Night Owl Mk. II

HomeSite 4.0
Created with Allaire HomeSite 4.0

Last Update: 01 Feb 99

Return to "Religion" essay

Back to Philosophy page

Please feel free to E-mail me with your own comments on this issue or on anything else included in my Philosophy of Life section. Debate is good!

Please report any problems with this page to the Webmaster!


Boldfaced statements are parts of the original essay (or a subsequent reply) to which the respondent has directed his comments.

Italicized/emphasized comments
prefaced by (R) are those of the respondent and are presented unedited.

My replies appear under the respondent's comments in blue text and are prefaced by my initials (MB).

This is the last of a two-part reply.

(R) There's also another website you should check out, at:
    It describes the poll you've placed so much faith in. This counter poll purports to cast doubt on the Larson/Witham survey, but unwittingly debunks itself by the woefully flawed polling methods it describes.

(MB) How can this describe a "counter poll" when what it is reporting *is* the Larson/Witham survey?!?!? How can you uphold the survey (part of it, anyway) on one hand while also tearing it down for "woefully flawed polling methods"? I think what we have here is another proof that religious belief is more important to believers than are facts or cogent and coherent reasoning. Let me know when you figure out how you want to treat this survey and its results.

(R) Here's how the poll was done:
"The follow-up study reported in "Nature" reveals that the rate of belief is lower than eight decades ago. The latest survey involved 517 members of the National Academy of Sciences; half replied. When queried about belief in "personal god," only 7% responded in the affirmative, while 72.2% expressed "personal disbelief," and 20.8% expressed "doubt or agnosticism." Belief in the concept of human immortality, i.e. life after death declined from the 35.2% measured in 1914 to just 7.9%. 76.7% reject the "human immortality" tenet, compared with 25.4% in 1914, and 23.2% claimed "doubt or agnosticism" on the question, compared with 43.7% in Leuba's original measurement. Again, though, the highest rate of belief in a god was found among mathematicians (14.3%), while the lowest was found among those in the life sciences fields - only 5.5%."

(MB) As I've already said, this is part two of the Larson/Witham survey for which I gave the complete citation in a previous paragraph.

(R) I took a semester of statistics and probability about 15 years ago, so I dug out my old text book and did some brushing up.
(MB) Oh, boy! I fall to my knees to pay homage to the extent of your education. (Should I mention that I was a mathematics and statistics major during the first two years of my college days? No. Why spoil the fun?)

(R) This poll fails badly on the following counts:
1.) Sample Size: The sample size of 258 given in the above description yields a margin of error of plus/minus 6.2 percent for this poll. Anything more that five percent (a ten percent range) is generally considered unacceptable by statisticians.

(MB) Quite true. But, you fail to understand that a statistically unacceptable result is not the same as an "invalid" result. You also fail to understand that general conclusions can still be reached from insufficiently accurate poll results. For example, consider a poll with a plus/minus 6.2% margin of error which reports that 80% of the population says "Yes" while 20% say "No" to a given question. What can and can't we correctly conclude? We can't accurately say that four times as many people say "Yes" or that one-fifth of the people will say "No", but we *can* say that many more people say "Yes" and that a small minority of the people say "No". The margin of error comes into play when the percentage difference between answers falls within that margin. When it doesn't, we can still reach correct conclusions from the data.

(R) This poll indicates that somewhere between zero and 14 percent of scientists believe in a personal God - a woefully imprecise range. In contrast, the Larson/Witham poll (with a sample size of 600 and a margin of error of 4.1 percent) shows between 35 and 43 percent do.
(MB) Here you are combining numbers from two different polls of two different populations. Please get your figures straight and try again.

(R) (Note that the accuracy of a poll is completely independent of population size. Instead, it depends entirely on sample size. A poll of 1000 people has a margin of error of plus/minus 3.1 percent, whether the base population is two thousand or two hundred million.
(MB) Quite true, although the relevance of that fact is questionable. The number of people polled has not been in question in this discussion.

(R) An excellent approximation of the margin of error of a poll can be arrived at by inverting the square root of the sample size. One other comment - the validity of the margin of error depends absolutely on rigorous randomness in selecting a sample. Sample bias makes margin of error (and the poll to which it applies) meaningless.)
(MB) I alluded to this fact earlier in my scenario postulating a poll of Creationists. Of course, you have tried to claim that a poll of scientists about the opinions of scientists is somehow "biased" when it doesn't produce the results that you want to see.

(R) 2.) Response Rate: Regardless of statements made to the contrary by the author of this article, response rates of anything less than 66 percent are considered very poor. The abysmal response rate of 50 percent cited for this poll greatly increases the risk that those who declined to respond belong to a significant subgroup whose attitudes and opinions the poll doesn't represent.
(MB) How do you determine that? Maybe they just don't like answering polls in general or maybe they just consider the subject matter of the poll to be irrelevant. If your assertion is correct, we should expect that the poll results would have been weighted in favor of the religious viewpoint since religious beliefs are powerful to those who possess them.

(R) The response rate of 60 percent given by this website for the Larson/Witham survey, while better, isn't very good either. However, this site says merely "60 percent" and "half" when giving the response rates for the polls, which is oddly imprecise in an article which otherwise cites statistics to three significant digits. It seems likely the author rounded off the response rates. If so, which way? "60 percent" might mean anything in a range between 55 and 64.9. Was the Larson/Witham survey in fact close to 66 percent and the other poll significantly less than "half?"
(MB) Doubtful. Even somebody using imprecise numbers isn't likely to be *that* far off. Are you suggesting that there is something sinister about the response rate figures? If so, how can you uphold the results of the part of the survey with which you agree?

(R) 3.) Sample Randomness: This is undoubtedly the most serious flaw in this poll. Limiting the polling sample to the National Academy of the Sciences represents a clear attempt at "cherry picking," which is the statistical equivalent of gerrymandering.
(MB) You have made this claim before, but have yet to explain why this would invalidate the survey results. You will especially need to do some hard explaining in light of the fact that this survey is a replication of the original survey and methods of Leuba (whose results you agreed with).

(R) It is tantamount to going to Yankee Stadium, polling the first 200 people who come out after a ballgame, and then claiming to have conducted a poll representative of all baseball fans. This poll doesn't accurately portrait the religious beliefs of scientists. It merely shows the beliefs of NAS members, which may or may not parallel the beliefs of other scientists.
(MB) Why wouldn't the beliefs of NAS members reflect the beliefs of scientists in general when NAS members are elected by those other scientists and since religious beliefs (or lack thereof) of candidates are not a factor in their election?

(R) (And not very accurately either, I might add.)
(MB) You might add it, but your assessment would appear to be even less accurate.

(R) Only by random polling, where every scientist has a nearly equal chance of being selected, can the views of all scientists be reliably estimated. This is what Larson and Witham did, and Leuba before them.
(MB) Once again, you show that you only consider the parts of their studies that you choose to believe and brush off the rest. It also shows that you are completely unaware of all of the supporting studies that I mentioned earlier.

(R) One final shot at this poorly conceived poll...
(MB) Better load your gun with something other than blanks...

(R) - the last statement in the paragraph quoted above states, "...the highest rate of belief in a god was found among mathematicians (14.3%), while the lowest was found among those in the life sciences fields - only 5.5%." This is laughable.
(MB) Actually, this finding has been upheld in many other studies. In general, belief levels decrease as the scientific discipline gets "harder" (i.e., more concerned with observation and evidence).

(R) How many mathematicians were represented in the poll? Twenty? That gives a margin of error of 22 percent for the views of mathematicians! All that can realistically be said from this poll is that fewer than 36 percent of mathematicians believe in God.
(MB) Not really. One can also say that no more than approximately 1/3 of mathematicians believe in God, that mathematicians are at least twice as likely to believe in God than are those in the life sciences, and that mathematicians have a *far* lower percentage of belief than do members of the general public while still being the most likely scientists to believe in God. Nowhere in here is there any support for any claims that belief in God among scientists is on the rise.

(R) Anyway, so much for the "blatant lie" you've accused Newsweek of making.
(MB) Yes, I'd say that I've sufficiently demonstrated my case.

(R) Newsweek's statement was based on a scholarly, meticulous survey which rigorously repeated a highly respected earlier survey.
(MB) ...and which ignored the entirety of the work done by the same individuals in both cases. It's absolutely hilarious how you call it "scholarly and meticulous" when you like the results and then turn around and call the same work "woefully flawed" when you don't like the numbers.

(R) Your accusation was based on a poll using methods, which if used by a politician for measuring public opinion or a quality control specialist to check the quality of a production run of widgets, would be next to worthless.
(MB) My "accusation" uses the very same work by the very same pollsters that you can't seem to decide how to interpret. Since I know the whole story, I know how to interpret it.

(R) You've stated that only someone completely unfamiliar with history could say that many scientists are arriving at the belief it is rational to believe in God, and further stated that religious belief among scientists is in rapid decline.
(MB) That's correct. The numbers prove my case, don't they? As stated before, I have many more studies which back up my statements.

(R) You're well off the mark on both counts.
(MB) Perhaps you'd now wish to revise this mistaken statement?

(R) I don't believe you're unfamiliar with history, but I do think you simply ignore those parts of it which don't support your views.
(MB) How can I do that when history supports my views?

(R) Most of the greatest names in science before the 19th century were devoutly religious men, including Galileo.
(MB) Quite true and I've never disputed that. Remember that this current discussion is about whether or not belief among scientists is in decline. If we both agree that most scientists were believers prior to the 20th century and we both agree that a minority of scientists are believers now (whether that percentage is as high as 40% or as low as 7%), is that not proof that belief among scientists is actually in decline? How then does history not support my views?

(R) The theories of Charles Darwin at mid-century caused a good deal of retreat from religion among scientists, especially when taken in conjunction with the philosophical argument the universe is eternal and has no beginning,
(MB) Good grief! What a mix of apples and oranges! Not only do Darwin's theories have absolutely nothing to do with the origin of the universe, the theory that the universe has no beginning didn't arise until over a century after Darwin wrote his Origin of Species.

(R) however, the theoretical evidence of the Big Bang postulated in the 1920's, ...
(MB) What we call the "Big Bang" theory was first proposed by George Gamow in 1946. I think you have it confused with the prediction of general relativity that the universe must be expanding. The solution to Einstein's equations which produced that prediction was found in 1922.

(R) ...followed by direct empirical evidence which supported the theory as provided by the COBE project, began to halt this withdrawal and even reverse it - to the point that the re-acceptance of religion by scientists has received wide media notice.
(MB) How can findings that verify a theory of science which demonstrates that the universe was *not* created by any God cause an increase in belief in that God?

(R) Newsweek, Time, Scientific American, Nature, and many others have published articles on this in recently years.
(MB) I think I've sufficiently demonstrated that such articles do *not* support what you wish to claim for them (assuming that they haven't left out important parts of the relevant facts).

(R) There is limitless material available on the internet about it. Counterbalance, at, provides a prime example of the carefully considered interest which exists in science and religion.
(MB) This is great! Have you actually read anything on that site? Ask yourself just what is being "counterbalanced". The articles on that site are all aimed at preserving religious belief by showing how it can be harmonized with science. There are exactly zero articles there which show how any Christian belief has been undermined by science. I suppose this is another example of an "unbiased and even-handed" site?

(R) The scientific support of atheism which you place such faith in is evaporating.
(MB) On the contrary, it is readily apparent that it has never been stronger and is continuing to grow.

Created with Allaire HomeSite 4.0 .......... Last Update: 01 Feb 99

Earthlink Network Home Page