to no good. It back fired and now they have even a bigger PR problem and try to spin themselves out of it.

That should be the end of the story.

How many threads do we want to make out of this? How many more posts do we want to read about what a dishonest scumbag Mathis is? It is tempest in a teapot, really, maybe involving a couple of bruised egos.

Snap out of it.

913. Lying for Jesus?

Comment #148871 by Bonzai on March 24, 2008 at 12:25 pm

Hitler was no more motivated by Christianity than Stalin was motivated by atheism. I agree with Mphil on this completely. It is unnecessary to try to pin Hitler's atrocities on Christianity. It just make the people who do that look desperate and simplistic,

Even if Hitler did consider himself a good Christian, which I strongly doubt, it has to be a very distorted and deformed kind of Christianity, which means the example of Hilter lends absolutely no force to the critique of Christianity in general.

Indeed, if one insists on latching on to such flimsy connections to argue that Hitler was somehow the product of Christianity, there is a much stronger case to link him to Darwin. (for those who don't understand the logic of conditional statement, I am not agreeing that there was such connection) After all he created his genocidal program after American eugenics rather than the Bible (ref:Edwin Black, "War against the weak")

914. The science of religion: Where angels no longer fear to tread

Comment #148661 by Bonzai on March 23, 2008 at 2:38 pm

42nd

Not exactly, because your goal is still your own survival and no one else's (at least at genetic and unconscious level), you only drag others because you have to. In prisoner's dilemma, each prisoner is trying to get his own arse out of trouble, not caring about the other one, but they still end up cooperating.

But then they seem to be just different levels of descriptions, like thermodynamics and statistical mechanics. "Groups" do exhibit cohesion and from what you are saying they do get selected as units. Whatever the "goals" of individuals may be, it doesn't invalidate the group level description.

Besides, if you are right we would expect a lot more traitors in wars and conflicts when working with a superior enemy often seems to provide a better chance for individual survival and gene propagation, But there are strong group incentives against such actions. You can of course argue in some convoluted way that group loyalty is better for individual survival at normal time, but wouldn't it be simpler to describe the group dynamics using group variables instead? I don't know.

Moreover, if you do reductionism all the way down, whatever beneficial to individuals may not be beneficial in terms of gene propagation, Example technically advanced societies provide better chances for individual survival but they also also have less children. It is not an coincidence either.

EDIT And why stop at genes?. The late physicist Heinz Pagels wrote in his book "the dream of reason" that Dawkins had it all wrong, the genes were themselves only the play things of DNAs, I don't remember how the argument went, but will look it up when I have a chance.

915. Does God answer prayer? ASU research says 'yes'

Comment #148656 by Bonzai on March 23, 2008 at 2:30 pm

Yes. The test MAY be powerful enough to show significance. If the difference is there.

There will almost always be differences because our models are only approximations. I said "almost" just as a technicality to cover myself, in fact I would be tempted to say "always".

You are attempting to fit a specific model to the data. This is a form of "difference hunting". You should really be testing for a specific difference, within certain confidence limits, not searching for any difference.

Searching for "any difference" is testing for a specific difference = 0. So in principle it is the same (in many cases assumed say, a normal distribution)

However, you have diverted from the hypothesis I was considering. It is that the rate of decay is not exponential, not some specific exponential.

But I was clear that I was talking about parametric tests. It was you who changed the question.

f you pick ever more sample data, you will be able to fit an exponential curve with ever greater statistical significance. Increasing the sample data will (virtually) never show less fit.

Yes, but then you may overfit by capturing too much noise. Significance alone doesn't tell you whether over fitting occurs. I mentioned a regression model with a 100 parameters and a R^2 close to one to make the point earlier.

916. The science of religion: Where angels no longer fear to tread

Comment #148649 by Bonzai on March 23, 2008 at 2:13 pm

yeah, but how is it a "misfire"?

917. The science of religion: Where angels no longer fear to tread

Comment #148643 by Bonzai on March 23, 2008 at 1:49 pm

It's called game theory. Check prisoner's dilemma for more details of this http://en.wikipedia.org/wiki/Prisoner's_dilemma

If it is "rational" thing,--albeit in an unconscious level,--how does it fit in with Dawkins' theory of "selfish genes misfiring"?

So yeah, basically if it takes 10 people to kill a mammoth (and nothing smaller is available for hunting) , then they need to work together even if they don't really give a damn about each other.

So that enhances the chance of survival of the group v.s other groups where members don't cooperate. Maybe I am naive like beelzebub, but that sounds like group selection to me.

918. Does God answer prayer? ASU research says 'yes'

Comment #148642 by Bonzai on March 23, 2008 at 1:49 pm

There is not more likelihood of a statistically significant departure from exponential decay rate if you follow the decay of 100,000 atoms than if you follow 1,000 atoms.

No, you are talking about something different.

That's why I asked you to be specific about your hypothesis to be tested.

Let's say we assume it is exponential decay, the number of decays follows Poisson distribution. But we don't know what the decay rate is, that is what we want to find out. This is a typical set up for a parametric test.

Let's say your model predicts a decay rate of 1.78, in whatever unit and you want to test that as your null hypothesis.

In reality, the rate may actually be 1.78012222. It has nothing to do with sample fluctuations or anything like that, it is just that your model doesn't specify enough number of decimal places. It is a "model error"

Now if you test with 200 nuclei you may very well get a p-value above alpha,--say you pre-set it to 1%,-- because the power of the test is not able to detect the difference of 0.0012222. But if you use a sample of 20000 nuclei, the test may be powerful enough to yield a p-value below alpha, thus showing significance.

The fact is in almost all real world situations our models are going to be a bit off because at some points approximations would have to be made, If you have a large enough sample, the test would be able to detect whatever tiny errors in the model and yield significant results. Thus, in principle you can show any effect to be significant by a large enough sample.

If the test detects a difference of epsilon, then the samples are statistically different given the parameters you have specified. But, if they aren't statistically different, you won't detect that difference.

My point has nothing to do with sample fluctuation, but the power of statistical tests in relation to sample size. See above.

919. Does God answer prayer? ASU research says 'yes'

Comment #148633 by Bonzai on March 23, 2008 at 12:51 pm

You are wrong. The p-value will only be less than or equal to alpha, showing statistical significance if there is a statistically significant difference You can't just find statistically significant difference by increasing sample size.

I just showed you that you can.

What is a statistical significance? It basically says that certain effect is not "random" due to the quirks of the sample. How do you decide that? Look at the p-value, you agree beforehand that if the p-value is less than certain threshold (alpha) then you decide that it is too unlikely to be chance, with the provision that you may be wrong with probability alpha (type I error) Behind this "how" part there are a whole lot of technical assumptions.

I showed with your T-test example that you can, by simply appropriately inflating the sample to get statistical significance for any pre-set alpha. It is a technical point which arises because of the mathematical formalism. Whether something is "truly" significant in a "philosophical sense" or just technically appears to be so because of a small p-value are not two identical questions and you are confusing the two.

If you insist that they are the same question, you have to provide a proof or a demonstration of the equivalence, which I don't think you can.

Really must run.

920. Does God answer prayer? ASU research says 'yes'

Comment #148621 by Bonzai on March 23, 2008 at 12:18 pm

You implied this, in my view, when you said that any specific hypothesis could be shown to be true by increasing sample size. The only way this is possible is if increasing sample size somehow produced statistical significance out of thin air.

No, I certainly did not imply that, in ANY statistical analysis alpha is preset, this is really common sense to anyone who has some knowledge in the area.

I am saying with any preset alpha (and beta)you can increase the sample size so that your p-value would turn out to be less than or equal to alpha, hence showing statistical significance, I have explained it very clearly with your t- test example. If you still don't get it, I think you just want to argue for argument's sake.

I am sorry, I got work to do, See you later.

921. Does God answer prayer? ASU research says 'yes'

Comment #148618 by Bonzai on March 23, 2008 at 12:10 pm

Arbitrary statistically significant difference.

Arbitrary, but pre-set. So you set it to 5%,--or whatever. But it stays the same for the whole investigation, you can 't
have a single test which is significant for ALL alphas.

922. Does God answer prayer? ASU research says 'yes'

Comment #148616 by Bonzai on March 23, 2008 at 12:05 pm

In that case you are even more wrong

arge samples don't "invent differences". They reveal them. Or not. That is the point

What? When did I say large samples "invent" differences? I said "DETECT". Alpha is ALWAYS pre-set, most people would set it to say, 5%.

A 10% chance fluctuation in a sample of 100 is reasonable to expect. A 10% chance fluctuation in a sample of 10,000 is far less likely.

The hypotheses pertain to the POPULATION, not the sample. So you're correct, but answering a wrong question,

923. Does God answer prayer? ASU research says 'yes'

Comment #148611 by Bonzai on March 23, 2008 at 11:57 am

Looking for a specific difference is not the point. It is about whether that difference matters.

No, you are missing the point. It is not "looking for difference", but to demonstrate that you sample size can be adjusted to detect arbitrary difference.

Now you don't have to consciously pick out an epsilon or find a sample size for that,--though you do sometimes because you do want your sample size to be large enough to tell, say, a 10% difference,-- but the test will pick up the difference automatically if the sample happens to be very large.

924. Does God answer prayer? ASU research says 'yes'

Comment #148608 by Bonzai on March 23, 2008 at 11:52 am

No. It's not about detecting difference. It is about detecting statistically significant differences. You are neglecting the P value

No, I didn't, I did mention alpha being pre determined, You can set it to 0.001, It doesn't matter, For the uninitiated, alpha sets the threshold of rejection. If p-value <= alpha, reject null hypothesis.

925. Does God answer prayer? ASU research says 'yes'

Comment #148605 by Bonzai on March 23, 2008 at 11:48 am

If epsilon is some real effect? If it's real, we want to find it, don't we?

But it may not be if it is very tiny, which would correspond to differences that fall within some experimental error if you really try to monitor it in some way. That's my whole point, significance alone is not enough.

926. Does God answer prayer? ASU research says 'yes'

Comment #148600 by Bonzai on March 23, 2008 at 11:36 am

Re. Your example, of course the answer is yes, it would.

This is a simple independent sample T- test

You null is H0: %match0 - %mathch1 = 0

alternative H1: is %match0 - %mathch1 not= 0

Suppose there is a discrepancy of say 10%, the test would be shown to be significance if it can detect the difference between the null and the alternatives HA's 0< |%match0 - %mathchA |<= 10 (H1 is a compound alternative)

To be able to detect this you must choose your sample size large enough so that the test is sensitive enough (ie with a predetermined beta), it depends on your alpha and beta as given beforehand. There is a formula for that, I won't write it here because of the fonts.

Now change the 10% to epsilon, with the same alpha and beta and using the same formula you can find the sample size to be sufficient large for the test to detect the difference of epsilon.

Well you can specify your epsilon to be so tiny that on a practical level it is indistinguishable to zero, but the test will pick that up if the sample size is large enough,

927. Does God answer prayer? ASU research says 'yes'

Comment #148591 by Bonzai on March 23, 2008 at 11:09 am

I am arguing against your idea that you can prove any hypothesis by increasing sample size.

You cannot "prove" any hypothesis, and I never said you can. But you can show any effect to be significant since that is only a p-value that you are after, I think you are misrepresenting my point here.

You are wrong about the radioactivity example. You can get quite small counts, and in some cases you can be dealing with pretty small numbers of atoms, and the decay can fall to undetectable in finite time. Infinity is not involved.

Well then state your hypothesis and tell me what the power is, You are doing the experiment, not I.

I am sorry it is really quite absurd to argue that sample size has nothing to do with power.,

(Now I was talking about parametric hypothesis testing, specifically the kind of the form H0: effect = 0, now if you are trying to do other things it may be different, so what is your hypothesis?)

You have one group of, say, 100 people who attempt to match a card (chosen by computer) that someone looks at, by ESP.

I was thinking about a single psychic who claims to have ESP.


'

928. Does God answer prayer? ASU research says 'yes'

Comment #148584 by Bonzai on March 23, 2008 at 10:50 am

You are wrong. Look over my radioactivity example

I have explained why that is a different class of problem, because your formula for power doesn't depend on n, you use an asymptotic formula where n--> infinity.

But that has nothing to do with the kind of human scale research we are talking about here, where n cannot be treated as infinity and taken out of the consideration. In these cases, larger n corresponds to higher power. I don't even know what you are trying to argue against, this is first year statistics and it should be obvious for you since you claim to have done graduate level work in statistics,

929. Does God answer prayer? ASU research says 'yes'

Comment #148582 by Bonzai on March 23, 2008 at 10:44 am

No. That is not a valid statistical approach. Please read what I have written in the last few posts. You need controls.

No one say it is. All I am saying is if statistical significance is all you are after then this would do. My whole point is there is more to sound statistics than just a p-value!

930. Does God answer prayer? ASU research says 'yes'

Comment #148577 by Bonzai on March 23, 2008 at 10:34 am

It seemed to me that you were originally saying that a particular hypothesis could be shown to be significant with a large enough sample. What you seem to be saying now is that if you keep sampling for long enough, you can find an arbitrary hypothesis which would seem to be significant. That isn't the same thing.

I was and am saying that you can show a particular hypothesis to be significant if you have a large enough sample.

But if each run is considered an independent trial, then the two are the same (if you want to be pedantic you can invoke some versions of ergodic theorem but it is hardly necessary for the kind of stuffs in social and life science).But you don't have independent trials in the kind of Markov chain modeling you said you are doing, so in that case they are not the same.

As for ESP, it depends on what you determine to be the criterion for acceptance. If all you require is that the subject should guess more than 50% of the time correct with whatever significance level, that can be done with a large enough, pre decided number of runs, But if you insist the subject has to guess x number of times correctly in a row, then it wouldn't work by increasing the number of runs. There is probability 1 (certain event) for any number of consecutive runs to be successful eventually("the infinite monkey theorem") but the waiting time can be the age of the universe depending on how many consecutive runs you stipulate.

931. Does God answer prayer? ASU research says 'yes'

Comment #148572 by Bonzai on March 23, 2008 at 10:12 am

No, this isn't the way statistics is used. You don't say "search for any difference, no matter how small, and keep going until you find it".

I never say that is what you should do, all I am saying is thatif all you care about is statistical significance you can produce significance if you do it

Since you may be doing this inadvertently if you run a bunch of data through SAS or something you should look at other things than just significance even if the test result turns out to be significant, that was all my original point,

If you do it deliberately to mislead people, then at least someone should be able to take you to task, being statistical significant alone doesn't say a lot, all it says is there is a good chance some effect is not zero, but it could be so tiny that it is well within the margin of error of measurements, which can be quite big for the social and life science.

932. Does God answer prayer? ASU research says 'yes'

Comment #148565 by Bonzai on March 23, 2008 at 10:01 am

Dr.Benway

A dishonest scientist gathers a mass of info, runs the ANOVAs, and reports on those measures that come up significant without mentioning all the things that weren't significant.

Yeah, but first of all, you don't run Anova's separately because that will magnify the chance of type I error, you instead do some kind of general linear models where all the factors are considered together, in that case all the p-values would be reported on the same report

Secondly, withholding data is actually cheating, and probably illegal. That is not what I am talking about.

933. Does God answer prayer? ASU research says 'yes'

Comment #148561 by Bonzai on March 23, 2008 at 9:52 am

Dr.Benway,

here are many ways to lie with statistics. Increasing sample size isn't one of them.

Practically, no, in most circumstances anyway because it is a very expensive way to lie, given the fact that it costs a lot to produce samples in the social sciences, But I was making a point about principle, which is not controversial at all.

I see where steve is coming from, In his line of work he deals with samples which are 1) huge, of the order of 10^23, say and 2) easily controlled and reproducible in the lab. These are not the environment of applied statistics in most human scale research.

In Steve's kind of scenario the formulae are somewhat different as well, For example it wouldn't make sense to compute power in terms of sample size because theoretically, he handles his sample size as "infinite" and use asymptotic formulae where n does not appear.

But in social and medical research n is important, A sample size of 200 and 2000 have very different power.

It is really not so difficult to understand. Bigger sample size increases the power, higher power means more sensitive to small differences between the null and the "actual" distributions, Since the null can almost never capture the true distribution because, among other things, we don't formulate our null hypotheses down to 5 or 6 decimal places in typical social science and life science studies, this small discrepancy can be detected, in principle with an increase in sample size. For cheap experiments where cost is not an issue, like making people read cards 1000 times, it has been used to show things like ESP are statistically significant.

P.S. Your point about rounding up and down being equally likely concerns numerical computations where one has to do a lot of roundings, it is not what I am talking about here.

934. Does God answer prayer? ASU research says 'yes'

Comment #148538 by Bonzai on March 23, 2008 at 9:14 am

OK, I also have M.Sc level stats

But let's do it. What is the hypothesis of your example and the power of your test?

EDIT

Just thought of this, in physics you may use an asymptotic formula for power because of the large sample size, so n may not be a meaningful parameter in determining the power in that setting. But in social research, which is what we are talking about here, The sample size would be a couple of hundreds to may be a few thousands at most. You wouldn't find a sample size of 10^23 so most of the asymptotic formulae you use in statistical mechanics type of modeling would not apply.

935. Does God answer prayer? ASU research says 'yes'

Comment #148532 by Bonzai on March 23, 2008 at 9:07 am

You can't simply come up with an arbitrary hypothesis and prove it with enough samples. The whole point of statistics is to look for data trends that are actually there.

Well just look it up.

How do you determine what is actually there and what is noise? The hypothesis testing itself doesn't tell you, you have to use other knowledge and heuristics (often knowledge of the specific disciplines) to decide that.

There is no hypothesis testing that can tell you you are over fitting, thus modeling a lot of noise, for example, you may get a huge R^2 with 100 parameters in a regression model and think that you did a good job.

936. Does God answer prayer? ASU research says 'yes'

Comment #148525 by Bonzai on March 23, 2008 at 8:54 am

No, it doesn't work like that. You can only detect significant differences if they are really there

No, look up the formula for computing power. I am very sure I am right on this one. You can "prove" ESP to be significant using this trick and I think somebody has actually done it.

Indeed this is standard material in any good first year statistics course. You can find that in text books.

EDIT In statistics text books this is usually under the heading :statistical significance and practical importance.

937. Does God answer prayer? ASU research says 'yes'

Comment #148520 by Bonzai on March 23, 2008 at 8:46 am

Steve,

The idea is actually very simple. A statistical hypothesis test is like a microscope which would resolve two overlapping probability distributions, the null and the alternative (let's just stick with a single alternative distribution)

The power of the test is like the resolution power of the microscope. If the power is low, it is difficult to tell the two distributions apart, so you may fail to reject the null even when you should, resulting in type II error. When the power of the test is high, it is sensitive enough to resolve small differences between the null and the alternative hypothesis even when they are very closely overlap.

The power increases as the sample size grows. With a sufficiently large sample size you can detect very minor difference between the null and alternative distribution, leading to the rejection of the null.

Now the null distribution is almost always "wrong" because of truncation errors in representing real numbers. For example, in your null the mean is 7.5, while in fact it is 7.5111. With a powerful enough test you can resolve the difference and get a significant result.

938. The science of religion: Where angels no longer fear to tread

Comment #148513 by Bonzai on March 23, 2008 at 8:32 am

How does altruism in kin groups translate to non kin groups? Members of kin groups have common genes but not in larger groups. Dawkins says it is a "mis-fire". It may turn out to be the only explanation, but it is not the kind of explanation which is so compelling that it would close off other alternative approaches once and for all.

939. The science of religion: Where angels no longer fear to tread

Comment #148510 by Bonzai on March 23, 2008 at 8:25 am

I think it is clearly wrong to say that behaviour that promotes the group over the individual is the opposite of Darwinism as conventionally understood

It may not be opposite, but I find convetional Darwinism has nothing to offer at this level other than providing ad hoc stories of plausibility, evolutionary psychology is like making up stories as one gets along. Sloan Wilson might very well be wrong, but I don't think anyone gets it right either, and certainly not Dawkins and his mematics.

940. It looks like Man crucified

Comment #148508 by Bonzai on March 23, 2008 at 8:20 am

The only section which seems to mention "post humans" is this:

And the title of the piece.

941. The science of religion: Where angels no longer fear to tread

Comment #148501 by Bonzai on March 23, 2008 at 8:11 am

Oh dear. Another attempt to sneak group selection in. Someone needs to re-read The Extended Phenotype.

I don't think he was trying to sneak in anything, he was quite upfront that group selection was not mainstream. However, I find that very dogmatic to think the Dawkins had spoken the last word on the topic and dismiss the whole idea that way.

In science there shouldn't be any taboo, especially in an area such as the evolutionary role of religion, which standard evolutionary theory does not apply directly. Anything that invoke evolution at the social level necessarily would involve a lot of extrapolations

The idea that there is group selection at the social level is not any more crackpotish than Dawkins' idea of memes, which is taken as gospel truth by many here,

942. It looks like Man crucified

Comment #148491 by Bonzai on March 23, 2008 at 7:53 am

Koreman

It seems the author is not talking about atheism but about 'post humans'. People who believe they are going to be godlike in the next four decades with highly advanced technology, augmented intelligence, implants, uploaded brains, eternal life and so on. Amongst those people there are a few who match the description.

Exactly my though.

943. Does God answer prayer? ASU research says 'yes'

Comment #148487 by Bonzai on March 23, 2008 at 7:41 am

Steve,

If you really want to show an effect, you should provide results that are likely to have happened by chance alone less than 1 in 1000 times (p <= 0.001)..

I should also add that even if the study were statistically significant, one has to look at other things such as strength of association and sample size etc in order to conclude that there is a real effect,--though in disciplines like social work they probably wouldn't bother.

It is the nature of hypothesis testing that you can find any effect to be statistically significant with a large enough sample size. For example, you can show that ESP is statistically significant that way.

944. Fleabytes

Comment #148407 by Bonzai on March 23, 2008 at 12:40 am

Thank you Alan F for the excellent post.

However, I do take issue with one point. You said that Einstein denied being an atheist because of political expediency in deference to the prevailing theistic prejudice of the time. You suggested that he might be a bit like U.S. politicians who conceal their atheism while paying lip service to God in order to get elected.

There is a chance that you could be right, though it would be very out of character for Einstein, who always saw himself as an iconoclast. Moreover, having had openly called the belief in a personal God, including the Judeo Christian God, "infantile" and "childish" I don't think he could have gotten much worse in the eyes of the pious majority by saying he was an atheist, if it was just about renouncing all gods.

From reading Einstein's polemics against "atheism", it is clear that his definition of "atheism" was somewhat different from conventional usage today. It appeared that he used the world "atheism" to mean not simply an absence of theistic belief, but a rigid ideology which saw the universe as just "matter in motion" devoid of mystery and beauty, everything was laid bare by ruthless logic. This kind of "atheism" didn't even have room for "Spinoza's God".

An "atheist", therefore to Einstein was not a just a non believer, but a positive Philistine, a vulgar materialist who was tone deaf to "the music of the spheres".(my paraphrase of E's own words to that effect)

Based on this definition it is easy to understand why Einstein wouldn't want to have anything to do with "atheism".

P.S.In early 20th century Europe, "Atheism" was a political banner often associated with various radical movements, He might be thinking of say, the Marxist-Leninists or the nihilists when he used the label "atheists"

945. Fleabytes

Comment #148318 by Bonzai on March 22, 2008 at 4:06 pm

Dr. Benway,

.Murky believers are less of a problem than fundamentalists because they simply don't believe much of what their religion teaches. That's not a strong argument in favor of belief, I'm afraid.

I never argue in favour of beliefs, To say something is harmless is not the same as recommending it, I would recommend it only to fundies.

946. Fleabytes

Comment #148316 by Bonzai on March 22, 2008 at 4:01 pm

All I am arguing is that that Churches have influence.

Well then I don't know what you are arguing against. Certainly I wouldn't disagree with that, it is a truism.

947. Fleabytes

Comment #148312 by Bonzai on March 22, 2008 at 3:56 pm

religion helps.

So can other things. Pseudoscience, for one.

948. Fleabytes

Comment #148311 by Bonzai on March 22, 2008 at 3:54 pm

It does have specific content. It has the content put into it by those in the religious organisation you joint

That depends on how seriously you are a joinder. If you are right,--you are making categorical statement here, not I,--why is it that most Catholics in the U.K and the developed countries in general do practice birth control? This is a main doctrine of the Catholic church, not a minor issue.

I didn't say joining the Church would have no effect on anyone, but it is you who argue
categorically that it would have a strong enough effect even on very loose believers, I think your case in not tenable.

Your example of trade union is a good one. I have to join a union and pay union dues for most jobs that I had. But I can't really say what their views are on a lot of issues.

949. Fleabytes

Comment #148306 by Bonzai on March 22, 2008 at 3:45 pm

We get "evidence" from our dealings and conversations with other people. This evidence can include views that coloured people are inferior (or "not our kind of people"), or that God tells us that what gay people do in bed is naughty.

You don't need religion for that.

950. Fleabytes

Comment #148303 by Bonzai on March 22, 2008 at 3:40 pm

Whenever anyone offers up a rational about how belief in God influences people and therefore affects their decisions you just ignore the point and say, well people who don't believe in God have the same fallibility too, or that belief in God is not the only motivation for that particular act. Who is arguing that?

No, I am saying a "murky" belief in God in itself doesn't have any specific content,--it depends on what you put in it,-- it doesn't tell you to do good or bad. In a way it is a screen, a vehicle, so it is barking up the wrong tree to say believing in God would make you do a, b and c, it doesn't follow. It also doesn't follow that if you have this one "irrational" belief, you must be irrational consistently, most of us have irrational "beliefs", if not religion then it would be something else.